A Survey on using Ant-Based techniques for clustering

第1篇Date: [Insert Date]Location: [Insert Venue]Duration: [Insert Duration]Participants: [Insert Names of Participants]Facilitator: [Insert Name of Facilitator]Introduction:The English Grammar Research and Development Activity was conducted to enhance the understanding and application of English grammar among the participants. The session aimed to explore new teaching strategies, discuss common challenges faced in teaching grammar, and share innovative methods to improve student learning outcomes.Agenda:1. Opening Remarks2. Introduction to Grammar Teaching Strategies3. Workshop on Grammar Activities4. Group Discussions and Case Studies5. Innovative Techniques for Grammar Instruction6. Interactive Grammar Games7. Conclusion and Feedback Session1. Opening Remarks:The facilitator began the session with a brief introduction,highlighting the importance of grammar in language learning and the role it plays in communication. The participants were reminded of the need to integrate grammar into their teaching practices effectively.2. Introduction to Grammar Teaching Strategies:This section focused on various strategies that could be employed to teach grammar more effectively. The facilitator discussed the following methods:- Contextual Learning: Using real-life contexts to teach grammar, making it more relatable and memorable.- Interactive Lessons: Incorporating activities such as role-plays, debates, and group discussions to engage students actively.- Technology Integration: Utilizing educational technology tools to create interactive and engaging grammar lessons.- Assessment for Learning: Using formative assessments to monitor student progress and adjust teaching strategies accordingly.3. Workshop on Grammar Activities:Participants were divided into small groups to work on designing grammar activities. Each group was given a specific grammar topic to focus on, such as verb tenses, sentence structure, or prepositions. The groups were asked to create a lesson plan that included an introduction, main activity, and conclusion.4. Group Discussions and Case Studies:Following the workshop, each group presented their lesson plans to the larger group. The facilitator facilitated a discussion on the effectiveness of the activities and provided feedback on areas of improvement. Case studies were also presented to illustrate common challenges faced by teachers and possible solutions.5. Innovative Techniques for Grammar Instruction:The session continued with an exploration of innovative techniques for teaching grammar, including:- Grammar Journals: Encouraging students to keep a journal where they can practice and reflect on their grammar learning.- Grammar Videos: Creating or using existing grammar videos to provide visual examples and explanations.- Grammar Apps: Introducing grammar apps that offer interactive exercises and quizzes to reinforce learning.6. Interactive Grammar Games:To conclude the workshop, the facilitator introduced several interactive grammar games that could be used in the classroom. These games were designed to make grammar learning fun and engaging for students.7. Conclusion and Feedback Session:The session ended with a feedback session where participants sharedtheir thoughts on the activity. The facilitator collected feedback on the content, structure, and practicality of the workshop. Participants expressed their gratitude for the opportunity to learn new methods and strategies for teaching grammar.Activity Record Details:Grammar Topics Discussed:- Verb Tenses- Sentence Structure- Prepositions- Subject-Verb Agreement- Word FormationActivities Designed:- Grammar Scavenger Hunt- Grammar Charades- Grammar Crossword Puzzle- Grammar Board Game- Grammar DebateFeedback from Participants:- "The workshop provided me with practical ideas that I can implement in my classroom immediately."- "I appreciate the emphasis on student-centered learning and the use of technology in grammar instruction."- "The group discussions and case studies were very helpful in understanding common challenges and finding solutions."Conclusion:The English Grammar Research and Development Activity was a successful event that provided valuable insights and resources for teachers. The participants left the session with a renewed enthusiasm for teaching grammar and a toolkit of innovative strategies to enhance theirstudents' learning experiences.第2篇Date: [Insert Date]Location: [Insert Location]Participants: [List of Participants]Facilitator: [Name of Facilitator]Duration: [Insert Duration, e.g., 3 hours]---I. IntroductionThe English Grammar Research and Development activity was held to enhance the participants' understanding of grammar teaching strategies, explore new methodologies, and discuss practical classroom applications. The session aimed to provide a platform for teachers to share their experiences, exchange ideas, and improve their teaching skills.II. Objectives1. To review and discuss effective grammar teaching methods.2. To explore innovative approaches to grammar instruction.3. To analyze the challenges faced in teaching grammar and brainstorm solutions.4. To develop a shared understanding of grammar assessment techniques.III. Activity Outline1. Warm-up (15 minutes)- Icebreaker activity: Participants shared their favorite grammar teaching techniques.- Brief discussion on the importance of grammar in language learning.2. Workshop Session (1 hour)- Presentation on current trends in grammar teaching.- Interactive activities to practice grammar teaching strategies.- Group work: Participants were divided into small groups to discuss and implement grammar activities.3. Case Studies (45 minutes)- Sharing of real-life classroom scenarios and challenges.- Analysis of case studies by the group, focusing on effective grammar teaching strategies.4. Innovative Approaches (30 minutes)- Presentation on innovative grammar teaching methodologies.- Discussion on the feasibility and effectiveness of these approaches.5. Grammar Assessment Techniques (30 minutes)- Overview of different grammar assessment techniques.- Group activity: Participants developed grammar assessment tools based on the discussed methodologies.6. Feedback and Reflection (30 minutes)- Feedback on the day's activities.- Reflection on personal teaching practices and areas for improvement.IV. Detailed Activity Record1. Warm-upThe session began with a lively icebreaker activity, where participants were asked to share their favorite grammar teaching techniques. This encouraged open communication and set a positive tone for the day's activities. The discussion quickly turned to the importance of grammarin language learning, with participants emphasizing the need for a balanced approach that combines practical application with theoretical understanding.2. Workshop SessionThe workshop session commenced with a presentation on current trends in grammar teaching. The facilitator highlighted the importance of using technology and multimedia resources to make grammar lessons engaging and interactive. Participants were then engaged in interactive activities designed to practice these strategies, such as grammar games and role-playing exercises.Following the workshop, participants were divided into small groups to discuss and implement grammar activities. Each group was tasked with creating a lesson plan that incorporated the strategies discussed in the workshop. The groups presented their plans to the larger group, receiving constructive feedback and suggestions for improvement.3. Case StudiesThe case studies session provided participants with the opportunity to analyze real-life classroom scenarios and challenges. Each participant shared a case study, and the group engaged in a detailed analysis of theissues presented. The facilitator guided the discussion, focusing on effective grammar teaching strategies that could be applied to similar situations.4. Innovative ApproachesThe session on innovative approaches to grammar instruction introduced participants to new methodologies such as project-based learning and gamification. The facilitator presented examples of these approaches and led a discussion on their feasibility and effectiveness in the classroom. Participants were encouraged to think creatively about how they could incorporate these strategies into their teaching.5. Grammar Assessment TechniquesThe final session of the day focused on grammar assessment techniques. The facilitator provided an overview of different methods, such as formative and summative assessments, and explained how they could be used to gauge students' understanding of grammar. Participants then worked in groups to develop grammar assessment tools based on the methodologies discussed.6. Feedback and ReflectionThe session concluded with a feedback and reflection activity. Participants provided feedback on the day's activities, highlighting the most valuable aspects and areas for improvement. They also engaged in a reflective discussion on their personal teaching practices, identifying areas where they could apply the new strategies and techniques learned during the workshop.V. ConclusionThe English Grammar Research and Development activity was a successful and informative session that provided participants with valuableinsights and practical tools for enhancing their grammar teaching skills. The facilitator and participants expressed their gratitude for the opportunity to learn from each other and look forward to applying the new strategies in their classrooms.VI. Follow-up Actions1. Participants will share the grammar activities and assessment tools developed during the workshop with their colleagues.2. A follow-up meeting will be scheduled to discuss the implementation of new strategies and share progress.3. A survey will be distributed to gather feedback on the workshop and identify areas for future improvement.---This activity record provides a comprehensive overview of the English Grammar Research and Development session. It captures the objectives, activities, and outcomes of the workshop, as well as the participants' feedback and reflections. The record serves as a valuable resource for future planning and improvement of grammar teaching practices.第3篇一、活动背景随着新课程改革的不断深入，英语教学在基础教育阶段越来越受到重视。

阅读调查英文作文模板英文回答：In conducting a reading survey, researchers employ a variety of techniques to gather data from participants. One common method is the use of questionnaires. Questionnaires present participants with a series of fixed-response questions, which may be open-ended (allowing forqualitative responses) or closed-ended (providing a set of pre-determined answer choices). By analyzing the responsesto these questionnaires, researchers can obtainquantitative and qualitative data on participants' reading habits, preferences, and comprehension levels.Another approach to conducting reading surveys is the use of interviews. Interviews involve face-to-face orvirtual interactions between a researcher and a participant. During an interview, the researcher asks the participant a series of open-ended questions about their reading experiences. Interviews allow researchers to gain deeperinsights into participants' thoughts, feelings, and motivations related to reading.In addition to questionnaires and interviews, researchers may also employ observational methods to conduct reading surveys. Observational methods involve observing participants' reading behaviors in a natural setting. Researchers may observe participants in their homes, schools, or libraries, recording their reading choices, reading times, and reading comprehension. By observing participants' reading behaviors, researchers can gain valuable insights into their actual reading practices.The specific methods used in a reading survey will depend on the research objectives, the participant population, and the resources available to the researcher. However, by carefully selecting and employing appropriate data collection techniques, researchers can gather valuable information on the reading habits, preferences, and comprehension levels of their participants.中文回答：阅读调查。

安徽省“耀正优”2023-2024学年高三上学期12月名校阶段检测联考英语试题学校:___________姓名：___________班级：___________考号：___________一、阅读选择Here’s all the information you need to know about getting a membership to Sam’s Club.Explore all the membership offers below!Sam’s Club Annual Membership.Sam’s Club has two annual membership options, Club Membership and Plus Membership:* Club Membership: This is the standard membership level and costs $50 per year.* Plus Membership: This is the top membership level and costs $110 per year. Though the Plus Membership is more expensive, it gets you an additional 2% back on your purchases (up to $500 a year)Sam’s Club 3-Month Free Trial MembershipA Sam’s Club 3-month Free Trial Membership allows you to experience everything that a Club Membership gets for 90 days.This free trial offer is limited to new members only and you can only get one of these trial memberships per household(一家人).The trial membership is set up to auto-renew(自动续订) and you will become a club member at the end of the three-month period automatically, so be sure to cancel before 90 days is up. You can cancel at any time during the free trial by calling 1-888-746-7726.Sam’s Club 1-Day Shopping PassSam’s Club offers a l-Day Shopping Pass, which allows non-member guests to shop at Sam’s Club for 24 hours without a membership. Though it came with a few restrictions: * Shopping using a 1-Day Shopping Pass will be charged a 10% fee on all purchases.* A Guest Membership is active for 24 hours starting at the time of purchases.* Scan &Go can only be used by members and is not available for guests.1．How much should a Plus Membership user pay for something bought by a Club Membership user priced at $30?A．$28.8．B．$29.4．C．$ 30．D．$30.6．2．What do we know about the 3-Month Free Trial Membership?A．It offers less benefit than a Club Membership.B．One can cancel it anytime before the end of 90 days.C．One can get two of these trial memberships per household.D．It is necessary for you to phone if you want to become a club member.3．Why is the 1-Day Shopping Pass provided?A．It is for non-member guests to purchase things.B．Shopping with the pass will get a 10% discount.C．Scan &. Go can be available to those with the pass.D．Those with the pass can enjoy the top-level service.As an engineer, I have designed electronic control systems for more than 30 years, and I expected to do so until I retired. My wife, Krisztina Valter, is a scientist. One day, she went to a scientific activity with me. Most knowledge of biology went over my head, but one speaker attracted my attention. Jochen Zeil, a professor who studies animal behavior, presented a model of how insects could work out a way to go to a target despite having small brains.At lunch time, my wife introduced us. Zeil and I had a very enjoyable discussion about his idea and biology. Before we parted, I joked that if he wanted another PhD student, he could count me in. About a month later he emailed me, “Haven’t heard from you. Have you enrolled (注册) yet?” And that’s how, at the age of 53, I became a part-time doctorial student in biology.My background in biology had been out of date for many years, so I needed to learn the basics fast. It was not easy, but I got a lot of help from other laboratory members and my wife. The more I learnt, the more I realized how little I knew. Every paper caused the need to read more. It was tiring, and at times disappointing, but brought a lot of fun. You read, think, and suddenly things fall into place.But there is something I have to get used to. If you repeat an engineering experiement, you expect to get the same result. This is not so in behavior biology. Your experiment subjects actually have a mind of their own—You can put an ant on a trackball, but you can’t make it walk on the ball.Now at the age of 61, this period is nearly at an end. New knowledge enriches you, regardless of how old you are. My advice is that if you have the opportunity to enter a new field, take it.4．What did the author think of most of the knowledge about biology presented at the activity?A．He couldn’t think too highly of it.B．He couldn’t make sense of it.C．He considered it quite easy.D．He felt it was so practical.5．How did the author become a doctorial student in biology according to paragraph 2?A．He expected to realize his dream in his heart for long.B．He said something amusing and Zeil took it seriously.C．He was always urged by his wife to learn biology.D．He tried hard to persuade Zeil to take him in.6．What is the theme of third paragraph?A．The author’s good relationship with other laboratory members.B．The author’s difficulty in learning and his discouragement.C．The author’s learning experiences and his reflection.D．The author’s purpose of learning the subject.7．What can we infer from the author’s words in the last two paragraphs?A．Experimenting with insects may get a different result.B．Old people ought to work hard to get a college degree.C．The author prefers behavior bi ology to engineering now.D．Doing engineering experiments is easier than doing bi ology ones.A newly developed algorithm (算法) can spot depression in Twitter users with 88.39% accuracy. Developed by researchers at Brunel University London and the University of Leicester, the algorithm determines someone’s mental state by analyzing the 38 data points from their public Twitter information, including their posts, their posting times, and the other users in their social circle.“We tested the algorithm on two large databases and compared our results with those obtained by other depression detection techniques,” said Prof. Abdul Sadka, Director of Brunel’s Institute of Digital Futures. “In all cases, we’ve managed to make it perform better than existing techniques in terms of their classification accuracy.”The algorithm was trained by using two databases that contain the Twitter history ofthousands of users, as well as additional information about those users’ mental health. Eighty percent of the information in each database was used to teach the robot, with the other 20% then used to test its accuracy. The robot works by first excluding (排除) all users with fewer than five posts and running the remaining information through natural language software to correct for misspellings and short-form words. It then considers the 38 different data points—such as a user’s use of positive and negative words, the number of friends and followers they have, and their use of emojis—and makes a decision on that user’s mental and emotional state.The team says that such a system could potentially mark a user’s depression before they post something into the public websites on the Internet, paving the way for platforms such as Twitter and Facebook to proactively (积极行动地) warn users about their mental health issues. What’s more, the robot can also be used after a post has been made on the public platforms, potentially allowing employers and other businesses to assess a user’s mental state based on their social media posts. It could be used for a number of reasons, the researchers say, including the use in emotional analysis, criminal investigation or employment screening. 8．How does the robot determine a Twitter user’s mental state?A．By accessing the user’s previous health status.B．By inquiring about the user’s friends and followers.C．By discovering the misspellings and short-form words in the user’s posts.D．By taking into consideration the different aspects from the user’s public information. 9．Why does the author mention Prof Abdul Sadka’s words in paragraph 2?A．To show that the newly developed algorithm is better at depression detection.B．To emphasize the differences between depression detection techniques.C．To reveal the seriousness of depression among Twitter users.D．To introduce some ways to detect depression.10．What is the third paragraph mainly about?A．What factors determine depression.B．The common symptoms of depression.C．Why researchers wanted to access people’s mental state.D．The training and working process of the algorithm system.11．What is one of the practical uses of knowing the users’ mental and emotional state?A．It can help improve the techniques of depression detection.B．It enables the platforms to provide medical service for their users.C．It can help companies select more suitable staff from some candidates.D．It makes it possible for websites to better protect their users’ privacy.With the recent ban on iPhone 12 sales in France due to radiation concerns, a potential snowball effect might appear, as other European Union member states could follow suit (效仿). In a statement to the press earlier this month, France’s Digital Minister Jean-Noel Barrot said radiation from the iPhone model was above what is legally permitted according to national technology watchdogs.The Specific Absorption Rate (SAR) for the iPhone 12, when worn close to the body, for example in trouser pockets, was found to be above the legal limit of 4wattsper kilogram. Apple now plans updates, in response to the radiation from iPhone 12, which is 5.74 watts per kilo, to be rolled out to all iPhone 12 owners across France.The French government warned that failure from Apple would need it to recall all the iPhone 12s already owned by consumers within the country. Currently the World Health Organisation states that there is no scientific evidence to support the idea that mobile phone radiation could be harmful to humans.Some organisations recommend caution for their use and support further definitive long-term research, which is why government safety guidelines are tentatively (实验性地) in place. It is true that phones emit radiation. However, it is generally assumed to be safe and is the same type used to transmit radio station frequencies.This is a different type compared to other forms of radiation such as x-rays, which we know to be harmful. So far, no good explanations have been found to demonstrate that radiation from mobile phones can affect people’s health. It is also unlikely that they will increase cancer risk as the energy emitted is generally thought of by scientists as being too weak to damage DNA．The issue has resulted in the French government expressing their attitude, and Apple is not the only manufacturer under scrutiny. Other brands such as Samsung and Motorola all have models which exceed the national limit. In fact, over 50 phone models across the major tech company sphere are included in the list by French technology watchdogs, with flagship products such as the Samsung Galaxy Note 10 and Motorola Edge all under scrutiny.While it is unlikely that mobile phone radiation is likely to damage health, France’sdramatic assertion(断言) over a tech giant like Apple has caused others to follow suit. Mathieu Michel, Belgium’s secretary of state for digitization, has also vowed to investigate the issue for the country, while Germany and Italy are also paying close attention to any developments.12．How much higher is the radiation from an iPhone 12 than that of the legal limit?A．1.32 watts per kilo.B．1.46 watts per kilo.C．1.57 watts per kilo.D．1.74 watts per kilo.13．What did the French government require Apple to do?A．To repair all the iPhone 12s sold in France.B．To update all the iPhone 12s sold in France.C．To ask for all the iPhone 12s to be returned.D．To replace all the iPhone 12s with other types.14．Which can replace the underlined word “scrutiny” in paragraph 6?A．Examination.B．Development.C．Instruction.D．Budget. 15．What is the best title for the news report?A．France promises to accuse Apple for iPhone 12B．France’s iPhone 12 ban leads to Apple’s protestC．France promises to reduce radiation level from iPhone 12D．France’s iPhone 12 ban could lead to EU-wide tech regulationQuick facts about OPTYear after year, international students from around the world come to the US in pursuit of one of the best educational offers in the world. 16 But for others the goal is to work in the US after graduation.What is OPT?17 Every international student in the US entitled (使有权利) to 12 months of practical training (i.e. work) in the US under OPT. Typically, if you are not a citizen of a particular country, then you are not allowed to work in that country without proper documents. Similarly, as an international student in the US without proper documents, OPT is one that allows you to do so.It is also important to note that students studying in the STEM fields get a 24-month extension in addition to the normal 12 months for a total of 36 months.18However, OPT only gives international students 12 months of employment. In the long run, foreigners would seek an H1-B visa in order to work in the US for the long term. 19 You will first need to find an company that would like to sponsor (赞助) you for your H1-B visa. This means that you have to apply for jobs, and on top of that, you have to make sure that if you accept a job offer, that company will want to pay the additional fee to sponsor you.Why not go straight for an H1-B after graduation?As stated previously, a company is required to sponsor you in order for you to be eligible (合格的) for an H1-B visa. This means that if you break the law, then the company sponsoring you will share responsibility in one way or another. Some companies might view sponsoring an international student as a legal responsibility. 20 If your goal really is to work in the US in the long run, then the OPT gives you a period of time to prove yourself to the company you’d like to work for.A．They wish to give others a hand.B．How can you apply for a suitable college?C．Thus they may not want to go that extra mile.D．OPT simply stands for Optional Practical Training.E．What if you want to work in the US for longer than that?F．Many students plan to make use of their degrees back home.G．Similar to applying for universities, there are requirements to apply for an H1-B visa.二、完形填空For 35 years, Floyd Martin has delivered the mail to residents in Marietta, Georgia. Onneighborhood with something more 29 : kindness. Some neighbor s 30 their mailboxes with flowers, balloons and signs, and the celebration continued when his final day of work ended.More than 300 neighbors attended the neighborhood party honoring Martin. Several stood up and said kind words to honor him. Then, Martin rose to thank the neighborhood. “Thank you for 31 me. We’ve gone through good times and bad times together.” He said 32 he had lost his parents and many of his friends, he knows he still has friends in the Mariett a neighborhood. “You are all here,” he said. “You were there when I 33 you, although you didn’t know it.”Martin doesn’t have any children of his own. 34 , he knows that even in retirement, he can return to this neighborhood and see the hundreds of kids and friends who love him. “I love you guys. I say that. I mean it,” he said. “And that’s what the world needs more of now love, caring and 35 and taking care of one another.”21．A．retirement B．presentation C．election D．promotion 22．A．broke down B．showed up C．dropped out D．cut in 23．A．skilled B．ambitious C．strict D．beloved 24．A．made B．adapted C．shared D．shortened 25．A．touched B．followed C．disturbed D．attacked 26．A．office B．park C．supermarket D．neighborhood 27．A．tasks B．allowances C．treats D．suggestions 28．A．Regardless of B．Aside from C．Far from D．In case of 29．A．invisible B．abstract C．complex D．meaningful 30．A．decorated B．painted C．designed D．covered 31．A．putting up with B．calling for C．caring about D．living up to 32．A．even though B．unless C．until D．supposing that 33．A．suspected B．described C．criticized D．needed 34．A．Otherwise B．Furthermore C．However D．Besides 35．A．hatred B．sympathy C．complaint D．annoyance三、语法填空阅读下面短文，在空白处填入1个适当的单词或括号内单词的正确形式。

谈话调研推荐报告报告内容是关于谈话调研的推荐。

建议将中文报告改为英文报告。

Report on Recommended Conversation Survey Methodology Introduction:Conversation surveys are an essential tool in gathering qualitative data and understanding people's opinions and experiences. This report aims to provide recommendations on how to conduct an effective conversation survey, including the selection of participants, survey design, and data analysis. The recommendations are based on extensive research in the field and will help researchers and practitioners optimize their conversation survey methodology.Selection of Participants:1. Targeted Group: Determine the specific group that should be targeted for the conversation survey. This could be based on demographics, interests, or specific experiences relevant to the research objectives.2. Sampling Technique: Utilize appropriate sampling techniques such as random sampling or purposive sampling to ensure the representation of the target group. Consider using diverse sample sizes and characteristics to enrich the data collected.Survey Design:1. Structured Questions: Design a set of structured questions to ensure consistency across respondents and facilitate easier dataanalysis. These can include closed-ended questions with predefined response options or Likert scale questions.2. Open-ended Questions: Include open-ended questions to allow participants to provide detailed and subjective responses, providinga deeper understanding of their experiences or opinions.3. Probing Techniques: Develop effective probing techniques to encourage participants to provide more in-depth and thoughtful responses. This can include follow-up questions or using prompts to elicit specific information.Conducting the Survey:1. Clear Communication: Clearly communicate the purpose of the survey and assure participants of the confidentiality of their responses. This will enhance participation rates and ensure accurate data collection.2. Active Listening: Actively listen to participants during the conversation survey to capture their thoughts and experiences accurately. Avoid interrupting or imposing personal biases that may hinder the collection of unbiased data.3. Note-taking and Recording: Take detailed notes during the survey and consider recording the conversation (with participants' consent) to ensure accurate data capturing. Transcribe recordings to allow for easier analysis later.Data Analysis:1. Thematic Analysis: Apply thematic analysis techniques to categorize and identify themes within the conversation survey data. This involves coding responses and identifying patterns or trends across the collected information.2. Comparative Analysis: Compare responses across differentparticipants or groups to identify similarities and differences. This helps reveal important insights and understand the broader perspectives.3. Validity and Reliability: Ensure the validity and reliability of the data by cross-referencing findings with other research methodologies or conducting member checks. This provides a more comprehensive understanding of the topic.Conclusion:Conversation surveys are a powerful research tool that allows for in-depth exploration and understanding of subjective experiences and opinions. By implementing the recommended practices in participant selection, survey design, survey conduct, and data analysis, researchers and practitioners can optimize the effectiveness and quality of their conversation surveys. Continuous improvement and adaptation based on feedback and evaluation will further enhance the value and impact of conversation survey methodology.。

2019-2020年高考英语语法提升《真题例析非谓语动词》专题讲解讲义非谓语动词作主语1．(2013·福建卷) ________basic first-aid techniques will help you respond quickly to emergencies.A．Known B．Having knownC．Knowing D．Being known解析：句意为：懂得基本的急救技能有助于对紧急情况作出快速反应。

本题考查非谓语动词作主语。

分析题干可知非谓语动词短语作主语，因为过去分词不能作主语，排除A项。

根据句意可知，空格处不是表达完成或被动意思，排除B和D项，故答案为C项。

答案： Cimportant for the figures________regularly.2．(2011·北京卷)It’ｓA．to be updated B．to have been updatedC．to update D．to have updated解析：考查非谓语动词作状语。

句中的时间状语regularly表示经常发生，因此排除B、D两项；又因为figures是被更新的，故选A项。

答案： A非谓语动词作状语3．(2013·湖南卷)________warm at night，I would fill the woodstove，then set my alarm clock for midnight so I could refill it.A．Staying B．St ayedC．To stay D．Stay解析：句意为：为了在夜里取暖，我总是把火炉填满，然后把闹钟定在午夜，以便我能再填满一次。

本题考查非谓语动词作状语。

根据上下文逻辑可知设空处表示目的，而四个选项中只有C项作目的状语，故答案为C项。

答案： C4．(2013·安徽卷)________in the early 20th century，the school keeps on inspiring children’ｓlove of art.A．To found B．FoundingC．Founded D．Having founded解析：句意为：这所学校建立于20世纪早期，并一直激励孩子们热爱艺术。

第34卷 第9期2011年9月计 算 机 学 报CH INESE JOURNA L OF COM PU TERSVo l.34N o.9Sept.2011收稿日期:2010-10-25;最终修改稿收到日期:2011-07-04.本课题得到国家自然科学基金(61073005,60803016)、国家/九七三0重点基础研究发展规划项目基金(2009CB320706)、国家/八六三0高技术研究发展计划项目基金(2009AA043401)和清华信息科学与技术重点实验室学科交叉基金项目资助.王建民,男,1968年生,博士,教授,博士生导师,中国计算机学会(CCF)高级会员,主要研究领域包括数据管理与信息系统、云环境中非结构化数据管理技术、业务过程与产品生命周期管理、软件保护与系统安全技术.E -mail:ji m wang@.余志伟,男,1985年生,博士,主要研究方向为软件混淆、软件水印、防篡改等软件保护技术.王朝坤,男,1976年生,博士,副教授,主要研究方向为数据管理、软件保护技术.付军宁,女,1985年生,硕士,主要研究方向为软件保护技术与云计算.Java 程序混淆技术综述王建民1),2),3)余志伟1),2),3),4)王朝坤1),2),3)付军宁1),2),3)1)(清华大学软件学院 北京 100084)2)(清华信息科学与技术重点实验室 北京 100084)3)(信息系统安全教育部重点实验室 北京 100084)4)(清华大学计算机科学与技术系 北京 100084)摘 要 软件混淆技术已经广泛应用于抵制逆向工程和重组工程.文中从混淆技术的历史发展角度对现有的混淆技术理论、算法、攻击模式和评估进行了综述,将Jav a 程序混淆算法分为类内混淆和类间混淆两个类别,并对其中的各类算法进行详尽的阐释.最后在现有工作的基础上,展望了软件混淆技术未来的发展与研究方向.关键词 程序混淆;软件水印;防篡改;软件版权保护中图法分类号T P 309 DOI 号:10.3724/SP.J.1016.2011.01578A Survey on Java Program Obfuscation TechniquesWANG Jian -M in 1),2),3) YU Zh-i Wei 1),2),3),4) WAN G Chao -Kun 1),2),3) FU Jun -Ning 1),2),3)1)(S choolof S of tw are ,T singh ua Univ er sity ,B eij ing 100084)2)(T sing hua L abor atory f or I nf or mation S cience and T echn olog y ,Beij ing 100084)3)(K ey L aborator y f or I nf ormation S yste m S ecu rity of M inistry of E duc ation ,Beij ing 100084)4)(Department of Comp uter S cience and T echnolog y ,T singh ua Univ er sity ,B eij ing 100084)Abstract Obfuscation techniques have been w idely applied in the defenses of reverse engineering and re -engineering attacks.From the view of the dev elo pment o f obfuscatio n,w e br iefly dis -cussed the principles,algor ithms,different kinds of attack appro aches and evaluating standard.Java prog ram o bfuscation algo rithm s can be divided into tw o types:One is obfuscation w ithin a class;the o ther is obfuscation betw een classes.Finally,based on the sur vey of these obfuscation techniques,the future research of Java prog ram obfuscation is also stated.Keywords obfuscation;so ftw ar e w atermark;tem p -pro ofing;so ftw are co py right protectio n1 引 言随着计算机软件的广泛使用,软件的安全问题严重威胁着软件产业的发展,主要表现为:软件攻击者获得试图攻击的软件备份之后,成功地破解软件.现有的调试和编辑工具,使直接检查或修改二进制程序代码变得更加容易.因此,通常情况下攻击者会重组或破解软件,导致软件的使用控制机制失效.破解后的软件可以非法向大众分发,更加助长了软件盗版问题.Jav a 程序由于平台无关性得以在Internet 上迅速传播和使用.但同时,Java 语言的这一特性也带来了保护知识产权方面的新问题.Java 语言为了支持平台无关性,采用了Class 这种字节码文件格式①.经过预编译之后,Java 源代码指令转化为字节码,而后在虚拟机上解释执行.由于Java 字节码的设计是为了使语言更简洁、更具有平台独立性和网络灵活性,它的指令集相对简单通用,每一个类编译成一个单独的文件,Class 文件保留方法变量名称等大量语义信息.这些特点都导致Java 字节码更容易被反编译为Java 源代码.攻击者通过反编译和反汇编技术,获得软件的全部或部分源代码,从而获取关键信息如核心算法、秘密信息等为自己所用.目前已有许多Jav a 的反编译工具,如Jad ②、M ocha [1]等,都是常用的针对Java 字节码文件的反编译器.因此,防止针对Java 字节码文件的静态分析和反编译攻击的要求,显得更加迫切.混淆技术可以被理解成为是一种特殊的编码技术,目前被广泛地应用于软件知识产权保护领域.通常所说的混淆是指对拟发布的应用程序进行保持语义的混淆变换,使得变换后的程序和原来的程序在功能上相同或相近,但是更难以被静态分析和逆向工程所攻击,如图1所示.图1 软件混淆技术应用本文从混淆技术的发展过程角度,对现有的混淆技术进行综述.首先,在第2节介绍Collberg 和Thom bor son 等人[2]最早提出的混淆技术的形式化定义;第3节介绍混淆算法的分类、各类算法的原理及其实现方法;第4节介绍针对混淆算法的理论研究以及攻击模式;第5节阐述针对混淆算法的评估度量;最后总结并展望混淆技术的前景.2 软件混淆技术基本原理软件③混淆技术是在1997年由新西兰Auck -land 大学的Co llberg 和Thom bo rson 等人[2]首先提出的.混淆技术属于一种重要的软件保护技术,其本质是利用程序等价变换的方法,将原始程序转变为语义等价、难以被理解或反编译的程序.同时,他们还阐述了对于混淆算法的分类及其评估方法,并分别就每一类混淆技术提出了若干算法.这些理论及算法对混淆技术的发展起到了提纲挈领的作用.在Collberg 和Thomborson 提出的理论中,混淆技术可以视为一种高效的程序变换技术,它将原始程序P 转换成新的程序P c .P c 与P 相比具有相同的外部行为,并且代码安全性能更强.其形式化定义如下:设T 是从原始程序P 到目标程序P c 的一个变换,P c =T (P ).如果该变换满足下列条件,就称P c =T (P )为一个软件混淆变换:①若P 无法结束或者以错误的状态结束,则P c 可结束也可不结束;②若P 结束,则P c 也必须结束并且产生和P 相同的输出结果;③对于完成某特定的计算任务,P c 比P 多消耗的运行时间在一个有限的范围内;④攻击者将P c 恢复为P 所耗费的时间大于将P 转换至P c 的时间.其中,P 为未经混淆的程序,即原始程序,而P c 为混淆后的程序.对于任意的合法输入,P c 与P 输出相同的处理结果,即满足程序语义等价性.3 软件混淆技术的分类Jav a 文件经过编译生成类文件.类文件分为不同的字段,包括魔数(magic)、版本(versio n)、常量池(co nstant pool)、访问标识(access flag s)、(this)类、(super )类、接口(interfaces)、域(fields)、方法(m ethods)和属性(attr ibutes).根据混淆算法针对的对象不同,可以将混淆分为类内混淆和类间混淆.类内混淆指混淆的作用范围在某个类文件内部,基于字节码的类内混淆作用对象主要是上述类15799期王建民等:Java 程序混淆技术综述①②③T he Java Virtual M ach ine S pecification.h ttp:///docs /books/jvms /Jad.h ttp://w w /jad.html为描述方便,本文对/程序0和/软件0不作区分.文件结构方法(metho ds)中的code字段.类间混淆指混淆作用于两个或者两个以上的类文件,实现的方式主要包括类合并以及类拆分等. 311类内混淆类内混淆实现的方式主要包括数据混淆、控制流混淆、切片混淆和针对特定工具的混淆.31111数据混淆数据混淆的原理是通过对常量、变量和数据结构这些程序的基本组成元素进行修改的方式,增大攻击者进行逆向工程的难度.数据混淆包括:变量存储和编码混淆、变量聚合混淆、顺序调整混淆、词法混淆以及移除注释和调试信息混淆.3111111变量存储和编码混淆(1)分裂变量混淆分裂变量混淆[2-3]将较为简单的数据结构或数据类型(例如int、boo lean等)分解为一些变量的组合,从而达到隐藏原始数据的效果.(2)将静态数据转换为与程序相关的数据以一个静态的字符串为例,可以通过混淆变换将其转化为一个函数或一段子程序,从而在程序执行的过程中通过调用生成相应的字符串[2].这种变换的实现并不困难,但应用时需要考虑以下这对矛盾的制约因素:①在常见的应用程序中都包含大量的静态数据,如果对所有的静态数据都进行混淆,将导致代码的执行代价和传输代价显著增加;②如果仅对关键的静态数据进行混淆,则为反混淆者寻找关键数据提供明显的提示.所以,如何把握①和②的平衡点,是需要在实践中解决的问题.(3)改变编码方式混淆和改变变量生命周期混淆改变编码方式混淆隐藏了真实的数据,使原程序更加复杂[2].下面是一个改变编码方式混淆的例子:用表达式i c=c@i+u替代整型变量i,其中c和u为整型常数.原程序:int i=1;w hile(i<1000){,A[i], i++}.混淆后的程序:int i=11;w hile(i<8003){, A[(i-3)/8],i=i+8}.此外,也可以把局部变量改变为全局变量,从而改变变量的生命周期.(4)变量替换改变变量的类型,例如将一个整型变量更改为一个整型对象.这种操作的目的在于,利用Java的自动回收垃圾机制及时清理一些变量的使用信息,从而增大逆向工程的难度[2].3111112变量聚合混淆(1)合并变量混淆合并变量混淆算法的原理是:在保持程序语义等价的前提下,将两个或两个以上的数值变量V1, V2,,,V n合并为一个变量V m[2].例如,两个32位整型变量可被合并为一个64位整型变量.为了增加混淆变换的弹性,可以考虑在程序中添加不影响原始变量值的伪操作.(2)数组重构混淆数组重构混淆有多种实现方式,主要包括数组分裂变换(将一个数组拆分为两个子数组)、数组合并变换、数组折叠变换和数组辗平变换[2].为了获得更好的效果,可以将某些上述变换组合使用. 3111113顺序调整混淆程序员在进行程序设计时,通常会把逻辑上相关的数据放在物理上相邻的位置.因此可以通过顺序调整把逻辑上相关的数据分散在物理上不相邻的位置,从而增大攻击者分析程序的难度.顺序调整混淆主要有重新调整实例变量顺序混淆、重新调整方法顺序混淆、重新调整数组顺序混淆[2].3111114词法混淆词法混淆的原理是通过对函数和变量的名称进行扰乱,使其违背见名知义的软件工程原则[4].这类混淆变换大多针对Java语言设计,其原理是依照Jav a虚拟机规范(T he Java Virtual M achine Spec-i fication)中有关类文件结构的规定,对常量池中存储类、域、方法和变量名字的/CON ST ANT-U tf8-info0类型数据项加以混淆.词法变换混淆具有单向的性质,且无额外的执行代价,在保护知识产权的实践中得到了广泛的应用.目前的Jav a字节码混淆器和混淆编译器大多支持这一功能,可见词法变换具有良好的可行性.为了提高混淆算法的隐蔽性,De A R和Van L O[5]提出了使用标识符交换来进行词法变换的方法,其思想是将标识符作交换,而不是将其更改成毫无意义的名字.标识符交换包括交换变量名、交换函数名和交换类名3个步骤.使用标识符交换进行词法变换提高了算法的隐蔽性,同时也保留了标识符,这使得该混淆算法不再具有单向性,为反混淆提供了可能性.Chan等人[4]在2002年提出了一种具体的基于修改和重用标识符的词法混淆算法.其基本的思想是尽可能多地对标识符进行重用,通过对标识符的多次重用来迷惑攻击者.此外,这种算法将复杂的标识符用简单的标识符代替,因此具有减小字节码文件体1580计算机学报2011年积的作用.3111115 移除注释和调试信息混淆程序的注释和调试信息在一定程度上为攻击者理解程序语义提供了便利,去除此类信息可以有效阻碍攻击者对程序的理解.移除注释和调试信息混淆的主要实现方法是去除注释和调试信息等源代码格式化信息,以及将原有标识符替换为无意义的标识符.这是一种单向的变换,源文件在经过混淆之后无法恢复[3].该混淆方法操作简单,但是强度较差,因为移除的调试信息量较小.这种移除信息的混淆方法对程序运行的时间和空间复杂度几乎没有影响[2].混淆器Cr em a ①是一个典型的应用实例.31112 控制流混淆程序的控制转换过程的信息是追踪定位程序状态的重要线索,如何保护这部分信息也是软件保护中很重要的一个环节.控制流图(Contr ol Flow Gr aph,CFG)是程序可能执行流程的图形化表示,它可以用来描述程序的控制转换.一个程序可以被分成由一系列无分支的代码组成的基本代码块,这些基本块作为控制流图的结点,而图的边即为各个基本块之间可能的跳转关系.控制流混淆的目的就是改变或复杂化程序的控制流,使程序更难以破译.控制混淆可采用的手段很多,比如应用不透明谓词增加伪造分支、加入可导致反编译错误的指令(例如在Java 字节码中添加go to 语句等)、将一段代码转换为内联函数调用等[2].3111211 控制计算混淆(1)通过不透明谓词加入不会被执行的分支语句对于一串语句S,利用不透明谓词可以构造出两种变换[6].第1种是增加恒真或恒假的不透明谓词,构造出一条不会被执行的分支,该分支上可以没有语句,也可以有和S 不同的语句;第2种是利用时为真时为假的不透明谓词,令它的两个分支上的语句都与原语句S 相同.其中第2种混淆变换能够较好地抵抗动态分析.(2)循环条件插入变换通过不透明谓词使一个循环的终止条件变得更为复杂,可以达到对循环进行混淆的目的[2].两种可能的变换原理如图2所示.其中f 和g 是判断谓词k ,j 的表达式.(3)将可化简的控制流转换为不可化简的控制流通过将一个结构化的循环变换为含有多个入口的循环,可以将可化简的控制流转换为不可化简的控制流[2].其原理如图3所示,主要是利用不透明谓词加入的假分支,生成循环交叉,从而使控制流图难以被反编译.(4)并行化代码并行化代码是编译器优化程序性能的重要方法之一[2].这一技术可以用于混淆程序控制流,从而增大攻击者理解程序的难度.具体地,有两种实现方式:①在程序中添加伪造的进程代码(dummy process );②将一组串行程序并行化.使用并行化代码算法,需要注意代码中的数据依赖关系,否则可能破坏程序的语义等价性.3111212 控制聚合混淆通过方法内联变换和将某些代码片段转换成为子程序,可以改变程序自身的控制流图.这种混淆方15819期王建民等:Java 程序混淆技术综述①Van V H.Crema -T he J ava obfuscator.http://w eb.inter.nl.n et/users/H.P.van.Vliet/crema.h tm l,1996法是单向的,具有较好的鲁棒性[2].(1)方法交叉调用和方法克隆其主要思想如图4所示,合并两个调用方法的参数列表和主体,并增加一个参数或全局变量(图4中的int V ),用以声明执行时具体应当调用哪一个方法.方法克隆的目的是混淆程序的方法调用关系,克隆的方法应与原方法看似区别但又具有相同的语义[2].(2)循环变换循环变换可以采用循环模块化(将循环区间划分为多个模块)、循环展开(拓宽循环的步频)和循环切割(将一个循环转换为多个循环)等方法实现.3111213 控制顺序混淆控制顺序混淆的实现原理为:随机打乱表达式、基本块、方法和类的顺序,使攻击者难以理解程序的正确意图.使用该方法要注意一些有依赖关系的数据在重新排序之后是否合法.class C { m ethodM 1(T 1a){S M 11;,;S M 1k ; }m ethodM 1(T 1b;T 2c){S M 21;,;S M 2m ;}}{C x =n ew C;x .M 1(a);x .M 2(b,c );}class C c {m ethod M (T 1a;T 2c;int V){if (V ==p ){S M 11;,;S M 1k ;}else {S M 21;,;S M 2m ;} }}{C c x =new C c ;x .M (a,c,V =p );x .M (b,c ,V =q );}图4 方法交叉调用对于控制混淆如何抵抗自动化分析工具的攻击,也有不少研究.W ang [7]提出了通过更改跳转指令实现控制混淆的一种方法,使得编译时候每一个跳转指令的目标都是未知的.Chow 等人[8]提出了一种相似的方法,通过一个有限状态自动机来控制程序的跳转语句.根据以上的控制混淆算法理论,Low [9]在1998年提出了针对Java 代码的控制混淆算法实现.对于不透明谓词的设计,Majumdar 等人[6]在2006年为分布式系统的混淆补充了一种分布式不透明谓词.应用该不透明谓词可以有效地抵抗多种自动静态分析的攻击.31113 切片混淆切片通常是用来帮助理解程序的,而混淆的目的是使程序更难以被理解.Dr ape 等人[10-11]提出了切片混淆算法,使得混淆过的程序能够更好地对抗切片分析攻击.定义Slice (P ,S,V)用来标记程序P 在状态S 时,对应的变量集V 的一个后向切片(backw ards slice ),指令out 表明V 中变量的输出.图5中显示了一个包含有变量i,x,y 的程序,当输出为y 时的程序切片:黑体部分为切片时可观测到的变量值.此时变量x 不在观测范围内.切片混淆(slicing obfuscatio n )算法的主要思想就是尽可能多地将多个变量的值放入到切片的观察范围之内,增加使用切片分析程序的攻击者的困难int s umprod (int n ){int i =0;int x=0;int y =1;while (i <n ){i++;x=x+i;y =y *i;}ou t (x );out (y );}图5 原始程序o ut (y )时的切片程度[11].切片混淆的主要方法有:增加恒假谓词、变量编码和增加循环变量.增加恒假谓词是在恒假谓词的假分支上增加令x 与y 相关的函数;变量编码是在不改变语义的情况下将y 的表达式重新编码为与x 相关的表达式;增加循环变量是在循环变量中添加与x,y 相关的变量.图6至图8分别描述了这几种方法的实现.int bogu s (in t n ){int i =0;int x =0;int y =1;while (i <n ){i ++;if (i <5||x <y )x =x +i ;else y =x *i ;y =y *i ;}ou t (x );out (y );}图6 增加伪造分支int varEnc (int n ){int i =0;int x =0;int y =1;while (i <n){i ++;x =x +i ;y =(y +i -x)*i +x;}ou t (x );out (y -x );}图7 变量编码1582计 算 机 学 报2011年int addVar (int n ){int i =0;int x =0;int y =1;int j =2;while (i <n &&j >0){i ++;x =x +i;y =y @i;j =j +y -x;}out (x );out (y );}图8 增加循环变量Majumdar 等人[12]从单词计量、结果求和、定位时间等角度验证了切片混淆的方法能够增强程序抵抗切片分析攻击的能力.31114 针对特定工具的混淆针对特定工具的混淆是针对自动化的反编译和反混淆工具提出的混淆方法,旨在阻止这类自动化工具的使用,或者增加其使用难度和代价[2].该类混淆方法与前面提到的3种类别的混淆,最大的不同之处在于,前3种混淆技术主要是针对程序的阅读者,即针对的是人;而后者针对的是自动化的反混淆以及反编译工具.针对特定工具的混淆可以通过与其他的混淆技术综合使用从而达到更好的鲁棒性[2].在实际操作中,可以在循环体内部加上伪造的无用变量,形成数据依赖,增加反混淆的自动工具分析难度.例如在程序的某处使用了反向循环的混淆,为了防止这一变换被反编译器分析破解,可以采用图9所示方法:在这个例子中,增加了B 数组,for 循环的前后向顺序会对B 数组的各个元素产生影响,从而形成数据依赖,阻止反编译器变换循环的方向.这一混淆方法的弹性,取决于添加伪造数据的复杂度.For (i=0;i<=10;i++){ A[i]=i;}int B [50];For (i=10;i>=10;i--){A[i]=i ; B [i ]+=i @i /5;}图9 增加伪造的数据依赖例如,针对Mo cha[1]这一反编译工具,我们可以采用这样的措施造成它反编译失败:在每个方法的每一条返回语句之后增加额外的指令.在不影响程序行为的前提下,造成M ocha 的崩溃.Batchelder 等人[13]在2007年较为详细地阐述了如何利用Java字节码与Java 源代码之间指令规范的不同,在字节码的级别进行可以导致反编译失败的控制流混淆.例如利用jsr -ret 指令取代普通的if -goto;针对模式匹配的反编译器,将lo ad 指令提到if 指令之前;在子类构造函数中对父类的调用进行外包等等.312 类间混淆Sosonkin 等人[3]在2003年提出了3种分别基于类的合并、类的拆分以及类型隐藏的类间混淆算法.31211 类合并类合并算法的主要思想是合并两个或多个类各自包含的变量和函数,根据需要重命名变量或函数标识符[3].若待合并的类中有标识符相同的变量或非构造函数,则将其改为不重复的新标识符;若待合并的类中有标识符及参数都相同的构造函数,无法随意更改标识符,则合并后为其中一个增加伪造的参数;若待合并的两个类之间有继承关系,则合并后增加一个布尔型的私有变量用于区分标识符相同的函数,图10是类合并的一个例子.Original classes :class A {private int i; public A(){ i=5; }public b oolean m(){ return i <0; }}class B extends A { public B(){ super(); i=10; }public b oolean m(){ return i <10; }}class C {void n(){ A a; if (,){a=n ew A(); }els e {a=n ew B(); }a.m (); }}Obfus cationObfuscated class es :clas s AB { private int i;private boolean isA; public AB(){ i=5;isA=tru e; }public AB(int j){ this(); i=10;isA=false; }public boolean m(){ if (is A){ retu rn i <0; }els e {retu rn i <10; } }}clas s C { void n (){ AB a; if (,){a=new AB(); }els e {a=new AB(38); }a.m(); }}图10 一个类合并算法的实现31212 类拆分类的拆分算法首先分析判断待拆分的类内部的继承和方法调用关系[14],观察它们是否能够进行拆15839期王建民等:Java 程序混淆技术综述分.对于可拆分的类,主要方法是将一个类C 拆分为C1,C2两个类,且C2是C1的子类,并确保C 中的每个方法和变量,在C2中均包含或由C1继承,图11是类拆分的一个例子[3].Original classes :clas s C {private int i; private dou ble d;protected Ob ject o;public C(){i=5;d=110;o=new Object();}public C(int iarg,double darg){i=iarg;d=darg;o=new Object();}public boolean m1(){retu rn i <0;}public void m2(){d=310; m3(3);} protected void m3(int iarg){ i=iarg;m4(n ew Object());}public voidm4(Object obj){o=obj;}}clas s D { void n (){ C c=new C();if (c.m 1){,}c.m2;c.m4;}}Obfuscation Ob fus cated classes:class C1{ private int i;private double d; public C1(){ i=5; d=110; } public C1(int iarg, dou ble darg){ i=iarg; d=darg; } public b oolean m 1(){return i <0;}protected voidm3(int iarg){i=iarg;m4(new Object());}public void m4(Ob ject obj){o=ob j.getClass(); }}class C2extends C1{protected Object o;public C2(){super();o=n ew Ob ject();}public C(int iarg,dou ble darg){super(iarg,darg); o=n ew Ob ject();} public void m2(){ d=310; m3(3); } public void m 4(Ob ject obj) o=ob j; }}class D { void n(){C2c=new C2(); if (c.m1){,} c.m2; c.m4; }}图11 一个类拆分算法的实现31213 类型隐藏类型隐藏算法是利用Jav a 中的接口来声明待混淆类的变量和方法,而原先的类则实现这些接口,具体实现无需改动[3].算法的关键部分是每一个接口只随机包含一个待混淆类中公开的方法的子集.除类合并算法之外,另外两种混淆算法对程序运行时间的影响较小.4 软件混淆算法的攻击模式411 混淆算法的理论研究Bar ak 等人[15]曾经证明,如果把混淆的过程看作是一个/虚拟黑箱0(virtual black bo x ),即要求被混淆过的程序的输入输出应当完全相同于原程序.满足这一性质的混淆器是不可能实现的.然而/虚拟黑箱0的要求过于严格,在较宽泛的限制中,混淆器是可以实现的.Lynn 等人[16]在Barak 的证明基础上给出关于程序混淆第一个/积极的0结果:提出了几种针对复杂的访问控制函数的混淆方法,并进行了验证.Wee 等人[17]也针对/虚拟黑箱0理论,阐述了关于点函数的混淆,证明在黑箱条件下对点函数混淆的存在性.关于混淆技术理论,还存在较大的研究空间.412 混淆算法的攻击模式41211 针对数据混淆算法的攻击假定T 是对程序P 的一个单向的混淆变换,当且仅当从原始程序P 除去某些信息后,无法通过混淆后的程序P c 恢复出P [18].词法变换是最典型的不可逆混淆算法.虽然对于经过词法变换的程序进行攻击不可能恢复程序的原貌,但只要理解程序各个模块的含义就可能对程序产生威胁.例如根据无法被混淆的系统API 名称等关键字,攻击者可以推测出该模块的大致功能.Cimato 等人[19]提出基于字节码中的标识符修正的反混淆方法.该方法可以用于攻击标识符重命名的混淆算法.41212 针对控制流混淆算法的攻击目前,针对控制流混淆算法的主要攻击方法是动态分析.U dupa 等人[20]指出了动态分析对大部分混淆算法的攻击作用,并阐述攻击模型.对于控制流混淆而言,变换后生成的程序中若存在始终不执行的分支,通过动态分析就能找到对破解程序有用的信息.针对控制流混淆,还有黑盒测试攻击[21],该方法通过对程序进行黑盒测试,了解各个类及其函数的功能,从而获取攻击者需要的信息.这种方法对大多数的混淆变换均能加以攻击.同时,该方法也存在着一定的局限性:黑盒测试缺乏自动分析工具,需要依靠大量的人力来完成分析工作.同时,Udupa 等1584计 算 机 学 报2011年。

山东省泰安市2023-2024学年高二上学期期中考试英语试题学校:___________姓名：___________班级：___________考号：___________一、阅读理解Four Off-the-Beaten-Path DestinationsDogon Rock City, Mali, AfricaDogon Rock City is an ancient city attached to vertical cliffs, housing over 5000 caves and 2000 houses. With its unique architecture and history, it is a must-visit destination for those seeking an adventure filled with history and culture. Visitors can witness the splendour and richness of this ancient civilization as they wander through its maze (迷宫) of courtyards and corridors.Salar de Uyuni, BoliviaSalar de Uyuni is the largest salt flat in the world, covering an area of over 12,000 square kilometers. It is a perfect place for those seeking a unique and peaceful natural environment. The flat is also famous for its beautiful salt bridges and salt hotels, which are built on salt lakes. Here, visitors can also see the world’s largest mirror and take unforgettable photos. Ushuaia, ArgentinaThe remote town of Ushuaia is located on the Beagle Channel and is the southernmost city in the world. It is the perfect destination for those seeking to explore Antarctica without breaking the bank. Here, visitors can also witness incredible wild up-close, such as penguins and seals.Ba Na Hills, VietnamBa Na Hills is an amazing alpine (高山的) resort with lush green mountains and charming views. Rising to a height of 1487 meters, it is a perfect destination for those seeking a break from Vietnam’s busy cities. Ba Na’s unique culture and cuisine provide an enriching experience for those seeking something different from Vietnam’s more famous attractions. Visitors can also take cable car rides to the top of the mountain and explore the beautiful scenery below.These off-the-beaten-path destinations, each with their own unique features and charm, provide an enriching travel experience for those seeking something different.1．What do Dogon Rock City and Ba Na Hills have in common?A．People there live a busy life.B．Many tourists go there every year.C．They both have characteristic culture.D．Visitors can take cable car rides to mountains.2．Which place fits you best if you’re expert at sea animals?A．Dogon Rock City.B．Salar de Uyuni.C．Ushuaia.D．Ba Na Hills.3．How is Salar de Uyuni different from the other destinations?A．It has breathtaking views.B．Tourists can enjoy cuisines made of salt.C．Tourists can be attracted by its splendid civilization.D．It offers tourists accommodation made of special materials.Cui Chenxi competes in the final of the women’s street skateboarding event during the Asian Games in Hangzhou on Wednesday.The teenager, who is the Chinese national team’s youngest athlete, won gold by scoring242.62 points. Until Wednesday, the title of China’s youngest Asian Games gold medalist be-longed to 15-year-old skateboarder Chen Ye. Nevertheless, the youngest participant at the Games is Mazel Paris Alegado, a 9-year-old Filipino skateboarder.Asked if she was nervous before the competition, Cui shook her head and said: “My dad told me to stay relaxed, enjoy the competition and showcase my skills. As long as I give it my all, I have no regrets.”Skateboarding, which is rooted in street culture, is a sport that emphasizes freedom and personal techniques. Cui only took up skateboarding in 2020 when the COVID-19 pandemic did not allow her to practice rollerblading (滚轴溜冰), which she began as a 3-year-old. Her father suggested that she try skateboarding at home instead. “I remember the day I stepped onto the board for the first time. It felt incredibly smooth. It was love at first sight,” Cui said.Two years later, she ranked third in the women’s street event at the Shandong provincial games. She soon joined the provincial skateboarding team and underwent a more comprehensive (全面的) and structured training program.In skateboarding, injuries are unavoidable, but when discussing the pain and the setbacks, Cui displayed a level of resilience beyond her age. “When I started skateboarding, I thought it was great fun. Later, I realized people suffer a lot of injuries while skateboarding. But that iswhere the spirit of skateboarding lies,” she said. Most of the time, getting injured is just part of the process, and she simply pushes through it and never gives up, she added.Cui is “the future of Team China”, according to Zeng Wenhui, the silver medalist. “I think she is quite excellent. We will stand together in moving forward in the future,” she said. 4．Who is the youngest athlete taking part in the Asian Games?A．Chen Ye.B．Cui Chenxi.C．Mazel Paris Alegado.D．Zeng Wenhui.5．What can we infer from the passage?A．Cui’s father has a positive effect on her.B．Cui’s fellow athletes don’t get along well with her.C．Cui spent a hard time from rollerblading to skateboarding.D．Cui received a comprehensive training program during the pandemic.6．What’s the spirit of skateboarding according to Cui?A．Life is movement.B．No cross, no crown.C．Never too old to learn.D．Freedom is a part of the essence of life. 7．What does the underlined word “resilience” refer to in paragraph 6?A．Calmness.B．Resistance.C．The ability to adapt.D．A state of feeling depressed.Ann has been severely paralysed (瘫痪)for more than 18 years. She cannot speak and normally communicates. But now, she has been able to speak through an image on the computer using technology that translated her brain signals into speech and facial expressions.The advance raises hopes that the technology could be on the edge of transforming the lives of people who have lost the ability to speak due to conditions like strokes (中风).The latest technology uses tiny electrodes (电极) put on the top layer of the brain to detect electrical activity in the part of the brain that controls speech and face movements. These signals are translated directly into a digital avatar’s (化身) speech and facial expressions including smiling, frowning or surprise.The team put 253 paper-thin electrodes on the top layer of Ann’s brain over an area important to speech. Afterwards Ann worked with the team to train the system to detect her unique brain signals for various speech sounds by repeating different phrases repeatedly. Thecomputer learned 39 distinctive sounds and a Chat GPT-style language system was used to translate the signals into understandable sentences. This was then used to control an avatar with a voice personalised to sound like Ann’s voice before the injury, based on a recording of her speaking at her wedding.The technology was not perfect, translating words incorrectly 28% of the time in a test run involving more than 500 phrases, and it generated brain-to-text at a rate of 78 words a minute, compared with the 110–150 words typically spoken in natural conversation. However, scientists said the latest advances in accuracy and speed suggest the technology is now at a point of being practically useful for patients. A crucial next step is to create a wireless form that could be put beneath the skull (颅骨).8．How can Ann speak?A．By tiny electrodes on her brain.B．Through an avatar using the latest technology.C．Through repeating different phrases repeatedly.D．By a technology that detects her brain signals.9．What is paragraph 3 mainly about?A．How the technology works.B．How the technology benefits the patients.C．How the technology translates the brain signal.D．How the technology is put into wide application.10．Which is unnecessary to make the avatar sound like Ann?A．Ann’s cooperation.B．Ann’s facial expressions.C．An AI language system.D．A recording of Ann’s speaking. 11．What do we know about the technology according to the last paragraph?A．It is a promising way to help patients.B．Its translating speed is too slow.C．It cannot be applied to patients.D．It is not accurate in translating at all.Fed up with tourists who only come in to check out the scenery, the owners of Queviures Murria, a grocery store in Barcelona, Spain, have put up a sign announcing a 5 euro fee for coming in just to look.Queviures Murria has been in business since 1898 and is one of the most eye-catching locations in all of Barcelona. The vintage (老式的) look of the store and the traditional design of the inside attract hundreds of tourists every day, but the problem is that many of them aren’t actually interested in the products sold inside—all kinds of traditional Spanish cured sausages, cold cuts, cured cheeses, oils, wines, etc. According to the store, many tourists enter the store without even saying a simple “hello”, wander around taking pictures, and then simply leave without buying anything. So it decided to charge people for “just looking” just to discourage them.Toni Merino, one of the managers at Queviures Murria, said that the idea to charge tourists a fee for simply coming into the grocery store and looking around without buying anything started out as a joke, but as the staff and the regular customers became increasingly inconvenienced by random tourists looking around, taking pictures, with absolutely no interest in the products on display, they decided to do it.Merino has clarified that he and his partners have no interest in actually charging the fee, and they don’t really have to anymore. By posting a warning about the 5 euro fee in the shop window, they have managed to discourage most unwanted visitors.After photos of the 5-euro fee announcement in the window of Queviures Murria spread quickly on social media, the Barcelona business faced a public outcry, and it was forced to explain the reasons and clarify that it didn’t really plan on carrying on the measure. 12．What can we learn about Queviures Murria?A．Tourists are not welcome there.B．It has a history of nearly 100 years.C．It charges people for taking photos.D．Tourists prefer its design to the products sold there.13．Why did the grocery store put up the sign?A．To earn a fame.B．To attract more tourists.C．To warn the impolite visitors.D．To provide a convenient shopping environment.14．What’s the public’s opinion on the measure?A．Opposed.B．Objective.C．Supportive.D．Unconcerned. 15．Where is this text probably taken from?A．A travel journal.B．A news website.C．A fashion magazine.D．A debate program.二、七选五As the new semester rolls in, it’s time to get ready for a fresh start to achieve your goals.16 Here are some practical tips to help you get through a fresh start with ease.Evaluate how you did last semester. Grades aren’t everything, but they can help you understand how well you’re doing in terms of your studies. So start by taking some time to think about these things. Did your grades improve, stay the same, or drop last semester? How can you make changes this semester? 17Set new goals for this semester. Before the semester begins, take some time to set clear and achievable goals for yourself. 18 Having goals will provide you with directions and motivation (动力) throughout the semester. To make the most of it, write your goals and hang them in a visible place.19 Keep your motivation high by reminding yourself of the goals you set. Then surround yourself with supportive people who inspire and encourage you. Celebrate your achievements along the way to maintain a positive attitude. Remember to stay adaptable and make adjustments to your plan as needed throughout the semester.Take care of yourself. Remember to prioritize (优先考虑) your physical and mental health. 20 Practice stress management skills, or engage in hobbies that help you relax. Taking good care of yourself will improve your focus, energy and overall productivity.A．Stay motivated and stay positive.B．Take an active part in various activities.C．Get enough sleep, eat nutritious meals, and exercise regularly.D．Identify what you want to finish academically and personally.E．It will save you time and help you communicate with your teachers.F．You’d better make the most of this time because starting off strong is the key to a successful semester.G．It’s helpful to write these things down, allowing you to make connections and take the next step more easily.三、完形填空people spent so much time at home during COVID-19,” says 28-year-old Li Jianlin, wholovers in China, and it 23 online fitness programs.“Many young people go to the gym, or workout regularly. 24 , a phased (阶段性的) and slow approach to recovering exercise after COVID-19 is the best way to 25 pre-COVID-19 fitness levels. So baduanjin, which uses breathing and concentration skills to improve body and mind, through eight well-designed results, has won the hearts of the young during 26 ,” says Li.Li, who graduated from the Capital University of Physical Education and Sports with a 27 in aerobics (有氧运动) in 2018, has 28 baduanjin exercise programs for the platform’s users, which were 29 earlier this year. Available in three different levels, the programs have received 30 feedback from users.“Seventy percent are young people, under the age of 30. The programs give detailed 31 and demonstrate the movements from different angles (角度),” says Li.Keep’s baduanjin programs have been 32 more than 5.41 million times, and about 36,000 users have added them to their 33 . “Since it also involves meditation (冥想) and 34 exercises, it’s also regarded as a way for young people to relax and deal with 35 ,” he adds.21．A．apparently B．especially C．ultimately D．slightly 22．A．creator B．visitor C．inventor D．governor 23．A．detects B．displays C．offers D．reveals 24．A．Therefore B．Hence C．Otherwise D．However 25．A．regain B．rebuild C．retell D．remember 26．A．fever B．vacation C．recovery D．treatment 27．A．class B．subject C．game D．major 28．A．found B．produced C．discovered D．switched 29．A．cancelled B．selected C．launched D．responded 30．A．warm B．realistic C．worthwhile D．abnormal 31．A．conduction B．plans C．materials D．instructions32．A．surveyed B．observed C．watched D．witnessed 33．A．favorites B．habits C．notes D．memories 34．A．singing B．jumping C．dancing D．breathing 35．A．stress B．anger C．doubt D．leisure四、用单词的适当形式完成短文阅读下面短文，在空白处填入1个适当的单词或括号内单词的正确形式。

2024年高考英语模拟试卷注意事项：1．答卷前，考生务必将自己的姓名、准考证号、考场号和座位号填写在试题卷和答题卡上。

用2B铅笔将试卷类型（B）填涂在答题卡相应位置上。

将条形码粘贴在答题卡右上角"条形码粘贴处"。

2．作答选择题时，选出每小题答案后，用2B铅笔把答题卡上对应题目选项的答案信息点涂黑；如需改动，用橡皮擦干净后，再选涂其他答案。

答案不能答在试题卷上。

3．非选择题必须用黑色字迹的钢笔或签字笔作答，答案必须写在答题卡各题目指定区域内相应位置上；如需改动，先划掉原来的答案，然后再写上新答案；不准使用铅笔和涂改液。

不按以上要求作答无效。

4．考生必须保证答题卡的整洁。

考试结束后，请将本试卷和答题卡一并交回。

第一部分（共20小题，每小题1.5分，满分30分）1．—I will be a vice president in a year or two.—You can’t be serious!_______.A．I can’t make it B．I can’t help it C．I won’t tell a soul D．I wouldn’t bet on it2．Tom’s comments on this issue are confusing because they appear to ______the remarks on the same issue made earlier by him.A．violate B．induce C．clarify D．contradict3．一When he know the result of today’s job interview?一In a couple of days．A．should B．may C．shall D．must4．While working in Kunming, he checked the weather each morning for months ________he realized it would be the same every day.A．when B．afterC．before D．since5．_______, I have never seen anyone who's as capable as John.A．As long as I have traveled B．Much as I have traveledC．Now that I have traveled so much D．As I have traveled so much6．一Excuse me，sir．You can’t enter0ffice without permission．一But the manager is expecting me．A．the；a B．an；the C．the；不填D．不填；不填7．--- Mom, can you give me an extra 200 yuan a month?--- Son, we have just bought a house, and from now on we need to practise strict .A．economy B．medicine C．self-control D．patience8．My mom once worked in a very small village school, which is__________only on foot.A．acceptable B．adequate C．accessible D．appropriate9．I hope that we will be able to make it through the tough times and back to the business of working together ________ our common goals.A．on behalf of B．in honor of C．on top of D．in search of10．Class Two, our class became the Basketball Champion of our school.A．Beating B．to beat C．Beaten by D．Having beaten11．________ your blog, I would have written back two days ago.A．If I read B．Should I readC．Had I read D．If I could have read12．At school, it is essential that every child ______ equally regardless of family background.A．treating B．treated C．be treated D．is treated13．If we use the new recycling method, a large number of trees .A．are saved B．will save C．will be saved D．have saved14．As to the long-term effects of global warming some believe that the damage has been done,______________________.A．otherwise we take steps to make up nowB．now that we take steps to make upC．whether we take steps to make up now or notD．unless we take steps to make up now15．The old woman who ________ in the deserted house alone for ten years has been settled in a nursing home now. A．lived B．has livedC．had lived D．has been living16．Sometimes, the kind of food we serve a person suggests ________ we show our gratitude.A．when B．whatC．why D．how17．—I'm going to order chicken and salad．What about you?—.I'll have the same.A．I'm afraid not B．It's up to youC．That sounds good to me．D．That depends18．Peterson, a great archaeologist, said: “Archaeologists have been extremely patient because we were led to believe that the ministry was ________ this problem, but we feel that we can't wait any longer.”A．looking out B．bringing out C．carrying out D．sorting out19．I thought it hard to complete the project then, but I ________ my mind.A．will change B．would changedC．have changed D．had changed20．Was it at the beginning _____ you made the promise ____ you would do all to help make it?A．that; that B．when; thatC．that; when D．when; when第二部分阅读理解（满分40分）阅读下列短文，从每题所给的A、B、C、D四个选项中，选出最佳选项。

I.J. Modern Education and Computer Science, 2019, 8, 42-47Published Online August 2019 in MECS (/)DOI: 10.5815/ijmecs.2019.08.05A Machine Learning based Approach for Mapping Personality Traits and Perceived Stress Scale ofUndergraduate StudentsAhmed A. MaroufDepartment of Computer Science and Engineering Daffodil International University (DIU),Dhaka-1207, Bangladesh.Email: marouf.cse@.bdAdnan F. Ashrafi, Tanveer Ahmed, Tarikuzzaman EmonDepartment of Computer Science and Engineering Stamford University Bangladesh (SUB),Dhaka, Bangladesh.Email: {adnan, tanveer.sub.094419, emon.cse}@.bdReceived: 02 May 2019; Accepted: 26 May 2019; Published: 08 August 2019Abstract—This paper focuses on the personality traits of students and stress scale they had to face in undergraduate level. With the advancement of computer science and machine learning based applications, we have tried to inter-correlate the terms. In the area of computational psychology, it is important to understand participants’ psychological behavior using personality traits and predict how he/she is going to react on a certain level of the stressed situation. For the experiment, we have collected data of around 150 participants. The personality traits data are collected using the standard survey named The Big Five Personality Test created by IPIP organization and stress scale measurements are collected using scale devised by Sheldon Cohen named as Perceived Stress Scale hosted by Mind garden. The data are taken from Bangladeshi computer science undergraduate students and kept anonymous. In this paper, we have applied nine different machine learning based classification models are built for mapping the traits with stress scales. For performance evaluation, we have utilized precision, recall, f1-score, and accuracy. From the experimental findings, we found that Sequential Minimal Optimization (SMO) and k-NN classifier gives the highest prediction accuracy which is approximately 70%.Index Terms—Big Five Personality Traits, Perceived Stress Scale (PSS), Machine learning Approach, Data mining, Sequential Minimal Optimization (SMO).I.I NTRODUCTIONUnderstanding human psychology is considered to be one of the internal or back-end features of modern computer applications. Nowadays, predicting the individuals personality traits are considerable feature in many applications such as personalized e-commerce sites, personalized mobile applications. These types of personalization system are performed based on some of the personal attributes given by users’ such as demographic attributes (gender, age, previously made choices etc.). Therefore, understanding the personality computationally and incorporating it with the system may give higher user-friendliness to users’. In the context of educational data mining and computational psychology, it is evident that understanding the personality traits earlier may give benefits to predict the future behavioral actions of particular participant.The personality of a human cannot be declared without any proper justification or experimentation. According to the Big Five Personality Traits [1], there are five traits to identify the overall personality of any individuals named Openness-to-experience, Extraversion, Agreeableness, Neuroticism and Conscientiousness. Predicting each of the traits separately may lead towards the overall prediction of personality. Using IPIP [1] questionnaire, the insights of each traits could be understood. And, perceived stress scale [2, 3] measurement is based on the different types of stress people usually feel. The different stress level namely low, moderate and high perceived stress is determined using the survey. The survey questions are based on the last month’s perceived stresses in different category such as personal, social, family and professional.The motivation of this paper is to find the correlation between the undergraduate students’ personality and their perceived stress scale during the undergraduate semester cycles. It is to be mentioned that the undergraduate students of private universities of Bangladesh usually have four months semester and each semester contains the midterm and final exam along with the class tests and assignment submissions. Therefore, a regular student have to be quite active and educational stress happens as well as personal life stress includes. How can we predict that a particular students may fail in any course or may not perform well in any course? To find the answer of this question, we have investigated the personal or non-educational life stress scale of each individuals in this paper. The prediction system may predict which types of students are tend to do better and which are tend to perform poor in their upcoming activities.In this paper, we have utilized the machine learning algorithms to map the general personality traits and perceived stress scale of undergraduate students. We have considered the most popular survey forms for collecting both the personality traits and stress scales. For experiments, we have utilized the existing machine learning techniques such as data sampling, classification algorithms. Among the various classification methods we have applied Naïve Bayes (NB), Decision Tree (DT), Simple Logistic Regression (SLR), Random Forest (RF), Linear Discriminant Analysis (LDA), Bagging (BA), Multilayer Perceptron (MLP), Sequential Minimal Optimization (SMO) and k-Nearest Neighbor (k-NN). The methods could be utilized in educational data mining approaches and computational behavior finding.For the rest of the paper, section II discusses on the related works in the literature, section III summarizes the machine learning methods used in this paper. In section IV, the data collection procedure is described. Section V and VI comprises the experimental methods and results. The insights and mapping between the traits and stress scale are discussed in section VII and finally concludes in section VIII.II.R ELATED W ORKSIn this section, we have summarized the novel works from existing literature related to our presented work. Identifying personality traits using questionnaire and setting labels according to it is performed by many researchers [4, 5, 6]. The authors have tried to extract relevant features from the textual data of social media and classify the personality traits accordingly. International Personality Item Pool (IPIP1) are the items or questions to be answered to devise a scoring mechanism for traits identification. For identifying the traits a set of questions are used, as the test takers have to answer the questions honestly and individual score of each question are set to get the total trait score. Depending on the number of items in the survey there are 100 IPIP, 50 IPIP even 20 IPIP. Depending on the behavior of test taker on different issues of practical life, these items are set by psychologists. Different ground truth datasets are labeled by researchers for computational prediction tasks [7, 8] such as myPersonality dataset. These dataset are collected from social media and been utilized by others for devising prediction methods.Linear Dirichlet Allocation (LDA) [9], Linguistic Inquiry and Word Count (LIWC), Term Frequency - Inverse Document Frequency (TF-IDF) [9], deep learning methods [6], Parts-of-Speech (POS) tags, Speech Acts, Sentiment feature [10] based algorithms are presented by the respective researchers.In recent years, the impact of social media is significant and computer science as well as psychology researchers are making essential contributions in this area. The use of behavioral features are been integrated in social media such as users behavior based community recommendation [21], strong friends based friend recommendation frameworks [22]. But, none of the work shows the relationships on the perceived stress of the user and his/her personality traits. Therefore, this pitfall has become our main opportunity to contribute.Perceived Stress Scale (PSS) [2, 3] is one of the most widely used psychological instruments for determining the perception of stress. The degree of stress depends on the items designed which are justifying how unpredictable, uncontrollable and overloaded respondents find their lives. The levels of stress are low, moderate and high which should be influenced by regular hassles, significant events and predictive validity of PSS which is usual to fall rapidly after four to eight weeks. The evaluation of perceived stress scale among the UK university students has been documented by A. Denovan in [23]. The study has been examining the factor structure, composite reliability, convergent validity and gender invariance among the 524 UK university students and four different factor based models are analyzed. In another work, the Hindi version of Depression, Anxiety and Stress Scale -21 (DASS-21) survey has been taken from the Hindi speaking head-neck cancer and oral potentially malignant disorder patients in [24]. K. Kumar et al. [24] justified the reliability and psychometric validity of the DASS-21. Moreover, the identification of negative personality traits from user-generated contents is significant as the neuroticism rate of a person may reflect his/her person greatly. In [24] a machine learning based approach has been adopted to detect neuroticism rate of a person.In this paper, we have tried to depict the machine learning methods to map the traits with the PSS.III.M ACHINE L EARNING T ECHNIQUESThe machine learning techniques used in this paper are Naïve Bayes (NB), Decision Tree (DT), Simple Logistic Regression (SLR), Random Forest (RF), Linear Discriminant Analysis (LDA), Multilayer Perceptron (MLP), Bagging (BA), Sequential Minimal Optimization (SMO) and k-Nearest Neighbor (k-NN). In this section we have summarized the above mentioned techniques. Naïve Bayes (NB) classifier [11] is based on probabilistic approach and this classifier has been used in various areas with variants. Relying on the Bayes’ Theorem NB has no sophisticated unvarying parameter estimation.As the name suggests, for decision making decision tree (DT) [12] is applied in the context of data mining classifications. Decision tree has proven beneficial in case of numerical and categorical data as it is possible to validate a training model using statistical tests. Random Forest (RF) is a decision tree based ensemble classifier [13] applied in our work. RF is intrinsically suited for multiclass problems which also work fine with mixture of numerical and categorical features. RF internally builds separate multitude decision trees whiletraining and it outcomes the class that is the mode of the classes and/or mean regression of each tree.Simple logistic regression (SLR) [14] is a linear logistic model using LogitBoost algorithm. As LogitBoost uses a symmetric model, a sufficient number of iteration is performed in simple logistic regression to train the model. Built-in attribute selection is performed in SLR as an additional advantage. Therefore, for our experiments, we have applied this classifier.Linear Discriminant Analysis (LDA) [15] is a generalization of Fisher’s linear discriminant, which is widely used in statistics, pattern recognition and machine learning. LDA works well when the measurements are not dependent variables for separate observations. Bootstrap Aggregating, in short, Bagging (BA) [16] is an ensemble meta-algorithm which is based on statistical classification and regression. While performing it reduces the variance and avoids overfitting.Multilayer Perceptron (MLP) [17] is a version of feed- forward artificial neural network. Consisting minimum three layers namely input layer, hidden layer and output layer, where the hidden and output layers utilizes nonlinear activation function. MLP is proved to be useful to solve problems stochastically, which usually solutions to complex problems.Support vector machine (SVM) has become one popular yet essential classifier used in various sections of data mining such as, medical data mining, image data processing, bioinformatics etc. Though it is proven to work better in many cases, the training process is slower in some context. Therefore a fast algorithm for training the SVM is introduced in [18] namely sequential minimal optimization (SMO). Applying quadratic programming (QP) optimization problem by breaking the problem into sub-problems SMO minimizes the time required to train the model. For our work, we have used Weka version of SMO having poly kernel.k- Nearest Neighbors (k-NN) [19] algorithm is a parameter independent classification method where the input consists of the k closest training samples. The training examples are considered as vectors in a multidimensional feature space having a class label. The appropriate value of k is dependent on data. For our experiments, we have changed the value of k and tried to find the highest accuracy of the algorithm and the corresponding k value.We have used 10-fold cross validation to train and test the proposed experimental method for each of the classifiers. The classifiers are implemented inside Weka [20] and we have utilized the tool to apply them directly on the collected data.IV.D ATA C OLLECTIONIn this section the data collection process is described. The participants used, sample questionnaires, dataset credentials and data cleaning, data pre-processing steps are illustrated here. A. ParticipantsFor this paper, we have collected data from 162 participants. The participants were very friendly and willing while giving honest answers of the questions. Among the participants 48 were females and 114 were males in the age boundary of 19 to 23. The participants all are Bangladeshi born students in their undergraduate level of study. All the participants’ data are kept anonymous and they duly signed consent form of data privacy and security concerns.B. QuestionnairesThe questionnaires used for the experiments are of two types. One is for personality traits and another is for perceived stress scale. We have used the 50-item IPIP question set for getting the numeric score of each personality traits of Big Five Model. The PSS is widely used survey form, which we have also used in our case without any major modifications. The participants were asked to fill up the hard copy printable forms found in the project of Open Source Psychometrics Project2. Each of the IPIP items were labeled between 1-5 score, where 1=disagree, 2=slightly disagree, 3=neutral, 4=slightly agree and 5=agree.The PSS is a 10-item question set that has to be labeled between 0-4 score, where 0=never, 1=almost never, 2=sometimes, 3=fairly often and 4=very often.C. Dataset CredentialsAs mentioned in the survey forms the scoring is given by the participants and the Overall score for each personality traits and total score of stress is determined. For using this data in machine learning approach, we have gathered these raw data into a formal dataset which contains the five personality traits values and the stress class. For each participant these data are collected in each row considering a feature set. The properties of the collected dataset are given in Table 1.Table 1. Properties of the collected DatasetD. Data Cleaning & Pre-processingThe dataset consists of the personality traits score. Openness-to-experience, Extraversion, Agreeableness, Neuroticism and Conscientiousness scores are determined using pre-set formulas and the no hard rule based data cleaning or pre-processing is performed over the data. Therefore, the noisy data is incorporated within the dataset, which may affect the performance, but the manually altered data will not give the real insight of the personality and stress measurements.V. E XPERIMENTAL M ETHODIn this section, the machine learning based model is implemented. The experiments are in form of supervised learning and here 10-fold cross validation is performed for each of the classification algorithm. The experimental model could be depicted in Fig. 1.Fig.1. Experimental method for mapping personality traits to stress scale.The typical machine learning based approach has been adopted to evaluate the performance of the mentioned classifier in section III.VI E XPERIMENTAL R ESULTSWe have implemented the experimental method using the arff file format in Waikato Environment for Knowledge Analysis (Weka) [20], developed by University of Waikato, which is a collection of machine learning algorithms including many classification algorithms. The performance metrics used for evaluation are precision, recall, f1-score and accuracy. The metrics could be calculated using the following equations (1), (2), (3) and (4), where TP is True Positive, FP is False Positive, TN is True Negative and FN is False Negative.Precision=TP / (TP + FP) (1)Recall = TP / (TP + FN) (2)F1 – score = (2 * Precision * Recall)/ (Precision + Recall)(3)Accuracy = (TP + TN)/ (TP + TN + FP + FN) (4)Table 2 shows the performance metrics for nine different classification algorithms. Except for NB, all other classifier has shown considerably satisfactory accuracy of above 60%. The classifiers are devised for the same algorithm, but they are specialized for different types of scenarios. Therefore, the accuracy metrics may have fluctuated results. For each of the classifiers, all four metrics are calculated and it is proved that SMO and k- NN gives the highest accuracy among the classifiers. SMO is anoptimized version of SVM and the advantages of quadratic programming in the back end may become influential in performance.Table 2. Performance Metrics of Different ClassifiersThe regular k-NN algorithm in Weka runs for k=1. For our work, we have calculated the performance metrics of k-NN algorithm for different k values in Table.The performance of the same algorithm differs as the neighborhood changes. The maximum achieved accuracy are same as SMO using k=13. Therefore, we may declare that for mapping personality traits with perceived stress scale SMO and k-NN are better classifiers among the traditional classifiers.VII. D ISCUSSIONFrom the above depicted results, we may claim some of the findings. The experimental findings are for the evaluation of the machine learning algorithm. Hence, the mapping between these two terms shows that:- The more extrovert persons are tends to getmoderately stressed.- The persons having fewer score in extraversionare less stressed.- The persons having higher score in openness-to-experience are tending to perceive high stress. - Except for some cases, it is evident that personshaving fewer score in neuroticism are tending to be less stressed.The claims are depicted from the data gathered. It is mathematically possible to build a regression based model to prove the above mentioned sayings. The students perceiving high level stress could have direct impact on their behavior as well as on academic results [25]. Using the machine learning approaches as presented in this paper could be utilized to predict the students’ achievements in terms of motivation, learning pattern and emotional intelligence, as in [25]. C. Bhanuprakash et al. [26] presented an informal approach to determine the bright graduate students from their classroom behavior patterns, which could be mapped also utilizing the perceived stress scales in classroom environments.DatasetData Cleaning & Pre-processingClassification ModelMapping personality Traits toStress ScaleVIII.C ONCLUSIONIn this paper, we have presented a machine learning based approach to map between the personality traits and perceived stress scale of individuals. The system is trained and tested by the survey data of 162 participants and got approximately 70% of accuracy. The participants for our work were undergraduate students. Therefore, one of the future scopes could be to determine the personalitytraits at the beginning of the semester and behave accordingly to the students who could get more stressed as the semester goes on. The impact factors of ICT tools in education [27] and using these tools may prepare a solid ground for researchers to apply machine learning in education data. Along with this idea, the recent ideas of neural networks have been utilized in [28] to predict the academic performances of computer science students. Applying the idea of mapping the behavior patterns and stress scale level with the academic performance of the computer science students may lead to the future scopes.A CKNOWLEDGMENTWe would like to acknowledge the participants from Stamford University Bangladesh and Daffodil International University for their valuable time and knowledge shared to our research works.R EFERENCES[1]S. Rothmann, E. P. Coetzer, "The big five personalitydimensions and job performance". SA Journal of Industrial Psychology. 29. doi:10.4102/sajip.v29i1.88. Retrieved 27 June 2013.[2]S. Cohen, T. Kamarck, R. Mermelstein, “A global measureof perceived stress”, Journal of Health and Social Behavior, 24, 386-396, 1983.[3]S. Cohen, G. Williamson, “Perceived Stress in a ProbabilitySample of the United States. Spacapan, S. and Oskamp, S.(Eds.) The Social Psychology of Health. Newbury Park, CA, Sage, 1988.[4] D. Quercia, M. Kosinski, D. Stillwell, J. Crowcroft, “Ourtwitter profiles, our selves: Predicting personality with twitter.”, In Privacy, Security, Risk and Trust (PASSAT) and 2011 IEEE Third International Conference on Social Computing (SocialCom), pp. 180–185, 2011.[5]P. Howlader, K. K. Pal, A. Cuzzocrea, S. D. M. Kumar,“Predicting facebook-users' personality based on status and linguistic features via flexible regression analysis techniques”, Proceeedings of the 33rd Annual ACM Symposium on Applied Computing, Pau, France, pp. 339- 345, April 09-13, 2018.[6]T. Tandera, D. Suhartono, R. Wong so, Y. L. Presidio,“Personality prediction system from Facebook users.”Procardia Computer Science, vol. 116, pp. 604–611, 2017.[7]M. Kosinski, S. Mats, S. Gosling, V. Popov, D. Stillwell,“Facebook as a research tool for the social sciences: Opportunities, challenges, ethical considerations, and practical guidelines.” American Psychologist,2015 Feb;70(6): pp. 543.[8]M. Kosinski, D. Stillwell, T. Grapple, “Private traits andattributes are predictable from digital records of human behavior”, In Proceedings of the National Academy of Sciences of the United States of America; 2013: PNAS. pp.5802-5805.[9]P. Holder, K. K. Pal, A. Cuzzocrea, S. D. M. Kumar,“Predicting Facebook-users' personality based on statusand linguistic features via flexible regression analysistechniques”,SAC '18 Proceedings of the 33rd AnnualACM Symposium on Applied Computing, pp. 339-345, Pau, France, April, 2018.[10]V. Kaushal, M. Patwardhan, “Emerging trends inpersonality identification using online social networks-A literature survey”, ACM Transactions on Knowledge Discovery from Data, vol. 12, no. 2, Article. 15, January 2018.[11]I. Rish, “An empirical study of the naive Bayes classifier”,In Proceedings of IJCAI-01 workshop on Empirical Methods in AI, pp. 41–46, Sicily, Italy, 2001.[12]S. R. Safavian, D. Landgrebe, “A survey of decision treeclassifier methodology”, IEEE Transactions on Systems, Man and Cybernetics, pp. 660–674, 1991.[13]L. Breiman, “Random forests”, Machine Learning, 45:5–32, 2001[14]Niels Landwehr, Mark Hall, Eibe Frank, “Logistic ModelTrees”, pp. 161-205, vol. 95, 2005.[15]R. A. Fisher, “The Use of Multiple Measurements inTaxonomic Problems”, Annals of Eugenics. vol. 7, issue.2, pp. 179–188, 1936.[16]L. Breiman. Bagging predictors. Machine Learning, vol.24 issue. 2, pp. 123–140, 1996.[17]R. Collobert, S. B engio, “Links between Perceptrons,MLPs and SVMs”. Proceedings of International Conference on Machine Learning (ICML), 2004.[18]J. C. Platt, “Sequential minimal optimization: A fastalgorithm for training support vector machines”,Technical Report MSR-TR_98_14, Microsoft Research, 1998.[19]N. S. Altman, “An introduction to kernel and nearest-neighbor nonparametric regression”, The American Statistician, vol. 46, issue. 3, pp. 175–185, 1992.[20]G. Holmes, A. Donkin, I.H. Witten, “Weka: A machinelearning workbench”,Proceedings of Second Australiaand New Zealand Conference on Intelligent InformationSystems, Brisbane, Australia.[21]M. M. Hasan, N. H. Shaon, A. A. Marouf, M. K. Hasan, H.Mahmud, and M. M. Khan, “Friend RecommendationFramework for Social Networking Sites using User‘sOnline Behavior”, IEEE- Computer and Information Technology (ICCIT), December 2015, pp. 539-543. [22] A. Denovan, N. Dagnall, K. Dhingra and S. Grogan,“Evaluating the Perceived Stress Scale among UKuniversity students: implications for stress measurementand management”, Journal of Studies in Higher Education, Vol. 44, Issue. 1, pp. 120-133, 2019.[23]K. Kumar, S. Kuma, D. Mehrotra, S. C. Tiwari, V. Kumarand R. C. Dwivedi, “Reliability and psychometric validity of Hindi version of Depression, Anxiety and Stress Scale-21 (DASS-21) for Hindi speaking Head Neck Cancer andOral Potentially Malignant Disorders Patients”, The Journal of Cancer Research and Therapeutics, January, 2019.[24] A. A. Marouf, M. K. Hasan, H. Mahmud, “IdentifyingNeuroticism from User Generated Content of SocialMedia based on Psycholinguistic Cues”, 2019 2nd IEEE Conference on Electrical, Computer and CommunicationEngineering (ECCE 2019), CUET, 7-9 February, 2019. [25]Muhammad U. Fahri, Sani M. Isa, " Data Mining toPrediction Student Achievement based on Motivation,Learning and Emotional Intelligence in MAN 1Ketapang", International Journal of Modern Educationand Computer Science(IJMECS), Vol.10, No.6, pp. 53-60, 2018.DOI: 10.5815/ijmecs.2018.06.07[26] C.Bhanuprakash, Y.S. Nijagunarya, M.A. Jayaram, " AnInformal Approach to Identify Bright Graduate Students by Evaluating their Classroom Behavioral Patterns by Using Kohonen Self Organizing Feature Map ", International Journal of Modern Education and Computer Science(IJMECS), Vol.10, No.8, pp. 22-32, 2018.DOI: 10.5815/ijmecs.2018.08.03[27] Bekim Fetaji, Majlinda Fetaji, Mirlinda Ebibi, Samet Kera," Analyses of Impacting Factors of ICT in Education Management: Case Study", International Journal of Modern Education and Computer Science(IJMECS), Vol.10, No.2, pp. 26-34, 2018.DOI: 10.5815/ijmecs.2018.02.03[28] Abimbola R. Iyanda, Olufemi D. Ninan, Anuoluwapo O.Ajayi, Ogochukwu G. Anyabolu, " Predicting Student Academic Performance in Computer Science Courses: A Comparison of Neural Network Models", International Journal of Modern Education and Computer Science(IJMECS), Vol.10, No.6, pp. 1-9, 2018.DOI: 10.5815/ijmecs.2018.06.01Authors’ ProfilesAhmed A. Marouf is currently pursuing M.Sc. Engg in Computer Science and Engineering (CSE) from Islamic University of Technology (IUT), Gazipur, Bangladesh. He received his Bachelor degree from the Department of Computer Science and Engineering (CSE), IUT in 2014 with major research in human computer interaction andpattern recognition.He is a graduate researcher of Systems and Software Lab (SSL) in the CSE department of IUT. His research interest lies within Human Computer Interaction, Biometric technologies. He is currently working as a lecturer in Department of Computer Science and Engineering (CSE) of Daffodil International University (DIU), Dhaka, Bangladesh. He also acts as the Technical Lead of DIU HCI (Human Computer Interaction) Research Lab.Adnan F. Ashrafi is currently pursuing M.Sc. Engg in Computer Science and Engineering (CSE) from Islamic University of Technology (IUT), Gazipur, Bangladesh. He received his Bachelor degree from the Department of Computer Science and Engineering (CSE), IUT in 2014 with major research in Bioinformatics and machinelearning.His research interest lies within bioinformatics, machine learning, DNA & RNA sequence analysis. He is currently working as a senior lecturer in Department of Computer Science and Engineering (CSE) of Stamford University Bangladesh (SUB), Dhaka, Bangladesh.Tanveer Ahmed received his Bachelor degree from the Department of Computer Science and Engineering (CSE), IUT in 2013. His research interest lies within Human Computer interaction, education data mining, computer networking and analysis intelligent systems.He is currently working as a seniorlecturer in Department of Computer Science and Engineering (CSE) of Stamford University Bangladesh (SUB), Dhaka, Bangladesh.Tarikuzzaman Emon received his B.Sc. degree in Computer Science and Engineering from East West University, Dhaka, Bangladesh in 2006, M.Sc. degree in Computer Networking from London Metropolitan University, London, U.K in 2011 and PG.Cert. in Computer System and Networking from University ofGreenwich, London, U.K in 2010.He is serving as an Assistant Professor in the department of Computer Science & Engineering at Stamford University Bangladesh, Dhaka, Bangladesh. His field of research interest includes Machine Learning, Computer Networking, Wireless Mobile Communication, Image Processing and Pattern Recognition.How to cite this paper: Ahmed A. Marouf, Adnan F. Ashrafi, Tanveer Ahmed, Tarikuzzaman Emon, " A Machine Learning based Approach for Mapping Personality Traits and Perceived Stress Scale of Undergraduate Students", International Journal of Modern Education and Computer Science(IJMECS), Vol.11, No.8, pp. 42-47, 2019.DOI: 10.5815/ijmecs.2019.08.05。