美国数学建模竞赛论文格式

全国大学生数学建模竞赛论文格式规范●本科组参赛队从A、B题中任选一题，专科组参赛队从C、D题中任选一题。

（全国评奖时，每个组别一、二等奖的总名额按每道题参赛队数的比例分配；但全国一等奖名额的一半将平均分配给本组别的每道题，另一半按每题论文数的比例分配。

）●论文用白色A4纸打印；上下左右各留出至少2.5厘米的页边距；从左侧装订。

●论文第一页为承诺书，具体内容和格式见本规范第二页。

●论文第二页为编号专用页，用于赛区和全国评阅前后对论文进行编号，具体内容和格式见本规范第三页。

●论文题目、摘要和关键词写在论文第三页上（无需译成英文），并从此页开始编写页码；页码必须位于每页页脚中部，用阿拉伯数字从“1”开始连续编号。

注意：摘要应该是一份简明扼要的详细摘要，请认真书写（但篇幅不能超过一页）。

●从第四页开始是论文正文（不要目录）。

论文不能有页眉或任何可能显示答题人身份和所在学校等的信息。

●论文应该思路清晰，表达简洁（正文尽量控制在20页以内，附录页数不限）。

●引用别人的成果或其他公开的资料(包括网上查到的资料) 必须按照规定的参考文献的表述方式在正文引用处和参考文献中均明确列出。

正文引用处用方括号标示参考文献的编号，如[1][3]等；引用书籍还必须指出页码。

参考文献按正文中的引用次序列出，其中书籍的表述方式为：[编号] 作者，书名，出版地：出版社，出版年。

参考文献中期刊杂志论文的表述方式为：[编号] 作者，论文名，杂志名，卷期号：起止页码，出版年。

参考文献中网上资源的表述方式为：[编号] 作者，资源标题，网址，访问时间（年月日）。

●在论文纸质版附录中，应给出参赛者实际使用的软件名称、命令和编写的全部计算机源程序（若有的话）。

同时，所有源程序文件必须放入论文电子版中备查。

论文及源程序电子版压缩在一个文件中，一般不要超过20MB，且应与纸质版同时提交。

（如果发现程序不能运行，或者运行结果与论文中报告的不一致，该论文可能会被认定为弄虚作假而被取消评奖资格。

你的论文需要从此开始请居中使用Arial14字体第一作者，第二作者和其他（使用Arial14字体）1.第一作者的详细地址，包括国籍和email(使用Arial11)2.第二作者的详细地址，包括国籍和email(使用Arial11)3.将所有的详细信息标记为相同格式关键词列出文章的关键词。

这些关键词会被出版方用作关键词索引(使用Arial11字体）论文正文使用Times New Roman12字体摘要这一部分阐述说明了如何为TransTechPublications.准备手稿。

最好阅读这些用法说明并且整篇论文都是遵照这个提纲。

手稿的正文部分应该是17cm*25cm（宽*高）的格式（或者是6.7*9.8英尺）。

请不要在这个区域以外书写。

请使用21*29厘米或8*11英尺的质量较好的白纸。

你的手稿可能会被出版商缩减20%。

在制图和绘表格时候请特别注意这些准则。

引言所有的语言都应该是英语。

请备份你的手稿（以防在邮寄过程中丢失）我们收到手稿即默认为原作者允许我们在期刊和书报出版。

如果作者在论文中使用了其他刊物中的图表，他们需要联系原作者，获取使用权。

将单词或词组倾斜以示强调。

除了每一部分的标题（标记部分的标题），不要加粗正文或大写首字母。

使用激光打印机，而不是点阵打印机正文的组织：小标题小标题应该加粗并注意字母的大小写。

第二等级的小标题被视为后面段落的一部分（就像这一大段的一小部分的开头）页码不要打印页码。

请用淡蓝色铅笔在每一张纸的左下角（在打印区域以外）标注数字。

脚注脚注应该单独放置并且和正文分开理想地情况下，脚注应该出现在参考文献页，并且放在文章的末尾，和正文用分割线分开。

表格表格（如表一，表二，...）应该放在正文当中，是正文的一部分，但是，要避免文本混乱。

一个描述性的表格标题要放在图表的下方。

标题应该独立的放在表格的下方或旁边。

表中的单位应放在中括号中[兆伏]如果中括号不可用，需使用大括号{兆}或小括号（兆)。

美赛格式要求范文美赛（MCM，Mathematical Contest in Modeling）是一项面向全球大学生的数学建模竞赛，每年举办一次。

在美赛中，参赛队伍需要从所给的问题中选择一个进行建模与分析，并以一篇报告的形式提交给评委会。

这篇报告的格式要求是非常重要的，正确的格式能够使报告更具条理性、易读性和专业性。

通常，美赛的报告要求在1200字以上。

以下是美赛报告常见的格式要求：2.摘要：摘要是美赛报告的起始部分，需要简洁明了地介绍问题的背景、目标、方法和结果。

摘要应当限制在200字以内，并能够很好地概括整篇报告的主要内容。

3.引言：引言部分需要对问题进行充分的描述和分析，包括问题的背景、现状和重要性。

同时，引言还应该明确问题的具体要求，并给出解决问题的思路和方法。

4.建模过程：建模过程是报告的核心部分，需要详细描述解决问题的思路和方法。

可以分为以下几个步骤：a.模型假设：在建模过程中，需要明确问题所涉及的假设条件，这些假设在以后的推导和分析中起到重要作用。

b.变量定义和符号说明：对于涉及到的变量和符号，需要给出明确的定义和说明。

这样可以使读者更好地理解报告中使用的符号和表达方式。

c.模型建立：根据问题描述和目标要求，建立相应的数学模型。

需要清晰地说明模型的基本假设、变量关系和优化目标等。

可以使用数学公式、图表和图像等方式进行表达。

d.模型求解：通过合适的数学方法和计算工具对模型进行求解。

需要给出详细的求解过程，包括数据处理、计算流程和结果分析等。

5.结果分析：在结果分析部分，需要对模型求解的结果进行合理的解释和分析。

对于复杂的结果，可以使用图表、图像和表格等形式进行展示。

同时还需要对结果的可行性、准确性、稳定性和实用性进行评价和讨论。

6.模型评价：对于所建立的模型，需要进行客观的评价和讨论。

包括模型的局限性、不确定性和可扩展性等方面。

同时还可以提出改进模型的建议和思考。

7.结论：在报告的结尾部分，需要对整个建模过程进行总结，并给出解决问题的最终结论。

美赛格式要求范文美赛（MCM/ICM）是美国大学生数学建模竞赛的英文缩写，是一项面向全球大学生的数学建模竞赛。

MCM/ICM每年提供若干个实际问题供参赛者选择，并规定参赛者提交一份由三人组成的队伍作品，要求队伍在规定的时间内解答问题并撰写一篇报告。

以下是美赛格式的一般要求：1.报告页数要求：参赛队伍通常需要撰写一篇1200字以上的报告。

具体的页数要求可以根据不同的问题和竞赛要求略有变化，但一般不超过20页。

尽管有页数限制，但在撰写报告时需要全面、清晰地阐述问题、解决方法和结论。

2.章节结构：一篇标准的美赛报告通常包括以下几个部分：-引言：介绍问题的背景和目的，明确解决问题的方法和目标。

-问题分析：对问题进行深入的分析和理解，包括重新表述问题、提出假设和限制条件，展开问题讨论。

-模型建立：建立一个或多个数学模型，以解决问题。

需要解释模型背后的理论基础和假设，并给出模型的描述和方程。

-模型求解：详细描述解决模型的方法、步骤和计算过程。

需要标注具体的计算公式、算法、图表和详细的计算结果。

-结果分析：对所得结果进行详细的解释和分析，包括结果的合理性和局限性，对模型的优缺点进行评价。

-结论与建议：总结所得结论，并提出可能的进一步研究方向和改进建议。

3.图表和数学符号的使用：美赛报告通常需要使用多个图表和数学符号，以支持和解释问题的分析和解决方法。

图表应该清晰、简洁，并配有必要的标注和说明。

数学符号应该统一、准确地使用，避免造成混淆。

5.语言表达：报告应使用准确、简练、清晰的语言表达问题、论证思路和解决方案。

语法、拼写和标点符号应正确无误。

尽管以上是一般的美赛报告要求，但具体的格式要求可能会因竞赛规则和题目的特殊性而有所不同。

建议参赛队伍在参赛前详细了解官方提供的竞赛规则和报告要求，并遵循官方给出的指导进行撰写报告。

同时，可以参考以往的优秀报告和获奖队伍的经验，借鉴其写作技巧和结构。

For office use onlyT1________________ T2________________ T3________________ T4________________Team Control Number21432Problem ChosenCFor office use onlyF1________________F2________________F3________________F4________________2012 Mathematical Contest in Modeling (MCM) Summary SheetTwo models to make conspirators nowhere to hide in social network With the development of high-technology, the number of white collar, high-tech crimes grow by more than 4% a year [1]. Bec ause of conspirators’ high IQ and professional knowledge, they are hard to be tracked down. Thus, we need use some special data mining and analytical methods to analyze social networks’ inherent law and finally offer help for litigating criminal suspect.M odel I is used for calculating everyone’s criminal possibility by the following 4 procedures: 1) Derive topics’ danger coefficient by Ana lytic Hierarchy Process (AHP);2) Set the discriminate line by Support Vector Machine (SVM); 3) Use the weight sum to c alculate everyone’s criminal possibility; 4) Provide a nomination form about conspiracy leaders by the Pagerank algorithm.Model II is an improved text analysis, used for more accurately analyzing the content and context of relevant information. The model includes four steps as follows: 1) Ascertain keywords and topics by counting their arisen times; 2) Syncopate me ssages’ sentence; 3) Match intelligently between messages and topics; 4) Get results by model I at last.We utilize models to evaluate requirement 1 and 2. The results show the fault of rates are 8.33% and 12.5%, which is acceptable.Table1. The results of requirement 1 and 2.conspirators criminal possibility leaders rankRequirement1Seeri 0.494 Julia 0.137 Sherri 0.366 Beth 0.099 Dolores 0.323 Jerome 0.095Requirement2 Sherri 0.326 Alex 0.098 Paige 0.306 Paige 0.094 Melia 0.284 Sherri 0.092To verify our two models and describe the ideas for requirement 3, we use models to analyze the 10 people’s example. The results of model II sho w our topics contain 78.8% initial information, better than the former 5 topics’ 57.7%. The results of model I can identify two shadowy conspirators, Bob and Inez. Thus, the models are more accurate and effective.According to the requirement4, we specifically discuss the effect of the thorough network analysis to our models. Meanwhile, we try to extend our models in distinguishing the safe page and unsafe page in Internet and the results derived from our models are reasonable.Two models to make conspirators nowhere to hideTeam #13373February 14th ,2012ContentIntroduction (3)The Description of the Problem (3)Analysis (3)What is the goal of the Modeling effort? (4)Flow chart (4)Assumptions (5)Terms, Definitions and Symbols (5)Model I (6)Overview (6)Model Built (6)Solution and Result (9)Analysis of the Result (10)Model II (11)Overview (11)Model Built (11)Result and Analysis (12)Conclusions (13)Technical summary (13)Strengths and Weaknesses (13)Extension (14)Reference (14)Appendix (16)IntroductionWith the development of our society, more and more high-tech conspiracy crimes and white-collar crimes take place in business and government professionals. Unlike simple violent crime, it is a kind of bran-new crime style, would gradually create big fraud schemes to hurt others’ benefit and destroy business companies.In order to track down the culprits and stop scams before they start, we must make full use of effective simulation model and methodology to search their criminal information. We create a Criminal Priority Model (CPM) to evaluate every suspect’s criminal possibility by analyzing text message and get a priority line which is helpful to ICM’s investigation.In addition, using semantic network analysis to search is one of the most effective ways nowadays; it will also be helpful we obtain and analysis semantic information by automatically extract networks using co-occurrence, grammatical analysis, and sentiment analysis. [1]During searching useful information and data, we develop a whole model about how to effective search and analysis data in network. In fact, not only did the coalescent of text analysis and disaggregated model make a contribution on tracking down culprits, but also provide an effective way for analyzing other subjects. For example, we can utilize our models to do the classification of pages.In fact, the conditions of pages’classification are similar to criminological analysis. First, according to the unsafe page we use the network crawler and Hyperlink to find the pages’ content and the connection between each pages. Second, extract the messages and the relationships between pages by Model II. Third, according to the available information, we can obtain the pages’priority list about security and the discriminate line separating safe pages and the unsafe pages by Model I. Finally we use the pages’ relationships to adjust the result.The Description of the ProblemAnalysisAfter reading the whole ICM problem, we make a depth analysis about the conspiracy and related information. In fact, the goal of ICM leads us to research how to take advantage of the thorough network, semantic, and text analyses of the message contents to work out personal criminal possibility.At first, we must develop a simulation model to analysis the current case’s data, and visualize the discriminate line of separating conspirator and non-conspirator.Then, by increasing text analyses to research the possible useful information from “Topic.xls”, we can optimize our model and develop an integral process of automatically extract and operate database.At last, use a new subject and database to verify our improved model.What is the goal of the Modeling effort?●Making a priority list for crime to present the most likely conspirators●Put forward some criteria to discriminate conspirator and non-conspirator, createa discriminate line.●Nominate the possible conspiracy leaders●Improve the model’s accuracy and the credit of ICM●Study the principle and steps of semantic network analysis●Describe how the semantic network analysis could empower our model.Flow chartFigure 1Assumptions●The messages have no serious error.●These messages and text can present what they truly mean.●Ignore special people, such as spy.●This information provided by ICM is reasonable and reliable.Terms, Definitions and SymbolsTable 2. Model parametersParameter MeaningThe rate of sending message to conspirators to total sending messageThe rate of receiving message to conspirators to total receiving messageThe dangerous possibility of one’s total messagesThe rate of messages with known non-conformist to total messagesDanger coefficient of topicsThe number of one’s sending messagesThe number of one’s receiving messagesThe number of one’s sending messages from criminalThe number of one’s receiving messages from criminalThe number of one’s sending messages from non-conspiratorThe number of one’s receiving messages from non-conspiratorDanger coefficient of peopleModel IOverviewModel I is used for calculating and analyzing everyone’s criminal possibility. In fact, the criminal possibility is the most important parameter to build a priority list and a discriminate line. The model I is made up of the following 4 procedures: (1) Derive topics’danger coefficient by Analytic Hierarchy Process (AHP); (2) Set the discriminate line by Support Vector Machine (SVM); (3) Use the weight sum to calculate everyone’s criminal possibility; (4) Provide a nomination form about conspiracy leaders by the Pagerank algorithm.Model BuiltStep.1PretreatmentIn order to decide the priority list and discriminate line, we must sufficiently study the data and factors in the ICM.For the first, we focus on the estimation about the phenomena of repeated names. In the name.xls, there are three pair phenomena of repeated names. Node#7 and node#37 both call Elsie, node#16 and node#34 both call Jerome, node#4 and node#32 both calls Gretchen. Thus, before develop simulation models; we must evaluate who are the real Elsie, Jerome and Gretchen.To decide which one accord with what information the problem submitsFirst we study the data in message.xls ，determine to analysis the number of messages of Elsie, Jerome and Gretchen. Table1 presents the correlation measure of their messages with criminal topic.Figure2By studying these data and figures, we can calculate the rate of messages about criminal topic to total messages; node#7 is 0.45455, while node#37 is 0.27273. Furthermore node#7 is higher than node#37 in the number of messages.Thus, we evaluate that node #7 is more likely Elsie what the ICM points out.In like manner, we think node#34, node#32 are those senior managers the ICM points out. In the following model and deduction, we assume node#7 is Elsie, node#34 is Jerome and node #32 is Gretchen.Step.2Derive topics’ danger coefficient by Analytic Hierarchy ProcessUse analytic hierarchy process to calculate the danger every topic’s coefficient. During the research, we decide use four factors’ effects to evaluate ：● Aim ：Evaluate the danger coefficient of every topic.[2]● Standard ：The correlation with dangerous keywordsThe importance of the topic itselfThe relationship of the topic and known conspiratorsThe relationship of the topic and known non-conspirators● Scheme ： The topics (1,2,3……15)Figure3According to previous research, we decide the weight of The Standard to Aim ：These weights can be evaluated by paired comparison algorithm, and build a matrix about each part.For example, build a matrix about Standard and Aim, the equation is followingij j i a C C ⇒:ijji ij n n ij a a a a A 1,0)(=>=⨯ The other matrix can be evaluated by the similar ways. At last, we make a consistency check to matrix A and find it is reasonable.The result shows in the table, and we can use the data to continue the next model. Step.3 Use the weight sum to calculate everyone ’s criminal possibilityWe will start to study every one’s danger coefficient by using four factors,, and .[3]100-第一份工作开始时间）（第一份工作结束时间第一份工作持续时间=The first factor means calculate the rate of someone’s sending criminal messages to total sending messages.The second factors means calculate the rate of someone’s receivingcriminal messages to total receiving messages.=The third factormeans calculate the dangerous possibility of someone’stotal messages.The four factorthe rate of someone’s messages with non-conspirators tototal messages.At last, we use an equation to calculate every one’s criticality, namely thepossibility of someone attending crime. ( Shows every factors’weighing parameter)After calculating these equations abov e, we derive everyone’s criminal possibilityand a priority list. (See appendix for complete table about who are the most likely conspirators) We instead use a cratering technique first described by Rossmo [1999]. The two-dimensional crime points xi are mapped to their radius from the anchor point ai, that is, we have f : xi → ri, where f(xi) = j i i a a (a shifted modulus). The set ri isthen used to generate a crater around the anchor point.There are two dominatingStep.4 Provide a nomination form about conspiracy leaders by the Pagerankalgorithm.At last, we will find out the possible important leaders by pagerank model, and combined with priority list to build a prior conspiracy leaders list.[4]The essential idea from Page Rank is that if node u has a link to node v, then the author of u is an implicitly conferring some importance to node v. Meanwhile it means node v has a important chance. Thus, using B (u) to show the aggregation of links to node u, and using F (u) to show the aggregation of received links of node u, The C is Normalization factor. In each iteration, propagate the ranks as follows: The next equation shows page rank of node u:Using the results of Page Rank and priority list, we can know those possiblecriminal conspiracy leaders.Solution and ResultRequirement 1:According to Model I above, we calculate these data offered by requirement 1 and build two lists. The following shows the result of requirement 1.By running model I step2, we derive danger coefficient of topics, the known conspiracy topic 7, 11 and 13 are high danger coefficient (see appendix Table4. for complete information).After running model step3, we get a list of every one’s criticality .By comparing these criticality, we can build a priority list about criminal suspects. In fact, we find out criminal suspects are comparatively centralized, who are highly different from those known non-conspirators. This illuminates our model is relative reasonable. Thus we decide use SVM to get the discriminate line, namely to separate criminal suspects and possible non-conspirators (see appendix Table5. for complete information). Finally, we utilize Page rank to calculate criminal suspects’ Rank and status, table4 shows the result. Thus, we nominate 5 most likely criminal leaders according the results of table4.They are Julia, Beth, Jerome, Stephanie and Neal.According to the requirement of problem1, we underscore the situations of three senior managers Jerome, Delores and Gretchen. Because the SVM model makes a depth analysis about conspirators, Jerome is chosen as important conspirator, Delores and Gretchen both has high danger coefficient. We think Jerome could be a conspirator, while Delores and Gretchen are regarded as important criminal suspects. Using the software Ucinet, we derive a social network of criminal suspects.The blue nodes represent non-conspirators. The red nodes represent conspirators. The yellow nodes represent conspiracy leaders.Figure 4Requirement 2:Using the similar model above, we can continue analyzing the results though theconditions change.We derive three new tables (4, 5 and 6): danger coefficient of topics, every one’s criticality and the probability of nominated. At last, we get a new priority list (table6) and 5 most likely criminal leaders: Alex, Sherri, Yao, Elsie and Jerome.We sincerely wish that our analysis can be helpful to ICM’s investigation. We figure out a new figure, which shows the social network of criminal suspects for requirement 2.Figure 5Analysis of the Result1)AnalysisIn the requirement 1, we find out 24 possible criminal suspects. All of 7 known conspirators are in the 24 suspects and their danger coefficients are also pretty high. However, there are 2 known non-conspirators are in these suspects.Thus, the rate of making mistakes is 8.33%. In all, we still have enough reasons to think the model is reasonable.In addition, we find 5 suspects who are likely conspirators by Support Vector Machine (SVM).In the requirement 2, we also choose 24 the most likely conspirators after run our CPM. All of 8 known conspirators are also in the 24 suspects and their danger coefficients are pretty high. Because 3 known non-conspirators are in these suspects, the rate of making mistakes is 12.5%, which is higher to the result of requirement 1.2)ComparisonTo research the effect of changing the number of criminal topics and conspirators to results, we decide to do an additional research about their effect.We separate the change of topics and crimes’numbers, analysis result’s changes of only one factor:In order to analyze the change between requirement 1 and 2, we choose those people whose rank has a big change over 30.Reference: the node.1st result: the part of the requirement1’s priority list.2nd result: the part of the requirement2’s priority list.3rd result: the priority’s changes of requirement 1 and 2.After investigate these people, we find out the topics about them isn’t close connected with node#0. Thus, the change of node#0 does not make a great effect on their change.However, there are more than a half of people who talk about topic1. According to the analysis, we find the topic1 has a great effect on their change. The topic1 is more important to node#0.Thus; we can do an assumption that the decision of topics has bigger effect on the decision of the personal identity and decide to do a research in the following content.Model IIOverviewAccording to requirement3, we will take the text analysis into account to enhance our model. In the paper, text analysis is presented as a paradigm for syntactic-semantic analysis of natural language. The main characteristics of this approach are: the vectors of messages about keywords, semanteme and question formation. In like manner, we need get three vectors of topics. Then, we utilize similarity to separate every message to corresponding topics. Finally, we evaluate these effects of text analysis by model I.Model BuiltStep.1PretreatmentIn this step, we need conclude relatively accurate topics by keywords in messages. Not only builds a database about topics, but also builds a small database for adjusting the topic classification of messages. The small database for adjusting is used for studying possible interpersonal relation between criminal suspects, i. e. Bob always use positive and supportive words to comment topics and things about Jerry, and then we think Bob’s messages are closely connected with topics about Jerry. [5] At first, we need to count up how many keywords in the whole messages.Text analysis is word-oriented, i.e., the word plays the central role in language understanding. So we avoid stipulating a grammar in the traditional sense and focus on the concept of word. During the analysis of all words in messages, we ignore punctuation, some simple word such as “is” and “the”, and extract relative importantwords.Then, a statistics will be completed about how many times every important word occurs. We will make a priority list and choose the top part of these words.Finally, according to some messages, we will turn these keywords into relatively complete topics.Step.2Syncopate sentenceWe will make a depth research to every sentence in messages by running program In the beginning, we can utilize the same way in step1 to syncopate sentence, deriving every message’s keywords. We decide create a vector about keywords: = () (m is the number keywords in everymessage)For improving the accuracy and relativity of our keywords, we decide to build a vector that shows every keyword’s synonyms, antonym.= () (1<k<m, p is the number of correlative words)According to primary analysis, we can find some important interpersonal relations between criminal suspects, i.e. Bob is closely connected with Jerry, then we can build a vector about interpersonal relation.= () (n is the number of relationships in one sentence )Step.3Intelligent matchingIn order to improve the accuracy of our disaggregated model, we use three vectors to do intelligent matching.Every message has three vectors:. Similarly, every topic alsohas three vectors.At last, we can do an intelligent matching to classify. [6]Step.4Using CPMAfter deriving new the classification of messages, we will make full use of new topics to calculate every one’s criticality.Result and AnalysisAfter calculating the 10 people example, we derive new topics. By verifying the topics’ contained initial information, we can evaluate the effect of models.The results of model II show our topics contain 78.8% initial information, better than former 5 topics’ 57.7%.T hus, new topics contain more initial information. Meanwhile, we build a database about interpersonal relation, and using it to optimize the results of everyone’s criminal possibility.Table 3#node primary new #node primary new1 0 0.065 6 0.342 0.2652 0.342 0.693 7 0.891 0.9123 0.713 0.562 8 0.423 0.354 1 1 9 0.334 0.7235 0.823 0.853 10 0.125 0.15 The results of model I can identify the two shadowy conspirators, Bob and Inez. In the table, the rate of fault is becoming smaller.According to Table11, we can derive some information:1.Analysis the danger coefficient of two people, Bob and Inez. Bob is theperson who self-admitted his involvement in a plan bargain for a reducedsentence. His data changes from 0.342 to 0.693. And Inez is the person whogot off, his data changes from 0.334 to 0.723. The models can identify thetwo shadowy people.2.Carol, the person who was later dropped his data changes from 0.713 to0.562. Although it still has a relatively high danger coefficient, the resultsare enhancing by our models.3.The distance between high degree people and low degree become bigger, itpresents the models would more distinctly identify conspirators andnon-conspirators.Thus, the models are more accurate and effective.ConclusionsTechnical summaryWe bring out a whole model about how to extract and analysis plentiful network information, and finally solve the classification problems. Four steps are used to make the classification problem easier.1)According known conspirators and correlative information, use resemblingnetwork crawler to extract what we may need information and messages.[7]2)Using the second model to analysis and classify these messages and text, getimportant topics.3)Using the first model to calculate everyone’s criminal possibility.4)Using an interpersonal relation database derived by step2 to optimize theresults. [8]Strengths and WeaknessesStrengths:1）We analyze the danger coefficient of topics and people by using different characteristics to describe them. Its results have a margin of error of 10percentage points. That the Models work well.2）In the semantic analysis, in addition to obtain topics from messages in social network, we also extract the relationships of people and adjust the final resultimprove the model.3）We use 4 characteristics to describe people’s danger coefficient. SVM has a great advantage in classification by small characteristics. Using SVM to classify the unknown people and its result is good.Weakness：1）For the special people, such as spy and criminal researcher, the model works not so well.2）We can determine some criminals by topics; at the same time we can also use the new criminals to adjust the topics. The two react upon each other. We canexpect to cycle through several times until the topics and criminals are stable.However we only finish the first cycle.3）For the semantic analysis model we have established, we just test and verify in the example (social network of 10 people). In the condition of large social network, the computational complexity will become greater, so the classify result is still further to be surveyed.ExtensionAccording to our analysis, not only can our model be applied to analyze criminal gangs, but also applied to similar network models, such as cells in a biological network, safe pages in Internet and so on. For the pages’ classification in Internet, our model would make a contribution. In the following, we will talk about how to utilize [9] Our model in pages’ classification.First, according to the unsafe page we use the network crawler and Hyperlink to find the pages’content and the connection between each page. Second, extract the messages and the relationships between pages by Model II. Third, according to the available information, we can obtain the pages’priority list about security and the discriminate line separating safe pages and the unsafe pages by Model I. Finally we use the pages’ relationships to adjust the result.Reference1. http://books.google.pl/books?id=CURaAAAAYAAJ&hl=zh-CN2012.2. AHP./wiki/%E5%B1%82%E6%AC%A1%E5%88%86%E6%9E%90%E6%B 3%95.3. Schaller, J. and J.M.S. Valente, Minimizing the weighted sum of squared tardiness on a singlemachine. Computers & Operations Research, 2012. 39(5): p. 919-928.4. Frahm, K.M., B. Georgeot, and D.L. Shepelyansky, Universal emergence of PageRank.Journal of Physics a-Mathematical and Theoretical, 2011. 44(46).5. Park, S.-B., J.-G. Jung, and D. Lee, Semantic Social Network Analysis for HierarchicalStructured Multimedia Browsing. Information-an International Interdisciplinary Journal, 2011.14(11): p. 3843-3856.6. Yi, J., S. Tang, and H. Li, Data Recovery Based on Intelligent Pattern Matching.ChinaCommunications, 2010. 7(6): p. 107-111.7. Nath, R. and S. Bal, A Novel Mobile Crawler System Based on Filtering off Non-ModifiedPages for Reducing Load on the Network.International Arab Journal of Information Technology, 2011. 8(3): p. 272-279.8. Xiong, F., Y. Liu, and Y. Li, Research on Focused Crawler Based upon Network Topology.Journal of Internet Technology, 2008. 9(5): p. 377-380.9. Huang, D., et al., MyBioNet: interactively visualize, edit and merge biological networks on theWeb. Bioinformatics, 2011. 27(23): p. 3321-3322.AppendixTable 4requirement 1topic danger topic danger topic danger topic danger7 1.65 4 0.78 5 0.47 8 0.1713 1.61 10 0.77 15 0.46 14 0.1711 1.60 12 0.47 9 0.19 6 0.141 0.812 0.473 0.18requirement 2topic danger topic danger topic danger topic danger1 0.402 0.26 15 0.15 14 0.117 0.37 9 0.23 8 0.15 3 0.0913 0.37 10 0.21 5 0.14 6 0.0611 0.30 12 0.18 4 0.12Table 5requirement 1#node danger #node danger #node danger #node danger 21 0.74 22 0.19 0 0.13 23 0.03 67 0.69 4 0.19 40 0.13 72 0.03 54 0.61 33 0.19 36 0.13 62 0.03 81 0.49 47 0.19 11 0.12 51 0.02 7 0.47 41 0.19 69 0.12 57 0.02 3 0.37 28 0.18 29 0.12 64 0.02 49 0.36 16 0.18 12 0.11 71 0.02 43 0.36 31 0.17 25 0.11 74 0.01 10 0.32 37 0.17 82 0.11 58 0.01 18 0.29 27 0.16 60 0.10 59 0.01 34 0.29 45 0.16 42 0.10 70 0.00 48 0.28 50 0.16 65 0.09 53 0.00 20 0.27 24 0.16 9 0.09 76 0.00 15 0.27 44 0.16 5 0.09 61 0.00 17 0.26 38 0.16 66 0.09 75 -0.01 2 0.23 13 0.16 26 0.08 77 -0.01 32 0.23 35 0.15 39 0.06 55 -0.02 30 0.20 1 0.15 80 0.04 68 -0.02 73 0.20 46 0.15 78 0.04 52 -0.0319 0.20 8 0.14 56 0.03 63 -0.03 14 0.19 6 0.14 79 0.03requirement 2#node danger #node danger #node danger #node danger 0 0.39881137 75 0.1757106 47 0.1090439 11 0.0692506 21 0.447777778 52 0.1749354 71 0.1089147 4 0.0682171 67 0.399047158 38 0.1738223 82 0.1088594 42 0.0483204 54 0.353754153 10 0.1656977 14 0.1079734 65 0.046124 81 0.325736434 19 0.1559173 27 0.1060724 60 0.0459948 2 0.306054289 40 0.1547065 23 0.105814 39 0.0286822 18 0.303178295 30 0.1517626 5 0.1039406 62 0.0245478 66 0.28372093 80 0.145155 8 0.10228 78 0.0162791 7 0.279870801 24 0.1447674 73 0.1 56 0.0160207 63 0.261886305 70 0.1425711 50 0.0981395 64 0.0118863 68 0.248514212 29 0.1425562 26 0.097213 72 0.011369548 0.239668277 45 0.1374667 1 0.0952381 79 0.009302349 0.238076781 37 0.1367959 69 0.0917313 51 0.0056848 34 0.232614868 17 0.1303064 33 0.0906977 57 0.0056848 3 0.225507567 6 0.1236221 31 0.0905131 74 0.0054264 35 0.222435188 22 0.1226934 36 0.0875452 76 0.005168 77 0.214470284 13 0.1222868 41 0.0822997 53 0.0028424 20 0.213718162 44 0.115007 46 0.0749354 58 0.0015504 43 0.204328165 12 0.1121447 28 0.0748708 59 0.0015504 32 0.193311469 15 0.1121447 16 0.074234 61 0.0007752 55 0.182687339 9 0.1117571 25 0.0701292Table 6requirement 1#node leader #node leader #node leader #node leader 15 0.1368 49 0.0481 7 0.0373 19 0.0089 14 0.0988 4 0.0423 21 0.0357 32 0.0073 34 0.0951 10 0.0422 18 0.029 22 0.0059 30 0.0828 67 0.0421 48 0.0236 81 0.0053 17 0.0824 54 0.0377 20 0.0232 73 043 0.0596 3 0.0377 2 0.0181 33 0requirement 2#node leader #node leader #node leader #node leader 21 0.0981309 7 0.0714406 54 0.0526831 43 0.01401872 0.0942899 34 0.0707246 32 0.0464614 81 0.00977763 0.0916127 0 0.0706746 18 0.041114248 0.0855984 20 0.0658119 68 0.028532867 0.0782211 49 0.0561665 35 0.024741。

ContentsⅠIntroduction (1)1.1Problem Background (1)1.2Previous Research (2)1.3Our Work (2)ⅡGeneral Assumptions (3)ⅢNotations and Symbol Description (3)3.1 Notations (4)3.2 Symbol Description (4)ⅣSpread of Ebola (5)4.1 Traditional Epidemic Model (5)4.1.1.The SEIR Model (5)4.1.2 (6)4.1.3 (6)4.2 Improved Model (7)4.2.1.The SEIHCR Model (8)4.2.2 (9)ⅤPharmaceutical Intervention (9)5.1 Total Quantity of the Medicine (10)5.1.1.Results from WHO Statistics (10)5.1.2.Results from SEIHCR Model (11)5.2 Delivery System (12)5.2.1.Locations of Delivery (13)5.2.2 (14)5.3 Speed of Manufacturing (15)ⅥOther Important Interventions (16)6.1 Safer Treatment of Corpses (17)6.2 Conclusion (18)ⅦControl and Eradication of Ebola (19)7.1 How Ebola Can Be Controlled (20)7.2 When Ebola Will Be Eradicated (21)ⅧSensitivity Analysis (22)8.1 Impact of Transmission Rate (23)8.2 Impact of the Incubation Priod (24)ⅨStrengths and Weaknesses (25)9.1 Strengths (26)9.2 Weaknesses (27)9.3 Future Work (28)Letter to the World Medical Association (30)References (31)ⅠIntroduction1.1.Promblem Background1.2.Previous Research1.3.Our WorkⅡGeneral Assumptions●●ⅢNotations and Symbol Description3.1. Notataions3.2. Symbol DescriptionSymbol DescriptionⅣSpread of Ebola4.1. Traditional Epidemic Model4.1.1. The SEIR Model4.1.2. Outbreak Data4.1.3. Reslts of the SEIR Model4.2. Improved Model4.2.1. The SEIHCR Model4.2.2. Choosing paametersⅤPharmaceutical Intervention 5.1. Total Quantity of the Medicine 5.1.1. Results from WHO Statistics5.2. Delivery System5.2.1. Locations of Delivery5.2.2. Amount of Delivery5.3. Speed of Manufacturong5.4. Medicine EfficacyⅥOther Important Interventions 6.1. Safer Treatment of Corpses6.2. ConclusionⅦControl and Eradication of Ebola 7.1. How Ebola Can Be Controlled7.2. When Ebola Will Be EradicatedⅧSensitivity Analysis8.1. Impact of Transmission Rate8.2. Impact of Incubation PeriodⅨStrengths and Weaknesses 9.1. Strengths●●●9.2. Weaknesses●●●9.3.Future WorkLetter to the World Medical AssociationTo whom it may concern,Best regards,Team #32150References ［1］［2］［3］［4］。

摘要：第一段：写论文解决什么问题1．问题的重述a. 介绍重点词开头：例1：“Hand move” irrigation, a cheap but labor-intensive system used on small farms, consists of a movable pipe with sprinkler on top that can be attached to a stationary main.例2:……is a real-life common phenomenon with many complexities.例3：An (effective plan) is crucial to………b. 直接指出问题：例 1：We find the optimal number of tollbooths in a highway toll-plaza for a given number of highway lanes: the number of tollbooths that minimizes average delay experienced by cars.例2：A brand-new university needs to balance the cost of information technology security measures with the potential cost of attacks on its systems.例3：We determine the number of sprinklers to use by analyzing the energy and motion of water in the pipe and examining the engineering parameters of sprinklers available in the market.例4: After mathematically analyzing the …… problem, our modeling group would like to present our conclusions, strategies, (and recommendations )to the …….例5：Our goa l is... that (minimizes the time )……….2．解决这个问题的伟大意义反面说明。

美赛附录格式
（中英文实用版）
Appendix Format for MCM/ICM
美赛附录格式的要求如下：
美赛附录部分的格式要求如下：
1.所有附录内容应集中在一个单独的附录文件中，不要在论文正文中插入附录。

2.附录文件应以“Appendix”作为文件名，并按照提交论文的要求进行排版。

3.附录中的每个部分应以“Appendix A”、“Appendix B”等格式进行编号。

4.附录中的表格、图形、方程等应与正文中的引用保持一致，并在附录中给出详细说明。

5.附录中的文本应保持清晰、简洁，并避免过多的重复内容。

6.附录中的数据、图表等应确保准确无误，并与正文中的分析结果相符合。

7.附录中的引用文献应按照美赛规定的引用格式进行标注。

8.附录中的公式、符号、单位等应按照美赛规定的格式进行排版。

The format requirements for the appendix section of the MCM/ICM competition are as follows:
1.All appendix content should be集中在一个单独的附录文件中，不要在论文正文中插入附录。

Journal Citation(to be inserted by the publisher)Copyright by Trans Tech PublicationsYour Paper's Title Starts Here:Please Centeruse Arial14First Author1,Second Author2and Others3(use Arial14)1Full address of first author,including country,email(use Arial11)2Full address of second author,including country,email3List all distinct addresses in the same wayKeywords::List the keywords covered in your paper.These keywords will also be used by the Keywordspublisher to produce a keyword index.(use Arial11)For the rest of the paper,please use Times New Roman12Abstract.This document explains and demonstrates how to prepare your camera-ready manuscript for Trans Tech Publications.The best is to read these instructions and follow the outline of this text. The text area for your manuscript must be17cm wide and25cm high(6.7and9.8inches,resp.). Do not place any text outside this e good quality,white paper of approximately21x29cm or8x11inches.Your manuscript will be reduced by approximately20%by the publisher.Please keep this in mind when designing your figures and tables etc.IntroductionAll manuscripts must be in English.Please keep a second copy of your manuscript in your office (just in case anything gets lost in the mail).When receiving the manuscript,we assume that the corresponding authors grant us the copyright to use the manuscript for the book or journal in question.Should authors use tables or figures from other Publications,they must ask the corresponding publishers to grant them the right to publish this material in their paper.Use italic for emphasizing a word or phrase.Do not use boldface typing or capital letters except for section headings(cf.remarks on section headings,below).Use a laser printer,not a matrix dot printer.Organization of the TextSection Headings.The section headings are in boldface capital and lowercase letters.Second level headings are typed as part of the succeeding paragraph(like the subsection heading of this paragraph).Page Numbers.Do not print page numbers:Please number each sheet slightly in the left corner near the bottom(outside the typing area)with a light blue pencil.Footnotes.Footnotes1should be single spaced and separated from the text.Ideally,footnotes appear on the page of their reference,and are placed at the foot of the text,separated from the text by a horizontal line.Tables.Tables(refer with:Table1,Table2,...)should be presented as part of the text,but in such a way as to avoid confusion with the text.A descriptive title should be placed above each table. The caption should be self-contained and placed below or beside the table.Units in tables should be given in square brackets[meV].If square brackets are not available,use curly{meV}or standard brackets(meV).1This is a footnote2Title of Publication(to be inserted by the publisher)Figures.Figures(refer with:Fig.1,Fig.2,...)also should be presented as part of the text, leaving enough space so that the capt-ion will not be confused with the text.The caption should be self-contained and placed below or beside the figure.Generally,only original drawings or photographic reproductions are acceptable.Only very good photocopies are acceptable.Utmost care must be taken to insert the figures in correct alignment with the text.Half-tone pictures should be in the form of glossy prints.If possible,please include your figures as graphic images in the electronic version.If TTP is required to scan and insert images,please keep the following points in mind:(a)the allotted space(for inserting illustrations)must exactly match the space made available inthe camera-ready version,so that the electronic version is identical to the hard copy with regard to page and line breaks.(b)the required positioning of any high-quality separate illustration must be clearly indicated onits reverse side.The size of the illustrations must exactly match the space left in the camera-ready manuscript.Equations.Equations(refer with:Eq.1,Eq.2,...)should be indented5mm(0.2").There should be one line of space above the equation and one line of space below it before the text continues.The equations have to be numbered sequentially,and the number put in parentheses at the right-hand edge of the text.Equations should be punctuated as if they were an ordinary part of the text.Punctuation appears after the equation but before the equation number,e.g.c2=a2+b2.(1)Literature ReferencesReferences are cited in the text just by square brackets[1].(If square brackets are not available, slashes may be used instead,e.g./2/.)Two or more references at a time may be put in one set of brackets[3,4].The references are to be numbered in the order in which they are cited in the text and are to be listed at the end of the contribution under a heading References,see our example below.SummaryOn your floppy disk,please indicate the format and word processor used.Please also provide your phone number,fax number and e_mail address for rapid communication with the publisher(will not be published).Please always send your disk along with a hard copy that must match the disk's content exactly.If you follow the foregoing,your paper will conform to the requirements of the publisher and facilitate a problem-free publication process.References[1]Dj.M.Maric,P.F.Meier and S.K.Estreicher:Mater.Sci.Forum Vol.83-87(1992),p.119[2]M.A.Green:High Efficiency Silicon Solar Cells(Trans Tech Publications,Switzerland1987).This document is available on the web at /download **Please submit your paper in hardcopy and also electronically to the conference editor.。