当前位置:文档之家› Deciding to remind during collaborative problem solving Empirical evidence for agent strate

Deciding to remind during collaborative problem solving Empirical evidence for agent strate

Deciding to remind during collaborative problem solving Empirical evidence for agent strate
Deciding to remind during collaborative problem solving Empirical evidence for agent strate

Deciding to Remind During Collaborative Problem Solving: Empirical Evidence for Agent Strategies

Pamela W.Jordan Intelligent Systems Program University of Pittsburgh

Pittsburgh PA15260 jordan@https://www.doczj.com/doc/455956288.html,

Marilyn A.Walker

ATT Research Laboratories

600Mountain Ave

Murray Hill NJ07974 walker@https://www.doczj.com/doc/455956288.html,

Abstract

Previous work suggests that reminding a conversational part-

ner of mutually known information depends on the conver-

sants’attentional state,their resource limits and the resource

demands of the task.In this paper,we propose and evaluate

several models of how an agent decides whether or not to

communicate a reminder.We elaborate on previous?ndings

by exploring how attentional state and resource bounds are in-

corporated into the decision making process so that reminders

aid the performance of agents during collaborative problem

solving.We test two main hypotheses using a multi-agent

problem solving simulation testbed:(1)an agent decides to

present salient knowledge only when it reduces overall prob-

lem solving effort(2)an agent can use its own attentional state

as a model of the attentional state of its partner when assessing

the effort trade-offs of communicating a reminder.Our results

support both hypotheses,suggesting that the models we pro-

pose should be further tested for multi-agent communication

in problem solving situations.

Introduction

Recent work in multi-agent communication has begun

to address the problem of information overload and se-

lective attention:the problem of how an agent decides

which subset of a potentially large set of facts should

be attended to(Joshi1978;Walker&Rambow1994;

Giunchiglia et al.1993)inter alia.In human-human com-

munication,both agents recognize that agents have selective

attention and will sometimes REMIND the other agent of mu-

tually known facts that should be selectively attended to.We

posit that models of reminding for human-human conversa-

tion may be applicable to systems for both human-agent and

agent-agent communication.Thus,given a situation of two

communicating agents,agent A and agent B,we draw from

analyses of human-human communication in order to form

hypotheses of how agent A decides to remind agent B of a

particular relevant fact.We then test our hypotheses using a

dialogue simulation testbed.

In previous work on reminding,Walker and Rambow

(henceforth W&R)discuss the following excerpt of a nat-

1Salient is a cognitive term,but facts in a cache can be viewed

as salient facts.The critical assumption is that the salient set is a

subset of what is known that is being selectively attended to.

The plan for the paper is as follows.First we present three hypotheses about the basis of an agent’s algorithm for deciding whether to remind.Next we describe the Design-World dialogue simulation testbed that we use for testing our hypotheses.Finally,we present our results,which support our two main hypotheses.

Deciding to Remind in Collaborative Problem

Solving

In this work,we adopt W&R’s de?nition of performance evaluation in collaborative problem solving.This perfor-mance measure assumes that the agents are working together as a team(Levesque,Cohen,&Nunes1990;Grosz&Sid-ner1990),and as a team are attempting to maximize perfor-mance.PERFORMANCE is the difference between an objective measure of the utility of a completed task and a cost mea-sure called COLLABORA TIVE EFFORT(Clark&Schaefer1989; Brennan1990;Zukerman&McConachy1993).

PERFORMANCE Task De?ned RAW SCORE

COLLABORATIVE EFFORT. COLLABORATIVE EFFORT consists of all the processing re-quired for both agents to carry out the problem solving task. In our processing architecture,this is composed of COM-MUNICA TIVE EFFORT,RETRIEV AL EFFORT and INFERENCE EF-FORT.To ensure that our calculations of collaborative effort are independent of the implementation,the calculation of COMMUNICATIVE EFFORT is parameterized by COMMCOST, which speci?es the cost of producing and understanding one message,the calculation of RETRIEVAL EFFORT is parameter-ized by RETCOST,which speci?es the cost of one retrieval from memory,and the calculation of INFERENCE EFFORT is parameterized by INFCOST,which speci?es the cost of mak-ing one inference,as de?ned below:

COLLABORATIVE EFFORT

(COMMCOST total messages for both agents)

+(RETCOST total retrievals for both agents)

+(INFCOST total inferences for both agents)

Our experiments focus on decisions that have to do with reminding another agent of mutually known information. Since at least one agent is already attending to this infor-mation,reminding cannot increase RAW SCORE;it can only reduce COLLABORATIVE EFFORT.Thus,our?rst hypothesis is what we call the PERFORMANCE hypothesis: PERFORMANCE hypothesis:Agent A decides to present salient knowledge to agent B only when it improves performance by reducing collaborative effort.

In particular,agent A’s decision model must determine whether or not performance will be improved if B is re-minded of the warrant for a proposal.When we attempt to de?ne a decision algorithm to test the PERFORMANCE hy-pothesis,the?rst question that arises is how agent A goes about calculating collaborative effort.Agent A must have access to two types of information.

First,the agent must be able to access an estimate of the costs of the various processes that contribute to collaborative effort.To provide an agent with cost estimates for various processes,we de?ne COMMCOST and RETCOST as parameters of the environment that an agent has access to.Since our decision algorithms for reminding do not involve inference effort,it is ignored by the agents,and INFCOST is set to0. Second,agent A must have a model of agent B’s attentional state to determine whether a warrant is retrievable,and to estimate total RETRIEV AL EFFORT.Where does A get such a model?One possibility proposed by W&R(Walker& Rambow1994)is that agent A maintains a detailed model of agent B’s attentional state.This approach is consistent with work on modeling the other agent’s cognitive or attentional state(Zukerman&McConachy1993;Grosz&Sidner1986). This possibility is the ESP hypothesis:

ESP hypothesis:Agent A maintains a detailed model of agent B’s attentional state.

We propose that another possible source of an attentional model is for agent A to use its own attentional state to ap-proximate which of B’s beliefs are salient,and to estimate total RETRIEV AL EFFORT.We call this possibility the SOLIP-SISTIC hypothesis:

SOLIPSISTIC hypothesis:Agent A uses its own atten-tional state as a model of B’s attentional state.

The SOLIPSISTIC hypothesis is plausible because agent A is not always in a position to evaluate what agent B is currently attending to,and because agents in conversation appear to expend a great deal of effort to stay coordinated(Brennan 1990;Thomason1990).In addition,in our analyses of human-human dialogues we found evidence that humans use an approximate model of one another since they sometimes make the wrong decision.2In the problem solving dialogue in(2),in which two human agents must negotiate the?oor plan for a two room house,speaker J chooses not to fully motivate the proposal in(J-2).

(2)D-1:The green rug looks good,let’s go ahead and get it.

I only have$50left–how about two yellow chairs for

$25each?

J-1:Are the yellow chairs for the living room?

D-2:Yes.

J-2:Okay,that’s?ne then.How about a?oor lamp for50 dollars for the living room?

D-3:Ok,we could use some light–what color is it?

J-3:Yellow,of course!I wouldn’t dream of putting a red or green lamp with yellow chairs.

In dialogue(2),J and D mutually know they have a goal to match the colors of items in a room,and J assumes this goal is salient.However the color matching goal must not be salient for D since D asks about the color of the proposed item in(D-3).A cognitively plausible cause of this error is that J based her decision not to communicate the warrant on her own attentional state.

The ESP and SOLIPSISTIC hypotheses give rise to two differ-ent bases for adaptive communicative strategies,depending on where agent A gets its model of agent B’s attentional state.Given these adaptive communicative strategies,we can then test the PERFORMANCE hypothesis by determining whether either of these strategies improves performance when com-pared to the static All-Implicit and Explicit-Warrant strate-gies.In addition,we can test whether it is worthwhile main-taining a detailed model by comparing the performance of the two adaptive strategies.

The Design-World Testbed

ROOM # 1

ROOM # 2

DESIGN WORLD COLLABORATIVE PLAN: 434 points

KEY:G

F

F

P

G

P

G

G

F = FUCHSIA

G = GREEN P= PURPLE

55

53

55

56

55

5452

54

Figure 1:One Final State for Design-World Standard Task:Represents the CollaborativePlan Achieved by the Dialogue,434points

We test the hypotheses described above using the Design-World dialogue simulation testbed.Design-World is a testbed for theories about collaborative problem solving dialogues,that supports experiments on the interaction of agents’resource limits,their communicative strategies and the complexity of the problem solving task.3This section is drawn directly from the description of the Design-World testbed in (Walker &Rambow 1994).

The Design-World task requires two agents to carry out a dialogue in order to negotiate an agreement on the design of the ?oor plan of a two room house (Whittaker,Geelhoed,&Robinson 1993).Figure 1shows a potential ?nal design plan negotiated via a (simulated)dialogue,such as that in (3).Both the arti?cial language that the agents communicate with and an italicized gloss generated from that language are shown:4

and is done automatically by adhoc methods.5

The agents don’t know that all items have been assigned and don’t engage in existential reasoning,i.e.they do not make infer-ences about what better choices other agents might propose on the basis of these utilities.

Modelling Collaborative Interactions

In each dialogue simulation,the agents’goal is to agree on 4pieces of furniture for each room so that the maximum utility is achieved.Negotiating an agreement between two agents consists of a cycle of four steps.First,individual agents perform means-end reasoning about OPTIONS to use various furniture pieces that they have.Second,individual agents deliberate about which options are preferable based on the utility of using the item in the plan.Since the AWM parameter limits what an agent is attending to,an agent may not be able to recall and identify its highest utility pieces at any given time.

Once a preferred option is identi?ed,the agent attempts to get the other agent to agree to a proposal involving that option.In general,agents’communicative intentions are realized through DISCOURSE ACTS such as PROPOSALS,AC-CEPTANCES,REJECTIONS and CLARIFICATIONS.In the third step,agents use the preferred options to make PROPOSALS to other agents to PUT a piece of furniture into a room.Then in the fourth step,the other agents ACCEPT,REJECT,or request CLARIFICA TION of these proposals.

This is illustrated by the dialogue excerpt in(3).After receiving utterance(A-1)from agent A,agent B conducts means-end reasoning about the plan-step that A has made a proposal about.It then evaluates A’s proposal by comparing it with the options it has generated by reasoning,and on the basis of this comparison(deliberation),it decides whether to accept or reject it.During this evaluation,it will attempt to retrieve the warrant propositions stored earlier in memory which are the beliefs that allow it to evaluate each proposal and to compare another agent’s proposal with the options that it has generated by its own means-end reasoning.Remember that B knows the utility information for all the items that he and A could propose,but that information may not be salient. Proposals,such as(A-1)and(B-1)in(3),are inferred to be implicitly ACCEPTED because they are not rejected(Car-berry1989).If a proposal is ACCEPTED,either implicitly or explicitly,then the option contained in the proposal becomes a mutual intention that contributes to the?nal design plan (Power1984;Sidner1994).Agents REJECT a proposal if deliberation produces an option with a higher utility.For ex-ample,in(B-2)B rejects the proposal of option-45in(A-2), proposing instead option-56.Either B could not recall the utility of option-45,or the utility of option-56is higher. Discourse Strategies for Hypothesis Testing

A discourse strategy is a particular way of achieving a DIS-COURSE ACT such as a PROPOSAL.Agents are parametrized for different discourse strategies by placing different expan-sions of discourse acts in their plan libraries.To test the hypotheses discussed earlier,we examine four strategies: (1)All-Implicit(2)Explicit-Warrant(3)Solipsistic(4)ESP. Each strategy varies the decision algorithm that is used when an agent is reasoning about whether to expand a proposal to include a reminder of the warrant for the proposal.

The All-Implicit strategy never includes warrants in pro-posals,leaving it up to the other agent to retrieve them from memory.An agent utilizing this strategy acts as though it assumes that everything the other agent knows is salient.In dialogue(3),both agents communicate using the All-Implicit strategy,and PROPOSALS expand to the PROPOSE communica-tive acts shown in(A-1),(B-1),and(A-2).

The Explicit-Warrant strategy always expands the PRO-POSAL discourse act to be a W ARRANT followed by a PROPOSE utterance(Suthers1993).An agent utilizing this strategy acts as though the other agent never retains anything in working memory,as though nothing is salient for the other agent.For example in(4)(A-1)is agent A’s WARRANT for his proposal in(A-2):

(4)A-1:Putting in the green rug is worth56.

A-2:So,let’s put the green rug in the study.

B-1:Putting in the green lamp is worth55.

B-2:So,let’s put the green lamp in the study.

The?nal two strategies sometimes include a warrant de-pending on the source of information that agent A uses about B’s attentional state in its decision algorithm.These are the strategies based on the ESP and SOLIPSISTIC hypotheses dis-cussed earlier;we call these the Solipsistic strategy and the ESP strategy.

In the ESP strategy,agent A maintains a detailed model of agent B’s attentional state.To implement the ESP strategy, we actually give agent A access to agent B’s mind in the simulation.ESP provides data on how a perfect attentional model of the other agent affects performance.

An agent using the Solipsistic strategy uses its own atten-tional state as a model for the other agent.To implement the Solipsistic strategy,we endow agent A with the capa-bility of keeping track of its own cognitive effort for all of the processing that A does.Agent A then uses its own re-trieval effort to determine whether it improves performance to remind B of a warrant for a proposal rather than letting B retrieve the warrant from B’s own memory. Evaluating Performance

Earlier,we discussed the model of performance that we as-sume,in which agents work as a team.Performance is the difference between the RAW SCORE for the task and COLLAB-ORA TIVE EFFORT.RAW SCORE is task speci?c:we simply sum the utility of the furniture pieces in each PUT action in the?nal plan.For example,the raw score for the design plan in Figure1is434points.

As we discussed above,COLLABORA TIVE EFFORT is com-posed of COMMUNICATIVE EFFORT and RETRIEV AL EFFORT. Our PERFORMANCE hypothesis is that the adaptive reminding strategies can improve performance by reducing collabora-tive effort.Thus,the decision algorithm for both the ESP and Solipsistic strategies is to say the warrant whenever the model of agent B’s AWM predicts that(1)the warrant is not salient or(2)the warrant will be more costly for agent B to retrieve than for agent A to communicate.

Note however,that any savings in retrieval effort is al-ways at the cost of an additional utterance,the reminder, which increases communicative effort.To provide a fair test of our adaptive strategies,we construct two adversarial di-alogue situations,one which favors reminding by making

each retrieval relatively costly,and one which favors not re-minding by making each communicated message relatively costly.We do this by using two different parameter settings for COMMCOST and RETCOST in our experiments when cal-culating collaborative effort.Retrieval effort dominates with unit cost settings of COMMCOST=.1and RETCOST=.001. Communicative effort dominates with unit cost settings of COMMCOST=.001and RETCOST=1106.

We experiment with four AWM ranges for the resource limits;LOW for very resource limited agents,MID for agents hypothesized to be similar to human agents,MIDH for agents less constrained than humans,and HIGH for resource unlim-ited agents.An AWM range indicates the maximum number of beliefs and intentions that are potentially salient for an agent at any processing step.6

We determined that a sample size of200dialogues per experimental condition is adequate for determining whether a strategy affects PERFORMANCE.To collect these samples we simulate200dialogues with the appropriate parameter settings,yielding a performance distribution for each strategy and set of assumptions tested.

We present our comparisons of dialogue strategies in plots of the differences in the mean PERFORMANCE of agent pairs using different strategies,such as the graph in Figure2.In Figure2,agent pairs using the ESP strategy are compared against agent pairs using the All-Implicit strategy.Differ-ences in the mean PERFORMANCE between the two strategies are plotted on the Y-axis against the four AWM parameter ranges on the X-axis.Each point in the graph at a particular AWM range is the difference in the mean performances of 200samples where the agent pairs both use the ESP strategy and200samples where the agents use the All-Implicit strat-egy.This graph summarizes the information from8total performance distributions(1600simulated dialogues).

To see which of the performance differences are signi?-cant we run planned comparisons using one-way analysis of variance(anova)where statistical signi?cance is determined by the modi?ed Bonferroni test(MB).When comparing two strategies for the same AWM range and unit cost settings, if the mean performance of strategy1is signi?cantly less than the mean of strategy2,according to the MB test,then strategy1is DETRIMENTAL and strategy2is BENEFICIAL for agents in that AWM range.

Experimental Results

In initial experiments,we duplicated W&R’s results.They found that the Explicit-Warrant strategy(always say the war-rant)compared to the All-Implicit strategy(never say the warrant)is bene?cial at the two highest AWM ranges,when retrieval effort dominates other processing effort,but that

Exp-Warr vs.All-Imp Adapt vs.All-Imp Adapt vs.Exp-Warr

MIDH MIDH MIDH Retrieval Dominates-Exp-Warr-Adapt--

---

Solipsistic vs. All-Implicit:

COMMCOST = 0.1, RETCOST = 0.001

ATTENTION/WORKING MEMORY RANGE P E R F O R M A N C E D I F F E R E N C E S

-40

-20020

40Figure 4:When retrieval effort dominates,Solipsistic is bene?cial compared to All-Implicit at MIDH and HIGH AWM ranges.

This shows that in at least some communicative situations,it may not be worth maintaining a detailed model of agent B,given the potential overhead of maintaining such a model,which we did not include in the calculation of collaborative effort.

To test this idea further,we added small percentages of the task de?ned RAW SCORE to collaborative effort to account for the overhead involved in maintaining the ESP model,and then compared performance again.We found that if the effort of maintaining the ESP model is greater than 4%of the RAW SCORE ,that Solipsistic performs as well as ESP at all AWM settings.

Discussion

Our goal was to test three hypotheses about deciding to re-mind:the PERFORMANCE ,SOLIPSISTIC ,and ESP hypotheses.We found support for the PERFORMANCE hypothesis:the decision to present a warrant should be based on whether doing so enhances overall performance.This is not particu-larly surprising.

To test the PERFORMANCE hypothesis,we evaluated two adaptive strategies for reminding,Solipsistic and ESP,in comparison with similar static strategies.Furthermore,we evaluated them under two different adversarial dialogue con-ditions,one which favors reminding and one which favors not reminding.We found that the adaptive strategies never perform worse than the static strategies.Furthermore,both adaptive strategies are superior to the two static strategies since a single adaptive strategy performs well over all dia-logue conditions.

We also found support for the SOLIPSISTIC hypothesis over

Solipsistic vs. Explicit-Warrant:

COMMCOST = 0.001, RETCOST = 1e-06

ATTENTION/WORKING MEMORY RANGE

P E R F O R M A N C E D I F F E R E N C E S

-40

-200

20

40

Figure 5:When communication effort dominates,Solipsistic is bene?cial compared to Explicit-Warrant at the MID AWM range.

the ESP hypothesis.The SOLIPSISTIC hypothesis is,that in deciding to remind,agent A can approximate B’s attentional state with its own.The ESP hypothesis is that agent A main-tains a detailed model of agent B’s attentional state to use in making decisions about reminding.To our knowledge,no-one has ever proposed or tested the SOLIPSISTIC hypothe-sis before.By comparing the Solipsistic strategy to the ESP strategy,we found that it may not be worth maintaining a de-tailed model of agent B’s attentional state,given the potential costs of doing so.

We interpret the results on the SOLIPSISTIC hypothesis to mean that the ?delity of the AWM model to the recency and frequency properties of human working memory produces an environment in which agent A’s attentional state is proba-bilistically correct as an approximation of B’s.With respect to modelling humans,this suggests that it may be a cogni-tively ef?cient strategy to use one’s own attentional state to estimate that of your conversational partner.With respect to building arti?cial agents,the results suggest that when two agents are engaged in a synchronous collaborative problem solving dialogue,the Solipsistic model may be a useful ap-proximation for agent A to use in deciding when to remind agent B of relevant facts.

Our methodology for achieving these results consists of a controlled experimental environment in which we can ma-nipulate variables that are relevant to our hypotheses.The environment implements a particular formal model of collab-orative problem solving.Thus,our experiments are directly affected by both the formal model and the experimental vari-ables that we manipulate,and provide a way to evaluate how experimental variables interact with the formal model.

Hanks,Pollack and Cohen discuss at length the im-portance of demonstrating that results collected in sim-ulation experiments generalize beyond the particulars of a testbed environment(Hanks,Pollack,&Cohen1993; Cohen1995).There are several reasons why we expect the results above to generalize.First,our simulation is based on a model of collaborative planning that is similar to other mod-els(Grosz&Sidner1990;Levesque,Cohen,&Nunes1990; Guinn1994;Cohen1995),and based on general assumptions about the underlying agent architecture(Bratman,Israel,& Pollack1988).Second,we tested reminding for a particular information con?guration:agent A makes a proposal and reminds B of the warrant that supports it.Since this is a general information con?guration that is found in any agent that deliberates,our models of reminding should extend to other problem-solving situations.Third,our AWM model has many similarities to a cache,thus Solipsistic strategies should be bene?cial in any architecture with a cache type memory such a SOAR(Lehman,Lewis,&Newell1991),as long as communication is approximately synchronous.Finally, because our performance evaluation depends on calculating performance for the team(Levesque,Cohen,&Nunes1990; Grosz&Sidner1990),we believe that our reminder models will generalize to any team-oriented problem-solving envi-ronment such as Phoenix(Cohen1995).

In future work,the implicated decision model can be tested in human-computer interaction,as in intelligent tutors,to see how well the model extends to non-homogeneous agents (Moore1994).Additional experiments could also determine whether the Solipsistic strategy breaks down under some circumstances,when the ESP hypothesis might hold,and whether adaptive strategies are bene?cial compared to the static strategies under less extreme processing assumptions.

References

Allen,J.F.1983.Recognizing intentions from natural language utterances.In Brady,M.,and Berwick,R.,eds., Computational Models of Discourse.MIT Press. Bratman,M.;Israel,D.;and Pollack,M.1988.Plans and resource bounded practical https://www.doczj.com/doc/455956288.html,putational Intelligence4:349–355.

Brennan,S.E.1990.Seeking and Providing Evidence for Mutual Understanding.Ph.D.Dissertation,Stanford University Psychology Dept.Unpublished Manuscript. Carberry,S.1989.Plan recognition and its use in under-standing dialogue.In Kobsa,A.,and Wahlster,W.,eds., User Models in Dialogue Systems.Berlin:Springer Verlag. 133–162.

Clark,H.H.,and Schaefer,E.F.1989.Contributing to discourse.Cognitive Science13:259–294.

Cohen,P.R.1995.Empirical Methods for Arti?cial Intel-ligence.Boston:MIT Press.

Giunchiglia,F.;Sera?ni,L.;Giunchiglia,E.;and Frixione, M.1993.Non-Omniscient Belief as Context-Based Rea-soning.In IJCAI93,548–554.

Grice,H.P.1967.William James Lectures.Grosz,B.J.,and Sidner,C.L.1986.Attentions,intentions and the structure of https://www.doczj.com/doc/455956288.html,putational Linguistics 12:175–204.

Grosz,B.J.,and Sidner,C.L.1990.Plans for discourse. In Cohen,Morgan and Pollack,eds.Intentions in Commu-nication,MIT Press.

Guinn,C.I.1994.Meta-Dialogue Behaviors:Improving the ef?ciency of Human-Machine Dialogue.Ph.D.Disser-tation,Duke University.

Hanks,S.;Pollack,M.;and Cohen,P.1993.Benchmarks, testbeds,controlled experimentation and the design of agent architectures.AI Magazine.

Joshi,A.K.1978.Some extensions of a system for infer-ence on partial information.In Pattern Directed Inference Systems.Academic Press.241–257.

Landauer,T.K.1975.Memory without organization:Prop-erties of a model with random storage and undirected re-trieval.Cognitive Psychology495–531.

Lehman,J.F.;Lewis,R.L.;and Newell,A.1991.Natural language comprehension in SOAR.Technical Report CMU-CS-91-117,Carnegie Mellon University.

Levesque,H.J.;Cohen,P.R.;and Nunes,J.H.T.1990. On acting together.In AAAI90.

Moore,J.1994.Participating in Explanatory Dia-logues:Interpreting and Responding to Questions in Con-text.MIT Press.

Pollack,M.,and Ringuette,M.1990.Introducing the Tile-world:Experimentally Evaluating Agent Architectures.In AAAI90,183–189.

Power,R.1984.Mutual intention.Journal for the Theory of Social Behaviour14.

Sidner,C.1994.An arti?cial discourse language for col-laborative negotiation.In AAAI94,814–820. Suthers,D.D.1993.Sequencing explanations to enhance communicative functionality.In Proceedings of the15th Annual Meeting of the Cognitive Science Society. Thomason,R.1990.Propagating epistemic coordination through mutual defaults I.In Parikh,R.,ed.,Proceedings of the Third Conference on Theoretical Aspects of Reasoning about Knowledge,29–39.Morgan Kaufmann. Walker,M.A.,and Rambow,O.1994.The role of cognitive modeling in achieving communicative intentions.In The 7th International Conference on Natural Language Gener-ation.

Walker,M.A.1996.Limited attention and discourse https://www.doczj.com/doc/455956288.html,putational Linguistics22(1).

Whittaker,S.;Geelhoed,E.;and Robinson,E.1993.Shared workspaces:How do they work and when are they useful? IJMMS39:813–842.

Zukerman,I.,and McConachy,R.1993.Generating con-cise discourse that addresses a user’s inferences.In Pro-ceedings of the Thirteenth International Joint Conference on Arti?cial Intelligence,1202–1207.Morgan Kaufmann Publishers,Inc.

新人教部编版三年级语文下册常用词语造句例句

三年级语文下册专项训练 常用词语造句 1、好奇 ★我看见新鲜的东西都觉得很好奇。 ★我看到路上有个黑色的袋子感到很好奇,想看看到底是什么。 练习:____________________________________ 2、引人注目 ★他以优异的成绩考到了全年级第一,格外引人注目。 ★她今天穿得像个公主,特别引人注目。 练习:____________________________________ 3、使劲 ★上坡了,我们都使劲推着车。 ★孩子高兴极了,使劲地亲了妈妈一口。 练习:____________________________________ 4、有趣 ★吹泡泡是一件很有趣的事。 ★我看一本有趣的漫画书,笑得我肚子都疼了。 练习:____________________________________ 5、一会儿……一会儿…… ★秋天来了,大雁向南飞,一会儿排成一字,一会儿排成人字。 ★天上的白云一会儿像可爱的动物,一会儿像美丽的仙女,真是千奇百怪。★动物园里的猴子一会儿爬上,一会儿窜下,可调皮了。 ★春天的天气像娃娃的脸,一会儿晴,一会儿雨。 练习:____________________________________ 6、一本正经 ★班长一本正经的告诉我,我的语文成绩在年级得了第一。 ★爸爸说话总是一本正经。 练习:____________________________________ 7、自言自语

★爷爷总是自言自语,看起来很孤独。 ★小明上课总是自言自语,管不住自己。 练习:____________________________________ 8、甚至 ★今天作业不但多,还很难,甚至做到了9点还没做完。 ★他品行不好,总是骂人,甚至打人。 ★他很爱学习,周末一整天都在屋里看书,甚至都忘了吃饭睡觉。练习:____________________________________ 9、仿佛(好像)(似乎) ★天上的白云仿佛是一只只绵羊。 ★美丽的草原仿佛是一片绿色的海洋。 ★蓝蓝的天空仿佛是一片没有尽头的大海。 练习:____________________________________ 10、欢蹦乱跳 ★猴子们欢蹦乱跳地在树林里玩耍着。 ★小鸟在枝头欢蹦乱跳的歌唱。 练习:____________________________________ 11、终于 ★经过努力,我终于考了100分。 ★经过老师耐心的讲解,我终于明白了这个题目。 ★经过努力,我终于顺利跑到了终点。 练习:____________________________________ 12、也许 ★他到中午也没来上课,也许是因为病了。 ★爷爷没有准时来接我,也许是忘了。 练习:____________________________________ 13、果然 ★妈妈说明天会下雨,第二天果然下雨了。 ★昨天小明说今天考试,今天果然考试了。

久别重逢造句 .doc

久别重逢造句 1.凤西木是个很冷情的人,日子双腿虚设未曾怨尤半声,而此时对待林占也同样表现出同样的本色,没有久别重逢的喜悦,也没有前仇旧恨的怨憎。 2.久别重逢似乎有着说不完的话,他与大雕聊了许久。 3.两个久别重逢的爱人去了汽车旅馆。 4.晚上我和盈婷挤在我以前睡过的那张床上,久久不能入眠,久别重逢的兴奋与辛酸让我百感交集,我真的应该早点回来的,而这份孝心却迟到了这么多年。 5.听你诉说,清如天籁,身边就是一个久别重逢的好友,一个朝夕相处的亲人,温馨和谐,舒适惬意。 6.浣溪怅然若失,多少次午夜梦回,重回梅姐的小屋,多少次在心中预演久别重逢的场景,没想到结局却是如此残酷。 7.充实在银行大堂里我遇到以前的老师有了久别重逢的谈话。 8.两人久别重逢,自有许多话说。 9.久别重逢,互诉衷情。 10.强烈而震憾的一响,两只宽大的手掌一拍即合,http://..久别重逢的故人的心犹如这手又紧紧合在了一起。 11.沉浸在久别重逢的喜悦中的秦向阳,没有察觉到君冷澈的冷淡,快步走到床边,关心地低头看童真真。 12.或许,我整理书籍得到的最大的快乐是一种惊喜或者更好的说是一种久别重逢的喜悦。

13.许多来宾感动得边看边落泪,当录影带中师父亲自演唱的骊歌歌声轻轻响起时,同修更是潸然泪下,那是久别重逢的母亲在呼唤爱儿的呼声,句句震彻心扉声声萦绕耳际。 14.但是他却没死,谷底另有洞天,她还见到了久别重逢的小龙女。 15.兄弟久别重逢,兴奋不已,于是二人重新落座,斟茶拨火,谈论起来。 16.我俩久别重逢,我心里无比激动,可是看着他优哉游哉的样子,我的心顿时凉了。 17.屋子里充满了久别重逢后的喜悦,然而就在这时,灵宝说他明天一早将回到灵霞村去。 18.久别重逢,互诉衷情。 19.久别重逢,自是高兴。 20.可是这是久别重逢啊,再次见面,他可不想看到一个一本正经的美女。 21.久别重逢的一对好朋友紧紧相拥,又哭又笑的好不激动。 22.旁人只见一幅母女久别重逢,抱头痛哭的画面,却不知她哭她妈的久别,我哭我妈的不见。 23.笼子里的雁伸出了它的头,那两只雁的颈就交缠在一起,就像人间夫妇久别重逢一样。 24.久别重逢,格外亲切。 25.好多加油站是我们福建人开的,虽然是莆田人,你一样可以感受到象久别重逢的亲人,哪乡音,哪乡情。

人教版 小学语文四年级上册 造句大全 (精编版)

四年级语文用下列词语造句: 笼罩:①大雾笼罩了整个江面,分不清天南地北。 ②在大雾笼罩下,远处的景物若隐若现。 依旧:①虽然老师再三教育,可他依旧不认真学习。 ②这件衣服她已经穿了两年,依旧和新的一样。 横贯:①325国道横贯鹤山,这有利于鹤山更好地发展。 ②雄伟的长江大桥横贯大江两岸。 恢复:①天色渐渐暗下去,船停泊靠岸,湖面上又恢复了平静 ②经过半年的疗养,他的健康已完全恢复了。 沸腾:①锅里的水沸腾了,一股股水蒸气直冒出来。 ②下课了,同学们纷纷跑出教室做游戏,整个校园顿时沸腾起来。 宽阔:①很多汽车在这宽阔的公路上行驶。 ②我喜欢独自一人在这个宽阔的广场上散步。 灿烂:①这天,风平浪静,天空被灿烂的朝霞染红了 ②在一个阳光灿烂的上午,我和爸爸妈妈到公园游玩。 规律:①这几乎成了规律,每个星期六他总要骑车到婆婆家玩。 ②我们应当从实践中,找出事物运动的规律来。 静寂:①天色渐渐暗下来,周围一片静寂。 ②突然一声枪响,冲破了这夜晚的静寂。 均匀:①他在吃药以后慢慢睡着了,发出均匀的呼吸声。

②妈妈把菜籽均匀地撒播在地里。 痕迹:①海龟爬过海滩上,留下了一道深深的痕迹。 ②草地上留下了一些动物走过的痕迹。 休想:①你上课不好好听讲,休想取得好成绩。 ②你不买票,休想进内参观。 逐渐:①到了秋天,树叶逐渐变黄了,一片一片地从树上落下来。 ②经过不断努力,他的学习成绩逐渐提高。 引人注意:①她今天穿着一条大红的裙子,格外引人注意。 ②动物园里,最引人注意的是大熊猫。 隐蔽:①游击队员隐蔽在山上密密的树林里。 ②趁他们还没发现,你要赶快隐蔽起来。 毫不可惜:①今天早上他把新买的手表弄丢了,但他觉得毫不可惜。 ②他毫不可惜地扔掉了爸爸买给他的玩具车。 随遇而安:①无论到了哪个地方,他都能随遇而安。 ②在任何环境下,战士们都可以随遇而安。 选择:①他选择了一个星期天去公园玩。 ②我们选择了一个较宽敞的地方一起做游戏。 搜索:他搜索了大半天,但什么也没有找到。 隐藏:我把东西隐藏起来,谁也不可能发现。

道歉造句大全

道歉造句大全 导读:本文是关于道歉造句大全,如果觉得很不错,欢迎点评和分享! 1、向你道歉,刚才胡言乱语,我不再乱讲话了,别介意啦! 2、他必须道歉,这他再明白没有,他心想,非低一下头不可了。 3、你向他赔礼道歉就行了,何必肉袒负荆。 4、向客户道歉,效果是非常显著的,千万不要低估他。对于没有达到期望值的客户,无论怎样强调道歉的重要性,都不过分。 5、他们最后不但同意道歉,而且还要付款。 6、士兵有权要求召开会议,批评军官和向上级领导告状,如果判明哪个军官有错误,他通常要在群众大会上道歉或作检讨。 7、她生气是因为你错怪了她,解铃还须系铃人,还是你去向她赔礼道歉吧。 8、他认识到了自己的错误,只好赔礼道歉。 9、吵架、冷战、道歉是年轻时恋爱必定经历的循环,而当这个循环终止,青春便即将散场。以为偶尔错过的那个人,终究会化成心底里化不开的影子。那个时候,其实真的一转身就是一辈子。 10、无论他怎么道歉,对方就是不肯高抬贵手,撤回告诉。 11、任何场合,保持应有的涵养。学会说谢谢、辛苦您、对不起。做错了事情要懂得道歉和改过。 12、你这样放肆,非向我道歉不可!不然,你给我马上滚!

13、我想赤裸的永远不遮掩,我要疯狂的永远不道歉,我最自己的永远,不怕你讨厌。 14、福拉姆敦见她姑妈急急走来并连声道歉表示怠慢,心中宽慰不少。 15、爱本无罪,请你们相信。当你被父母教训了,请你记住,爱在此时是为你好;当你和同学吵架了,请你记住,你们之间有份爱,谁错了就应该去道歉的;当你恨一个人时,请记住,这个丗界是美好的,忘记一切,重新来过。 16、难得是诤友,当面敢批评;有时不能忍,猝然发雷霆;继思不太妥,道歉亲上门;于是又合作,相谅心气平。 17、但是万幸的是,她接受了我的道歉,而且我向她保证两个人不在一起的时候绝不想东想西,庸人自扰。 18、他不愿道歉、有着18世纪复古风的性格,即使在华盛顿惯常的有关名誉和地位的蜜罐中,依然我行我素。 19、内塔尼亚胡为在错误的时间公布这一决定做了矫揉造作的道歉,却拒绝撤销这一决定。 20、与其花时间去向如何跟朋友和好,还不如直接去向朋友道歉。 21、白云,黑土向你道歉,请你睁开眼看我多可怜,今天的你我还能否重复昨天的故事,我的这张旧船票还能否登上你的破船! 22、我说过我要道歉,乔丹夫人,我现在就要这样。 23、在许多博客眼里,这个道歉让莎朗斯通显得为了拯救自己

现代汉语连词意义及用法总结

连词 连词是用来连接词与词、词组与词组、句子与句子,表示某种逻辑关系的虚词。连词可以表并列、承接、转折、因果、选择、假设、比较、让步、递进、条件、目的等关系。一般说来,连词有很多是由副词、介词发展而来的;很多副词、介词又是由动词发展而来。所以动词、介词、连词需要区别。 和 =跟、=与、=及、=同<方>。表示并列关系、联合关系。 车上装的是机器和材料。他的胳膊和大腿都受伤了。工人和农民都是国家的主人。工业与农业。批评与自我批评。我同你一起去。图书、仪器、标本及其他。(注意:用“及”连接的成分多在词义上有主次之分,主要成分放在“及”的前面) 以及 连接并列的词或词组(“以及”前面往往是主要的):院子里种着大丽花、矢车菊、夹竹桃以及其他的花木。 既=既然 既来之,则安之。既然他不愿意,那就算了吧。既然知道做错了,就应该赶紧纠正。你既然一定要去,我也不便阻拦。既要做,就一定要做好。既然这样,还不去做?用在上半句话里,下半句话里往往用副词就、也、还跟他呼应,表示先提出前提,而后加以推论。 既然。。。就。。。既然。。。也。。。既然。。。还。。。 继而=既而 先是惊叹,既而大家一起欢呼起来。人们先是一惊,继而哄堂大笑。先是一个人领唱,继而全体跟着一起唱。表示紧随在某一情况或动作之后。 而 1、连接动词、形容词、词组、分句。 (1)连接语意相承的成分。伟大而艰巨的任务。战而胜之,取而代之,我们正从事一个伟大的事业,而伟大的事业必须有最广泛的群众的参加和支持。 (2)连接肯定和否定相互补充的成分:此种花香浓而不烈,清而不淡。马克思主义叫我们看问题不要从抽象的定义出发,而要从客观存在的事实出发。有转折的意思。 (3)连接语意相反的成分,表转折:如果能集中生产而不集中,就会影响改进技术、提高生产。 (4)连接事理上前后相因的成分:因困难而畏惧而退却而消极的人,不会有任何成就。 2、有“到”的意思。一而再,再而三。由秋而冬。由南而北。 3、把表示时间、方式、目的、原因、依据等的成分连接到动词上面。匆匆而过、挺身而出、为正义而战、因公而死、视情况而定、三十而立。 而且 表示更进一步,前面往往有“不但、不仅”跟它呼应:性情温和而且心地善良。他不仅会开汽车,而且会修汽车。不但战胜了各种灾害,而且获得了丰收。 因而=因此 表示结果。下游河床狭窄,因而河水容易泛滥。 因为 常跟所以连用,表示因果关系。因为今天事情多,所以没去成。 何况 用反问的语气表示更进一层的意思。这么多事情一个人一天做完是困难的,何况他又是新手。他在生人面前都不习惯讲话,何况要到大庭广众之中呢?(even)

一来二去造句 .doc

一来二去造句 1.这一来二去,几个小时不知不觉的过去了。 2.因为李航为人直爽,虽说碍于就业压力,变得圆滑起来,但对乔峰还是非常真诚的,一来二去,两者就成为了极好的朋友。 3.他们常在一起踢足球,一来二去地也就熟了。 4.一来二去的,两个人你恩我爱,好的不得了。 5.周大神,是左邻右舍对周四儿的叫的外号,因为周四儿会算卦,会看手相,邻里家谁有点什么事都找周四儿算,一来二去就都管他叫周大神了。 6.的常客,一来二去、耳闻目染的,渐渐地对大慧公司关注了。 7.三者杂而不和,使其冲击愈演愈烈,再加上云昊不知道如何缓解,盲目的催动真丹来调和三股真元,一来二去之下,使冲击显得更加激烈。 8.这赵宝来就是玩玉器这一行的,不时会去他师父陆老那请教,有时还请陆老给他掌掌眼,一来二去这赵宝来就和这陆老的关门弟子丁铭熟了。 9.就这么一来二去的,杨教授就记住了这个老爱来打断自己研究的学生了。 10.一来二去的,此女一定会想办法帮主公圆了这个念想儿!如此一来,岂不美哉? 11.一来二去倒也认识了光头男和马尾少女,因为这两人隔三差五也会到训练场修炼。

12.一来二去,我摸着一些门道:有的喜阴,就别放在太阳地里,有的喜干,就别多浇水。 13.两个人彼此都有好感,平时多接触一下,一来二去的也就能走到一起了。 14.一来二去这种念头在她心中拧成了死结,这让她一时间显得有些无所适从起来。 15.可是意外的是,妍姗姗把大部分财产拿给李林去了,紧接着,妍云夕动手术又花费了不少钱,这么一来二去,剩下的,也确实没有多少钱了。 16.一来二去变得熟识了,小西有事找到李程,李程都会调侃的说“小西,干嘛,是不是想我了呢?还是终于要和我表白了!”。 17.一来二去,大头也就相信了苗人,苗人感恩戴德,临走时还给大头留下了电话。 18.一来二去,整个江府里的男男女女都知道了,唯一瞒着的只有那个极其不肖,从不归家的江大少爷。 19.可是小胖子就认准楚凡了,没事就找楚凡玩,一来二去的,天龙城的人就能看见,一个大傻子带着一个小胖子在街上四处乱走。 20.都是这个酒保招待的,一来二去,两人也就熟悉了,偶尔还会瞎聊几句打发时间解解闷。 21.她的一记扫腿也被我挡了下去,一来二去倒有了相持不下的意思。

小学三年级语文常用词语造句例句

三年级语文下册专项训练 语文教学组 常用词语造句 1、好奇 ★我看见新鲜的东西都觉得很好奇。 ★我看到路上有个黑色的袋子感到很好奇,想看看到底是什么。 练习:____________________________________ 2、引人注目 ★他以优异的成绩考到了全年级第一,格外引人注目。 ★她今天穿得像个公主,特别引人注目。 练习:____________________________________ 3、使劲 ★上坡了,我们都使劲推着车。 ★孩子高兴极了,使劲地亲了妈妈一口。 练习:____________________________________ 4、有趣 ★吹泡泡是一件很有趣的事。 ★我看一本有趣的漫画书,笑得我肚子都疼了。 练习:____________________________________

5、一会儿……一会儿…… ★秋天来了,大雁向南飞,一会儿排成一字,一会儿排成人字。 ★天上的白云一会儿像可爱的动物,一会儿像美丽的仙女,真是千奇百怪。★动物园里的猴子一会儿爬上,一会儿窜下,可调皮了。 ★春天的天气像娃娃的脸,一会儿晴,一会儿雨。 练习:____________________________________ 6、一本正经 ★班长一本正经的告诉我,我的语文成绩在年级得了第一。 ★爸爸说话总是一本正经。 练习:____________________________________ 7、自言自语 ★爷爷总是自言自语,看起来很孤独。 ★小明上课总是自言自语,管不住自己。 练习:____________________________________ 8、甚至 ★今天作业不但多,还很难,甚至做到了9点还没做完。 ★他品行不好,总是骂人,甚至打人。 ★他很爱学习,周末一整天都在屋里看书,甚至都忘了吃饭睡觉。

and的用法及含义

and的用法及含义 And是英语中一个普通的连词,然而and并非只作并列连词用,它还具有一些不太常见的表达方式和意义,应根据上下文的特殊环境,作出判断才能准确理解其用法和意思。 1.and作并列连词,译为“和、并且”等,当连接三个以上的并列成分时,它放在最后一个成分之前,其余用逗号分开,例如: He bought a book and a pen.他买了一本书和一支笔。 Solid,liquid and gas are the three states of matter.固态、液态和气态是物质的三种状态。 2.名词+and+名词,若这种结构表示一个概念时,and+名词相当于介词with+名词,译为“附带、兼”的意思,例如: Noodle and egg is a kind of delicious food.(and egg=with egg)鸡蛋面是一种美味食物。 Whose is this watch and chain?(and chain=with chain)这块带表链的手表是谁的? 3.名词复数+and+同一名词的复数,强调连续或众多的含义。例如: There are photos and photos.照片一张接着一张。 They saw film hours and hours last week.上星期他们一小时接一小时地看电影。 4.形容词+and+形容词,这种结构形似并列,实际并非并列结构。例如: This roon is nice and warm(=nicely warm). The coffee is nice and hot(=thoroughly hot). 5.用and连接动词的用法:and+动词作目的状语。动词go(come,stop等)+and+动词,此时,and+动词相当于in order to+动词,例如: ①I'll go and bring back your boots,(go and bring back=go in order to bringback)我去把你的靴子拿来。 ②and+动词,起现在分词的作用,表示方式或伴随情况。例如: He sat and waited.(and waited=waiting)他坐着等。 ③and+同一动词表示动作长时间地“继续”或“重复”,例如:

用一不造句

用一不造句 导读:本文是关于用一不造句,如果觉得很不错,欢迎点评和分享! 1、幸福对我来说应该是尖尖的、闪闪发亮的星星。锐利的尖角不经打磨过,所以一不小心碰触到星星的尖角就会被刺痛,不过正因为痛楚,这才能体现出生存的意义。 2、你是一尊象牙雕刻的女神,大方、端庄、温柔、姻静,无一不使男人深深崇拜。 3、每个人都可以将自己的烦恼写成一本书,厚厚的一摞下来,多年后回忆起来,也不过是年少轻狂时的大梦一场。梦中的自己欢乐喜忧,一一不为之牵动。梦醒之后,想起梦中的自己所烦恼的事情,微微一笑,也就这么过去了。 4、澎湃的大海,巍峨的高山,绚丽的彩虹,这无一不是大自然的杰作。大自然不仅给了我们美丽的景色,更交给了我们做人的道理。大自然,真是人类的母亲,人类的老师! 5、你怎么一点同情心都没有?俗话说君子动口不动手,我一不动口二不动手,我就活动活动心眼儿,我都快死的人了,你跟我较什么真儿呀? 6、骄傲是毒药,只会让人软弱无力,倒在前进路上;骄傲是万丈深渊,让你一不小心跌下深谷再无法爬上来;骄傲是汹涌的浪涛,它会将你自大的船只打翻。

7、别看弟弟年纪小,可象棋下得很精,每一步都能步步为营,我要是一不留精就会输给他。 8、学习和思考,两者缺一不可才能获得知识。 9、在我们的课本中还描写了蟋蟀的住宅,它用自己小小的,柔软的身体,建造出复杂又细致的工程,令人佩服,正当我看着它,读着它的时候,一不留神这个小东西逃跑了,哎,我的蟋蟀标本做不成了。 10、乌龟和兔子比赛,兔子一不留神,就输给了乌龟,所以胜利是不是必然的,一定是经过努力才能得到的。 11、有些人,你以为可以见面的。有些事,你以为可以一直继续的。然后,也许在你转身的那个刹那。有些人,你就再也见不到了。当太阳落下,又升起来的时候,一切都变了,一不小心就在也回不去了。 12、我小声吟唱着,害怕一不注意就走调了。我的两脚微曲,不敢绷直,只要一崩直就会不停地发抖,整个身体就像泄了气的皮球,没有力气来支撑。 13、校园的生活是充满挑战的,只要一不小心,不认真,就会被别人远远地甩在后面。思维是随时都紧张的。 14、光洁白皙的脸庞,透着棱角分明的冷俊;乌黑深邃的眼眸,泛着迷人的色泽;那浓密的眉,高挺的鼻,绝美的唇形,无一不在张扬着高贵与优雅,这,这哪里是人,这根本就是童话中的白马王子嘛! 15、时间像掌中的水,一不留神,就从指缝中流光了。

胆大妄为造句大全

13、令人不解的是,许多大公司的管理者被管理时尚和口号所 左右,迷恋于胆大妄为、突飞猛进、富有创意的领导行为,拒绝改 变路线,全然不顾已经完全改变了的形势。 14、这个曾国藩也太胆大妄为恣肆无忌了!咸丰忿忿不平的想。 15、一个浑浑噩噩的家伙,往往胆大妄为,毫无忌惮。 16、或者,是不是以一种最愚蠢的胆大妄为的方式开启了人类 灭亡的结局? 17、这个主竟使裘很得意。她就喜欢做胆大妄为的事情。 18、你这样胆大妄为,干尽坏事,是不会有好结果的。 19、想不到他监守自盗,偷天换日,真是胆大妄为。 20、犯人因绝望而越来越胆大妄为。 21、这件事太不顾羞耻,太不顾名誉和利害关系了,他不会这 样胆大妄为。 22、这在对国内企业间恶意收购行为闻所未闻的中国来说,实 在是胆大妄为得不同寻常。 23、这台机器很快变得更加胆大妄为起来。 24、这几个歹徒真是胆大妄为,竟敢在光天化日之下抢劫他人 财物。 25、刚才,我不在的时候,说不定你又干了些别的胆大妄为的 淘气事。 26、越靠近村庄的地方问题越严重,在数目膨胀的猴群中,猴 子们的行为越来越胆大妄为,偶尔还会攻击人类,甚至从住屋内偷 取食物。 27、嫌疑人行动诡秘,胆大妄为,连续作案,且其反侦查意识 极强,现场几乎没有留下有价值线索。

28、他是个胆大妄为的人,从来不怕任何事。 29、你真是胆大妄为,竟敢考试作弊。 30、法律也是一样,因为从不施行的缘故,变成了毫无效力的 东西,胆大妄为的人,可以把它姿意玩弄;正像婴孩殴打他的保姆 一样,法纪完全荡然扫地了。 31、楚枫的逆天战力,楚枫的胆大妄为,楚枫嗜血成性,楚枫 的疯狂行为,已是在九州大陆传的沸沸扬扬。 32、孙方目光瞄向史大庸,如果此刻突然将史大庸控制住,以 史大庸性命威胁,家丁们多半会乖乖束手就擒,那么葫芦岩这群胆 大妄为的山贼此次冒险便会成功。 33、为什么戴维做出这些胆大妄为的事情? 34、他敢和我分庭抗礼,真是胆大妄为。 35、这些人都是胆大妄为的家伙,虽然我们可以打他们个措手 不及,但我们要谨慎小心,否则他们就可能使我们受到某种损害。 36、胆大妄为的是你——冷秋月,也不过是姥姥身边的一条狗,主人还没有叫,就先撒欢儿了,拼命去讨好主人,让人恶心? 37、这个盗贼胆大妄为到了极点,竟然偷到了公安**的头上。 38、出了个胆大妄为的乖张孩子当然令整个家族坐立不安。 39、在我们这个社会主义国家里,是不容许不法分子胆大妄为的。 40、在猫王故居雅园举办的“猫王展”上,一名胆大妄为的小 偷施展空空妙手盗走了一把手枪。 41、少喝酒,可以很好地释放情绪,而一旦喝多了,酒壮怂人胆,不仅会导致暂时推迟记忆,还会让饮酒者做出一些胆大妄为的 事情来。

一举两得造句大全

一举两得造句大全 导读:本文是关于一举两得造句大全,如果觉得很不错,欢迎点评和分享! 1、刘老师的博客既能学知识,又能学做人,真是一举两得。 2、走进低碳生活,我们既能节约能源,又能健康快乐,真是一举两得。 3、在寒假,爸爸给我报了奥数班学习数学。这样我在寒假里,既能玩一玩,也能学到数学,真是一举两得,我同意这个方案。 4、寒假,我和爸爸将回四川乐山,这样即看了爷爷,又去参观了东山大佛,真是一举两得。 5、萌萌真可谓是一举两得呀,她只做了一件好事,却获得了两个荣誉。 6、他做好事一举两得,伙伴们知道了,都纷纷跑去做好事。 7、我早上为了不浪费时间,一边听英语一边写作业,真是一举两得。 8、利用废物,既可变废为宝,又可减少空气污染,是一举两得的好事。 9、我学着玩有趣的魔方,既锻炼了手的灵活性,也让脑子转得更快,一举两得。 10、妈妈教我合理的运用时间学玩分明,这样我就可以一举两得,事半功倍了。

11、今天我帮妈妈搬了一箱面条,即锻炼了身体,又帮了妈妈,真是一举两得。 12、只要我们动手把教室搞干净,我们既可以在环境好的地方学习,又可以美化校园卫生,真是一举两得。 13、我们大都市应该多多植树,这样不仅可以美化环境,还可以服务于大众,净化出新鲜的空气,真是一举两得! 14、我去百货商店买些布,顺便到托儿所把我儿子接回来,一举两得。 15、我们到公园跑步,顺便照些相,一举两得。 16、我回老家了,可以跟表哥们玩,又可以上网吧上博客跟大家在一起,真是一举两得。 17、这次期末我考试得了全班第二名的好成绩,老师表扬了我,同时也证明了我地努力没有白费,真是一举两得。 18、我要建议妈妈上下班别骑摩托车,和我一起步行,既保护了环境,又锻炼了身体,真是一举两得。 19、我今天因考试得了一百分,在校老师表扬我,在家父母奖励我,一举两得的滋味真爽。 20、如果我能用功学习的话,成绩就会提高。这样老师和家长就不会生气了,那将是一举两得的事。 21、你实行半工半读,既可保证生活,又能继续学业,可说是一举两得。 22、过新年了,可以一举两得,既能吃上美味的年夜饭,也可

用关联词造句的语文二年级句子练习题

用关联词造句的语文二年级句子练习题(*) 用因为……所以……造句 因为简单,所以快乐 我因为自信,所以最棒 因为我小,所以爸爸让我先说 是因为新开发,所以票价便宜呀 我因为没有雨伞,所以一直在等 我的爷爷因为好学,所以知识渊博 因为公主很聪明,所以才这样规定 也正是因为短暂,所以才格外宝贵 是因为自卑,所以才学习一再下降 因为爸爸很胖,所以妈妈叫他胖胖 正因为好看,所以破了都不舍得扔 人间因为有爱,所以拥有美好明天 这是因为我肚子痛,所以没去上学

因为他没有捂耳朵,所以才这么说 因为这里是座庙,所以可以烧香 他因为自己跑得快,所以特别骄傲 天因为我出门匆忙,所以忘记带雨伞 因为不想让海德发现,所以盖上了草 第二天因为我太兴奋,所以起得很早 美美因为挤不过他们,所以被淘汰了 (*) 用因为……因此……造句 也许它是因为它的形状像一根石柱,因此得名的吧 “无”牌因为不堪寂寞,因此将小樱身上的牌全部夺走 基础部分明明能全做对的,都是因为没认真审好题目,因此失了分 一本书生在一个农家里,因为那家太穷买不起多少书,因此这本书觉得很无聊 这是几周前的一个早上,因为前一天我和一位同学发生了矛盾,因此心里不太开心 今天是语数英联赛的日子,因为这次竟赛到我们小学六年级的素质加分,因此大家都看得比较重要 世代文人墨客,大多都因为母亲对儿女的无私而感动,因此为无

私的母爱留下了无数爱的赞歌 昨天下午,我因为在那位*的同学录上把我的好姐妹的绰号写了上去,因此而惹恼了我的好姐妹 弱者不敢面对挫折,因为他们害怕自己的再一次努力又遭无情的拒绝,因此一直在失败的阴影中迷失了方向 女人不应该放弃自己的理想,因为任何人都要在社会上走动才能找到归属感,因此女人应该有一份自己的事业 今天是开学后的第三个星期,星期二我被老师批评了是因为自己在自习课上借课外书看,因此劈头盖脸的批评了 一方面我不能从轿车和电动自行车的中间穿过去,另一方面又是因为我不能转向另一个方向,因此就导致我和电动自行车撞上了现在,因为它是一项易学易懂的手工艺爱好,因此流行非常广泛,受到不同年龄的人们的喜爱 但是回到家后我发现我感冒了,因为怕妈妈会骂我,因此我在妈妈面前装的精神饱满,暗地里却无精打采 因为有太多人要指引,因此怠慢了你 因为亲情是伟大的,因此我们感恩亲情 因为我有老师的辅佐,因此我做得十分精致 因为羊肉泡馍物美价廉,因此深受百姓的喜爱 因为“黑斑”背上、耳朵和眼睛周围都是黑色的,因此得名 因为四个人的力量比一个人的力量要大得多,因此我们要团结起来

用那么那么的词语造句

用那么那么的词语造句 造句,动词词语,是指用词语组织句子。今亦以指初等学校语文练习内容之一。用那么那么的词语造句,一起来看看。 1、天那么高,那么蓝,真是让人心旷神怡啊! 2、美丽的春天,是那么美丽,那么俏皮,像一个可爱的女孩. 3、看着妈妈连夜给我做的书包,我心里那么的感动,那么的难以平静。 4、你既然那么想造句,那么我就造一个句子吧! 5、你是那么的笨,那么的傻。 6、夏天的夜晚,是那么的宁静,那么的美丽。 7、天那么高,那么蓝,真是让人心旷神怡啊! 8、他那么高,那么瘦,身材太好了。 9、月亮是那么皎洁,那么明亮。 10、乡村的夜晚,只见一轮圆月挂在当空,薄薄的雾气笼罩着大地,这一切显得那么安详,那么宁静,那么唯美,那么超凡脱俗。 11、母亲对孩子是那么的无私奉献,那么的无微不至。 12、美丽的春天,是那么美丽,那么俏皮,像一个可爱的女孩。 13、妹妹是那么天真,那么可爱! 14、这个故事中描述的情节是那么曲折动人,人物是那么栩栩如生,真叫人难忘! 15、西藏的雪山是那么的高大,那么的圣洁,让人心里有种想要膜拜的感觉

16、大家都是那么无聊,那么闲。 17、湖水是那么清,那么静。 18、夜晚的月亮是那么澄澈,那么委婉. 19、天那么高,那么蓝,真是太美了。 20、她是那么美丽,那么善良。 21、夜晚的月亮是那么澄澈,那么委婉。 22、你既然那么想造句,那么我就造一个句子吧! 23、今天的阳光那么明媚,那么温暧,真舒服啊! 24、世界那么的神奇、那么的博大。 25、爸爸骄傲的说,儿子像我一样的那么帅那么牛。 26、白云是那么纯洁,那么淘气。 27、那朵花那么鲜艳,那么香,让人赞不绝口。 28、夜晚的月亮是那么澄澈,那么委婉。 29、美丽的花朵是那么鲜艳,那么艳丽。 1、夜晚的月亮是那么澄澈,那么委婉。 2、母亲对孩子是那么的无私奉献,那么的无微不至。 3、妹妹是那么天真,那么可爱! 4、他那么高,那么瘦,身材太好了。 5、夏天的夜晚,是那么的宁静,那么的美丽。 6、冬日的早晨,天气是那么的冷,街上的人是那么的少。 7、这孩子那么可爱,那么聪明,怎能不讨别人喜欢呢? 8、天那么高,那么蓝,真是让人心旷神怡啊!

颦蹙造句 .doc

颦蹙造句 1.墨欣寒颦蹙着眉头,看着不远处站在台上的逸风。 2.黛眉颦蹙,连忙将宇枫搀扶进山洞,眸子光芒转动有些不忍宇枫。 3.她注意到我在看她,就微微皱起了眉头,这一颦蹙倒令她更加好看了。 4.不施粉黛而颜色如朝霞映雪,眼球乌灵闪亮,双眉微微颦蹙,惹人爱怜。 5.沉睡之中的艾莉似乎正在做着噩梦,她的秀眉微微颦蹙着,安放在被中的一双小手紧紧撰着。 6.他骂的如此投入,没有注意到卖菜老板娘从找他的一叠零钱里飞快的抽走一张,没有注意到身旁也在挑菜的美女秀美颦蹙,面露惊奇。 7.婉儿颦蹙一笑道,原来她虽然在后面安静的坐着,看似在发呆,其实她对周围的情况早已经做了观察,这是她杀手的潜意识,这不,她已经发现了郑小余看她了。 8.看着床上的小丫头双眉颦蹙,在梦中也还在害怕吗?宇文律不禁握住叶晨冰冰冷的纤手,轻声说道“有我在,傻丫头、”。 9.走了几步,推开玻璃窗,看着远方颦蹙道。 10.,说完,司机摇了摇头,离开了,我莫名地望着车远去的样子,双眉颦蹙,有一种说不出的丝凉,风不时撩动我的头发。 11.少女长相妩媚,眉目含情,即使此刻眉头颦蹙,依然别有一番

风情。 12.白净的脸蛋完美无瑕,但那双凤目里已隐约有些杀气,黛眉颦蹙。 13.他看着少女颦蹙着的眉头,扁起来的嘴巴以及混杂着雨水和泪水的脸庞。 14.他七尺身躯,瘦骨嶙峋,蹙脊蹒跚,眉峰颦蹙,双眼嘘咪,形似羸弱,貌似枯槁,一副疲惫不堪之态。 15.只见镜中那本来倾国倾城的少女此时的脸色却是红润退了许多,换上来的是些许苍白,柳眉不自觉地颦蹙着,就连那檀口上都是有一些干涩。 16.她自然猜透了琉璃的想法,眉头微微颦蹙。 17.病痛使她时时颦蹙双眉。 18.墨倾城双眉颦蹙,抚了抚洛清狸眉间迭起的褶纹,懒懒散散地打了个哈欠,在他看来,没有什么比睡觉更重要的了。 19.颚颔微垂引得帝王丢了江山,嘴角微翘醉得王孙公子失去家业,柳眼微咪颦蹙之间妇人竞相效仿。 20.谁知道女孩颦蹙着黛眉,似乎本来心情就不是很好,这下被石空一撞,当场就要发作。 21.老王颦蹙着眉头不知道在想些什么。 22.瑞尔斯刚刚问完,卡狄修斯就双眉颦蹙了起来。 23.她捂着脑袋,双眉颦蹙,微扬的精致脸孔,一如白瓷,在阳光下闪烁着透明细腻的光,尽收于他眼底。

一举两得的近义词

一举两得的近义词 推送的一举两得的近义词,希望给大家带来帮助 一举两得的近义词 面面俱到、一箭双雕、一石二鸟、两全其美、事半功倍、一箭双鵰. 一举两得造句 1、在寒假,爸爸给我报了奥数班学习数学。这样我在寒假里,既能玩一玩,也能学到数学,真是一举两得,我同意这个方案。 2、走进低碳生活,我们既能节约能源,又能健康快乐,真是一举两得。 3、如果我们每天都整理房间,那将是一举两得的事。因为那样既把房间打扮得漂亮了,也会渐渐地养成习惯。 4、妈妈教我合理的运用时间学玩分明,这样我就可以一举两得,事半功倍了。 5、萌萌真可谓是一举两得呀,她只做了一件好事,却获得了两个荣誉。 6、我今天因考试得了一百分,在校老师表扬我,在家父母奖励我,一举两得的滋味真爽。 7、过新年了,可以一举两得,既能吃上美味的年夜饭,也可以收到许多红包,新年真快乐 8、我们大都市应该多多植树,这样不仅可以美化环境,还可以

服务于大众,净化出新鲜的空气,真是一举两得 9、爸爸让我边扫地边收拾一下自己的东西,不一会地扫干净了,我的东西也收拾好了,真是一举两得。 10、快考试的时候,妈妈让我做什么事情时都要想着英语单词,()可到头来,我的英语却考得很差,这种一举两得的方法真是用不得呀。 11、妈妈叫我把棒棒糖给妹妹,我把棒棒糖给妹妹时,故意说是我和妈妈一起给的,这样我和妈妈的书就不会被妹妹撕了。 12、这次期末试虽然我没考好,但是可以从中找到不足还能鞭策骄傲的自己,没考好也能一举两得嘛 13、刘老师让我们在寒假写下学期的生字帖,这样既可练习写字,又可预习下个学期的字词,真是一举两得呀 14、我学着玩有趣的魔方,既锻炼了手的灵活性,也让脑子转得更快,一举两得。 15、有些人不爱护环境,我特别想发明一种机器,它既能让不爱护环境的人主动捡垃圾,又能让这些人主动爱护花草树木多好啊 16、我每个星期都持之以恒地写晚报造句和作文,不但越写越好了,而且学会了合理安排时间,一举两得呀 17、我早上为了不浪费时间,一边听英语一边写作业,真是一举两得。 18、这次期末我考试得了全班第二名的好成绩,老师表扬了我,同时也证明了我地努力没有白费,真是一举两得。

三年级下册语文试题14单元名词造句 人教新课标

一、请用下列词语写句子,句型:拟人句。 1)燕子: 2)尾巴: 3)细叶: 4)荷花: 5)蜻蜓: 6)翠鸟: 7)列车: 8)风雪: 9)友情: 10)山谷: 11)森林: 12)洪水: 13)大雁: 14)讲台: 15)暑假: 16)墨水: 17)钻石: 18)水罐: 19)钢铁: 20)蔬菜: 二、辨字组词。(20分) 1.寓()悔()峰()度()

2.偶()诲()锋()席() 3.常()逢()装()泰() 4.裳()蓬()妆()奏() 5.喘()拔()延()罐() 6.瑞()拨()廷()灌() 三、在括号里填上适当的量词。(4分) 1.一( )微风一( )客人一( )火车一( )房子 2.一( )楔子一( )善心一( )公路一( )村庄 3.一( )金光一( )清泉 4.死记硬背是一种传统的教学方式,在我国有悠久的历史。但随着素质教育的 开展,死记硬背被作为一种僵化的、阻碍学生能力发展的教学方式,渐渐为人们所摒弃;而另一方面,老师们又为提高学生的语文素养煞费苦心。其实,只要应用得当,“死记硬背”与提高学生素质并不矛盾。相反,它恰是提高学生语文水平的重要前提和基础。一( )图画一( )赞歌 四、将下列句子补充完整 1.穿花蛱蝶深深见, 2.亲身下河知深浅, 3.一寸光阴一寸金, 4.要练说,得练看。看与说是统一的,看不准就难以说得好。练看,就是训 练幼儿的观察能力,扩大幼儿的认知范围,让幼儿在观察事物、观察生活、观察自然的活动中,积累词汇、理解词义、发展语言。在运用观察法组织活动时,我着眼观察于观察对象的选择,着力于观察过程的指导,着重于幼儿观察能力和语言表达能力的提高。少壮不努力,

用胆大妄为造句

用胆大妄为造句 1、这几个歹徒真是胆大妄为,竟敢在光天化日之下抢劫他人财物。用胆大妄为造句 2、这个盗贼胆大妄为到了极点,竟然偷到了公安部门的头上。 3、这家伙胆大妄为,早就该进监狱了。 4、他在公司里之所以这么胆大妄为,是依仗他跟经理的连襟关系。 5、这艘胆大妄为、自行其是的军艇,对周围情况的变化根本不在意。 6、想不到他监守自盗,偷天换日,真是胆大妄为。 7、他敢和我分庭抗礼,真是胆大妄为。 8、一个真正的企业家,不能只靠胆大妄为东奔西撞,也不可能是在学院的课堂里说教出来的。他必须在市场经济的大潮中摸爬滚打,在风雨的锤炼中长大。 9、一个具有创造精神的人不是空想家,也不是胆大妄为的冒失鬼. 10、为什么戴维做出这些胆大妄为的事情? 11、刚才,我不在的时候,说不定你又干了些别的胆大妄为的淘气事. 12、出了个胆大妄为的乖张孩子当然令整个家族坐立不安. 13、犯人因绝望而越来越胆大妄为。 14、这个主竟使裘很得意.她就喜欢做胆大妄为的事情. 15、这台机器很快变得更加胆大妄为起来 16、过于自信的;胆大妄为的,自大的。 17、这些人都是胆大妄为的家伙,虽然我们可以打他们个措手不及,

但我们要谨慎小心,否则他们就可能使我们受到某种损害。 18、这伙歹徒如此胆大妄为、厚颜无耻,让我很是震惊。 19、他是个胆大妄为的人,从来不怕任何事. 20、像这样的胆大妄为应该受到指责. 21、中国又有谁能够如此胆大妄为地安排这种会见呢? 22、你竟敢擅自开矿,真是胆大妄为. 23、这是一个典型的傻瓜胆大妄为的例子. 24、或者,是不是以一种最愚蠢的胆大妄为的方式开启了人类灭亡的结局? 25、他杰出的姑妈们胆大妄为地评论了一番他的外貌.用胆大妄为造句 26、在日本帝国舰队中,这是自杀性的胆大妄为. 27、这个曾国藩也太胆大妄为恣肆无忌了!咸丰忿忿不平的想。 28、胆大妄为的是你——冷秋月,也不过是姥姥身边的一条狗,主人还没有叫,就先撒欢儿了,拼命去讨好主人,让人恶心? 29、镖师们背后叫她眉眼高低,眉眼高低竟然是胆大妄为的陈大人的女儿,真的是让人不可思议。 30、试想在那个时代里,人人对意识形态的禁忌避之惟恐不及,闻捷以戴罪之身,居然胆大妄为,要一尝浪漫爱情的禁果,难怪自讨苦吃。 31、令人不解的是,许多大公司的管理者被管理时尚和口号所左右,迷恋于胆大妄为、突飞猛进、富有创意的领导行为,拒绝改变路线,全然不顾已经完全改变了的形势。

“之”的意义和用法

“之”的意义和用法 在古代汉语中,“之”字可作实词,也可作虚词,在不同的语境里有不同的意义和用法,同学们在备考时应引起重视。 一、“之”作实词 1、“之”作动词 这时,“之”的意思是“到……去”“往”“到”“至”。 例1 奚以之九万里而南为?(《逍遥游》) 例2项伯乃夜驰之沛公军。(《鸿门宴》) 例3辍耕之垄上。(《陈涉世家》) 2、“之”作代词 “之”作动词的情况在古代汉语中不多见,作代词相对常见一些。“之”作代词时,一般用作宾语,代人、事、处所,所代的对象大多出现在上下文中。作代词时,“之”又分为指示代词和人称代词。 (1)“之”作人称代词,可译为“他(们)”“她(们)”“它(们)”,用作宾语。 例4君为我呼入,吾得兄事之。(《鸿门宴》)

例5生乎吾前,其闻道也固先乎吾,吾从而师之。(《师说》) 例6虽有(通“又”)槁暴,不复挺者,鞣使之然也。(《劝学》) 例7是何异于刺人而杀之,曰“非我也,兵也”?(《寡人之于国也》) 例8盖将自其变者而观之,则天地曾不能以一瞬。(《赤壁赋》) 从上面所举的几个例子我们可以发现,“之”作人称代词时,一般用在动词之后。 “之”所代的人称在上下文中不一定有具体交代,要根据语言环境进行判断,灵活翻译,如“人非生而知之者”中的“之”,其上下文都找不到先行词,它是泛指“知”的对象,即知识、道理等,可略而不译,也可根据上下文之意译为“知识”或“道理”。 (2)“之”作指示代词,译为“此”“这”“那里”“这样”“这个”等,此时,“之”可作定语,也可作宾语。 例9以故其后名之曰“褒禅”。(《游褒禅山记》) 例10由山以上五六里,有穴窈然,入之甚寒。(《游褒禅山记》) 例11均之二策,宁许以负秦曲。(《廉颇蔺相如列传》) 例12之二虫又何知!(《逍遥游》)

一举两得的典故

一举两得的典故 一举两得:做一件事情得到两方面的好处。为大家整理了一举两得的典故,希望大家喜欢。 一举两得的典故: 东汉初,被任命为建威大将军的耿弇率兵奔赴齐地,前去围剿以张步为首的割据势力。当时张步据守剧地,派他的兄弟张蓝率精兵二万据守西安县,另有诸郡太守合万余人据守临淄。西安县与晒淄相距四十里。耿奔便进军画中邑,正好处于西安、临淄两城之间。安营扎寨之后,耿弇分析了一下两城的据守情形。西安城小却十分坚固,而且张蓝兵士精良;临淄名声很大,但守兵松散,易于攻下。于是,他来了个声东击西,扬言要攻打西安,实则发兵前往临淄。这时,有的将帅提出异议,认为随该速去攻打西安,耿弇坚持道:“不然。西安已听说我们要去攻打,正日夜守备,临淄在毫无防备的情况下遇到我们,必会慌乱不堪,这就很容易拿下。攻下临淄,西安就变成了一座孤城,张蓝与张步又相距较远,他一定会吓得弃城而逃。这样我们就是一举而两得了。”果然,耿弇之师只用半天功夫便占据了临淄,张蓝听说,惊慌失措,率领部将弃西安逃往剧城,耿弇攻一城而得到了两座城池。 【拼音】yījǔliǎng dé 【释义】举:做事。做一件事得到两方面的好处。

【语法】联合式;作谓语、定语;含褒义【近义词】一石二鸟一箭双雕两全其美一箭双雕事半功倍 一举两得造句: 1. 利用废物,既可变废为宝,又可减少空气污染,是一举两得的好事。 2. 你所干的那件事可是一举两得。 3. 封山造林,既能生产木材,又能保持水土,一举两得。 4. 我早上为了不浪费时间,一边听英语一边写作业,真是一举两得。 5. 萌萌真可谓是一举两得呀,她只做了一件好事,却获得了两个荣誉。 6. 妈妈教我合理的运用时间学玩分明,这样我就可以一举两得,事半功倍了。 7. 如果我们每天都整理房间,那将是一举两得的事。因为那样既把房间打扮得漂亮了,也会渐渐地养成习惯。 8. 过新年了,可以一举两得,既能吃上美味的年夜饭,也可以收到许多红包,新年真快乐! 9. 他做好事一举两得,伙伴们知道了,都纷纷跑去做好事。 10. 既不伤脾胃,又能吓这小妮子一跳,一举两得!嗯,对,就是这样!阳不韦刚想到这里,前边的公孙绾却忽然停了下来。 11. 此计一举两得,既揭开了魔窟之谜,一偿楼主夙愿,又将血衣楼的敌人通统不费吹灰之力一网打尽。

相关主题
文本预览
相关文档 最新文档