随着Web2.0思想观念及其技术的广泛应用,基于社交媒介的UGC对经济、政治、社会、军事、外交及其他方面都产生了重要的影响,基于UGC的情感倾向性自动识别研究具有重要的理论意义和实际应用价值。本研究针对当前情感倾向性自动识别中亟待解决的挑战性问题,研究并提出了基于特征选择和倾向分析联合优化的UGC情感倾向性自动识别方法。本研究将所提出的方法应用于实际的中文和英文、两类和五类UGC情感倾向性自动识别中。基于两种不同语言的语料库:中文豆瓣网电影评论文本和英文IMDB电影评论文本,本研究构建了基于特征选择和倾向分析联合优化的情感倾向性自动识别模型,并对该模型进行了检验。一系列的实验结果表明,本研究所提出并构建的基于特征选择和倾向分析联合优化的情感倾向性自动识别模型能够提高UGC情感倾向性自动识别的效果,从而说明了该模型对于中文和英文自动情感分析的有效性。

The global innovation landscape is changing open innovation.In open innovative team,electronic argumentation based on Web is becoming one of the most primary and important innovative activities.It is very important to mine,identify and visualize the argumentation topic of mass argumentation information in the process of open team innovation.These benefit not only mastering the whole team progress and the latest development rapidly and accurately,but also recommending appropriate knowledge and field experts to team members according to their argumentation topics.In order to strike on the main problems of classic text topic mining,an automatic topic identification method is studied and proposed in this paper.A semantic computing method based on argumentation ontology and argumentation tree structure is proposed and built during documentation modeling phase.AntSA algorithm is employed for short-text clustering during topic mining phase.The contribution ratio of the nouns in each category node to argumentation topics is proposed and computed to identify topics of each category.Consequently,the visualization system of topic identification for open team innovation argumentation is designed anddevelopment based on the proposed method.The visualization system is able to automatically identify and intuitively exhibit semantic relationships and structural relationships between various argumentation topics.Experiment study of topic identification method for open team innovation argumentation is conducted as well.

People increasingly tend to publish their reviews and comments on social media platform.Consequently,websites containing reviews are becoming targets of opinion spam.Integrating work from psychology and deception behavior in social media,we propose 11deception linguistic cues which are divided into three categories.Guiding by design science theory,we then develop an online review spam detection system and compare different collections of deception features.The result of experiment demonstrates the precision of identification of fake review is nearly 80%on our fake review data set that deliberately written to sound authentic.We discuss the validity of deception linguistic cues in fake review detection.

Knowledge innovation is the key to improve core competence of enterprises.The global innovation trends have been shrugging off the physical constraints in the context of globalization.In the open environment,cross-organization and cross-region open innovation teams become the main functional bodies for knowledge innovation.Social tagging enables knowledge transition and cooperation between team members according to their own wills and encourages knowledge innovation for open teams.However,due to the freedom of tagging,how to build tag correlation so as to effectively support knowledge management and knowledge innovation in open environment is the critical issue that needs to be solved.In this paper,dynamic tag correlation web for open team knowledge innovation is proposed and studied.This method is applied to case study afterwards.By comparison of results from experimental group and control group,differentiation between strong tag correlation and weak tag correlation can be distinguished.In addition,timeliness with popular tags is explored the tag correlation web.It can be seen that the dynamic tag correlation web for open team knowledge innovation can represent the tag correlation dynamically and support knowledge innovation in open environment.

Speech act classification is significant for question understanding and intention estimation in the domain of dialogue systems,machine translation and Q&A systems.The speech act classification is an active area of research both in information retrieval and natural language processing.We summarize the theory of speech act,speech act taxonomies,classification features,feature selection methods,classification algorithms and evaluation methods,and discuss the future research and development trend of speech act classification.

Decision making is essential to management.A lot of argumentation information is produced in group decision-making.There are multiple relationships between the argumentation information and decision solution.In this paper,a automatic identification method of multiple argumentation information relationship in group decision-making is researched and put forward.A automatic identification model of the argumentation information relationship in group decision-making is built.Furthermore,the method is applied to actual group decision process.The results of the application show that the method realizes the automatic identification of the strongly supportive,supportive,neutral,strongly against,and against relationship between decision solution and argumentation information effectively.It can help group members to organize the large amount of argumentation information effectively and increase the efficiency of information organizing in group decision-making.

Knowledge network performance in organization is analyzed in the paper.Firstly the systematic identity between knowledge network and finite network is studied,and a knowledge network performance model in organization is proposed.Based on the model,the method to measure the knowledge contribution of an individual or a group is discussed in the paper.This will help to evaluate the performance of an individual or a group in a reasonable way and improve the effectiveness of knowledge management.In the end,a case study is given and the model is verified.

In the 1990s,creation was regarded as the main resource of competitive advantage for firms replacing efficiency and quality.A lot of information is produced in team creation.In this paper,the automatic identification method of the information relationship in team creation is researched and put forward.Furthermore,the method is applied to setting up the automatic identification model of information relationship in actual team creation process.In the model,the primary word frequency method is improved from the following three aspects: classifying feature words into six levels according to their tone,introducing the new feature of clause count and the new feature of the last comment's attitude.The results of application show that the method realizes the automatic identification of the relationship between solution and comment effectively.It can help team members to organize the large amount of team comments effectively and increase the efficiency of information organizing.

The group members should be endowed with weights to compute the forces of conflict sides to analysis consensus states of group involving in argumentation supported by hall for workshop of metasynthetic engineering. This paper discusses the concept of the group members' weights orienting to argumentation procedure and how it differs from the weights orienting to preference integration. The algorithm is also discussed.

ANN theory and method is applied. The evaluation model of argumentation information record modes is developed. The model is applied in the study on Web - based Group Argumentation Support System (GASS). The results show that the method can evaluate record modes quantitatively based on concrete argumentation activity. Thereby, the method can provide more effective support to group decision.

知识创新是许多组织的重要竞争因素,现代组织越来越多地利用团队解决复杂的或者创新性的任务.然而,对于团队知识创新,目前普遍缺乏有力的创新支持工具.针对这样的问题,该文根据知识创新团队的工作过程,设计了团队创新支持系统的体系结构.最后,对系统的使用流程和开发实现进行了描述.

针对我国制造行业机械产品研究与开发团队的创新过程目前普遍缺乏有力的创新支持手段的现状,根据知识创新团队的工作过程,提出了团队创新支持系统的需求,设计了团队创新支持系统的体系结构,并分析了系统对机械产品研究与开发团队的支持过程.

Knowledge creation is an important competitive factor in many organizations. In order to compete in complex and turbulent markets, these firms must create, share, and manage knowledge that will give them a competitive advantage. One method of knowledge creation that has been used by organizations for many years is focus group. One drawback of the team approach is that information decentralization often results in organizational knowledge fragmentation and a loss of organizational learning. Moreover, because the creation information and the member proposing the information were not tracked in time, the team creation cannot be reasonably and effectively promoted. This study seeks to find the effective and efficient tracking approaches to promote team knowledge creation. The method of Backpropagation neural netowork (BPNN) is used to set up the model reflecting the match relation between knowledge creation activities tracking approaches.

Knowledge creation is an important competitive factor in many organizations. One way of knowledge creation of organizations for many years is mainly teamwork. However, teams are usually short of the effectively supportive system and method to team knowledge creation. In this paper, according to the work process of knowledge creation team, the Integration Frame of Web-based Team Creation Support System is put forward. The demand of the system is analyzed. The construction of Team Creation Support System is designed. And the key technologies are illustrated. Finally, the prototype is validated.