[Purpose/Significance] This paper aims to improve our understanding of the influence of characteristics of academic virtual community on the effectiveness of knowledge exchange, so as to provide some suggestions for its high-quality development.[Method/Process] Based on the previous relevant research and the characteristics of five types of academic virtual communities, this paper extracted eight characteristics from norms, content, and interaction of academic virtual communities, and then constructed a factor model of the influence of these characteristics on the effectiveness of knowledge exchange.Then the situational questionnaire was used to collect data, and SmartPLS 3 was used to process and analyze 241 questionnaires.[Result/Conclusion] Community participation restrictions, reporting convenience, perceived reliability, perceived variation, push timeliness, perceived interaction intensity and perceived interaction real-time had a significant positive impact on users' perceived learning.Review strictness, reporting convenience, perceived interaction intensity, and perceived interaction real-time had significant positive effects on users' perceived contribution.

[Purpose/Significance] Based on the entropy theory, the entropy flow index system of virtual academic community is constructed.By evaluating the order degree of the virtual academic community system, the status of the community is judged and the running trend is predicted.According to the experimental results, the feasible development strategy of the virtual academic community system is provided to strengthen the exchange of academic information.[Method/Process] After defining the positive and negative entropy concepts of the community and constructing the corresponding index system, Science Net was selected as the research sample, and the Brusselator model was used to determine whether Science Net become a dissipative structure.And then the entropy was used to calculate the order degree of the system in order to discuss its operating state.[Result/Conclusion] Although Science Net currently does not reach the dissipative structure, it is very likely that it will become a dissipative structure in the future.In order to enable it to develop from disorder to order, and then from low-level to high-order, strengthening the input of the emotional participation of members of Science Net and improving the standardization of industry competition in the virtual academic community should be the main focus of the improvement of Science Net.

[Purpose/Significance] The novel coronavirus(COVID-19) is raging around the world, and the scientific community has done a lot of research on the coronavirus.Sorting out the scientific research results of coronavirus and understanding the evolution rule and research progress of the thesis in this field can theoretically guide the future research direction of coronavirus and help scientific research progress and innovation.[Method/Process] This paper used topic modeling and network analysis methods, taking the COVID-19 research data set opened by the White House Office of Science and Technology Policy as an example.According to the outbreak time of 3 typical coronaviruses (SARS in 2003, Middle East Respiratory Syndrome(MERS) in 2012, and Novel Coronavirus(COVID-19) in 2019), they were divided into 4 research phases(1955-2002, 2003-2011, 2012-2018, 2019-2020).The topic content law, topic document distribution law, topic author network law of each stage of coronavirus from the semantic level were interpreted.[Result/Conclusion] The overall theme evolution rule of the coronavirus from a diachronic perspective during 1955-2020 was analyzed, the research direction of the coronavirus was accurately grasped, and global scientific research cooperation in this field was promoted to jointly build "human destiny community".

[Purpose/Significance] In order to reduce the number of labeled samples required for text classification tasks, a semi-supervised classification method that requires only a small number of labeled samples for training is proposed.[Method/Process] This article used a multi-granular text modeling method to obtain different levels of features of the same text, uses the idea of ensemble learning to integrate the base classifier, obtained the supplementary label in the semi-supervised training module, and finally in the predictive classification module Obtain the final classification result.[Results/Conclusions] In two different data sets, the MG-SSC (Multi Granularity Semi-Supervised Classification model) model achieved good classification results when only a small number of labeled samples were used for training.The model proposed in this paper can effectively handle semi-supervised classification tasks and reduce the cost of text classification for users.

[Purpose/Significance] Image resources play an important role in this protection and inheritance of intangible cultural heritage(ICH).The research in the paper is expected to help the preservation, display and service of digital ICH resources.[Method/Process] Aiming at the single image & text sequence problem existing in the organization and presentation of digital ICH image resources, the paper designed a semantic ontology model for them.The editing and instantiation were implemented using ontology development software Protege.The concept model of entities stored in two-dimensional tables was designed according to the ontology model.Moreover, the relational tables were converted into RDF format data with the help of D2RQ platform, and then published.[Results/Conclusion] The realization of complex semantic network about ICH image resources provides a new way to present linked data from the current linear information presentation to nonlinear diversification for further multimedia domain ontology.

[Purpose/Significance] This paper aims to make the construction and evolution of ontology faster and more convenient, it can provide users with more semantically relevant and valuable information, and better meet the needs of users for online tourism information.[Method/Process] Based on the research of tourism ontology construction, the construction method of online tourism website information ontology with user participation is proposed.Then the construction process of online tourism website information ontology by using seven step method was designed.Finally, using the data of the horse beehive online travel website to verify and analyze the method and effect of ontology construction.[Result/Conclusion] The experimental results showed that the information ontology of online tourism website can be constructed by using the information released by users in the website, and the constructed ontology can provide users with more comprehensive and more semantic information to a certain extent, optimize the retrieval effect of users.And the feasibility and advantages of the ontology construction method proposed in this paper are verified.

[Purpose/Significance] Nowadays, large enterprises are faced with the problems of low data quality and lack of data governance tools.The world is transforming from information technology to data processing technology, and data has become the core asset of enterprise development.Due to the unstructured text of enterprise data, computers cannot understand natural language directly.Therefore, enterprise data governance is particularly important.[Method/Process] In this paper, enterprise data assets are taken as the research object, and semantic ontology model is constructed by optimizing semantic representation of entities related to enterprise data with the help of semantic web technologies.Linked data technology is used to realize the visualization among enterprise data, process and other attributes.On this basis, the system update strategy, data transmission capture strategy and data query strategy are further proposed.[Result/Conclusion] Based on the enterprise data governance system constructed in this paper, the global visualization of enterprise data can be realized.By giving different rights and responsibilities to the members of the organization, the data sharing can be maximized, so as to improve the enterprise data governance level and enhance the quality of enterprise data.

[Purpose/Significance] The process of excavation, management, and research of cultural relics produced a wealth of spatio-temporal data.With the help of the characteristics of cultural relics, such as time-space and attributes, a spatio-temporal data model can be constructed to express the entire process of cultural relics changes.Grasp the changing dynamics and laws of cultural relics and promote the management and protection of cultural relics.[Methods/Process] Based on the analysis of temporal and spatial characteristics and temporal and spatial changes of cultural relics, an event-object-based temporal and spatial data model was proposed, and bronze ware was selected as the analysis object to establish its temporal and spatial expression model.[Results/Conclusions] The model can express the changes in the temporal and spatial information of bronzes.The temporal and spatial database which was established can query the temporal and spatial information of bronzes and extract some event information that affected the changes.Combining with temporal and spatial visualization tools, it is beneficial to better Management and protection of cultural relics.At the same time, it also provided a certain foundation for cultural relic knowledge discovery and knowledge service, and had certain theoretical and practical significance.

[Purpose/Significance] As a new generation of Internet service mode, Mobile Visual Search (MVS) has low utilization rate, poor user stickiness, and poor user experience.This paper explores the influencing factors of MVS user experience, in order to provide theoretical reference and practical guidance for MVS managers, designers and operators.[Method/Process] Based on the Grounded Theory, the author collected data through in-depth interviews and established a model of factors influencing the MVS user experience through open coding, spindle coding, selective coding and theoretical saturation test.[Result/Conclusion] The influencing factors of MVS user experience mainly included Accuracy, Authenticity, Completeness, Convenience, Timeliness, Accessibility, Flexibility, Tangibility, and Empathy, which had theoretical and practical guiding significance for identifying the key factors influencing MVS user experience.

[Purpose/Significance] This article used a socialized Q&A platform as an example to explore the characteristics and distribution of public psychological health information needs during the COVID-19 pandemic, and provided reference and help for government and social media operators to understand the current status better and improve their information service.[Method/Process] Through five steps of data collection, data processing, K-means clustering, LDA topic model and demand identification, all 14168 question data on the psychological health forum of the Q&A platform "Zhihu" during the epidemic period were mined.Through comparative analysis, we explored the thematic characteristics of public psychological health information needs in the stage of pre-pandemic, pandemic and post-pandemic.We constructed a framework of public psychological health information needs during the COVID-19 pandemic, and discovered the distribution and topic features of different social groups' psychological health information needs during that period.[Result/Conclusion] During the COVID-19 pandemic, the need for psychological health information in a family context is more likely to arise;The public mostly needs psychological health information related to depression.The public's awareness of psychological health problems prevention and receiving professional psychological assistance need to be improved;The expression accuracy of psychological health information needs under the online environment needs to be enhanced.

[Purpose/Significance] The online music users' willingness to pay is the key to the protection of music copyright and sustainable development of the industry.Based on the exploratory research, this paper aims to explore the factors affecting online music users' willingness to pay.[Method/Process] The model was built based on the theory of customer perceived value, UTAUT and the theory of consumer devotion.And then build a model to analyze the sample which size was 569.[Result/Conclusion] Perceived value, interpersonal influence, consumer devotion and individual paying consciousness directly affected users' willingness to pay.Utilitarian value and hedonic value positively affected perceived value.Financial cost and codification cost had no significant influence on perceived value.Online word-of-mouth negatively moderates the influence of perceived value on willingness to pay.The online music platform can make operational policies based on these conclusions.

[Purpose/Significance] As the phenomenon of knowledge anxiety is more and more manifested in the consumer behavior of platform users, how to make better use of this consumer psychology to carry out knowledge marketing has become the focus of platform enterprises.[Method/Process] Based on the theory on perceived value, user stickness, consumer purchase decision, this paper made a deep analysis on the influence of knowledge anxiety to users' willingness to pay for knowledge from the perspective of user's individual emotion management, and considered the user stickiness as the regulating variable, which constructed the theoretical model on knowledge anxiety-perceived value-willingness to knowledge payment.[Result/Conclusion] Through making hierarchical regression analysis on sample data, the results showed that the knowledge quality anxiety, knowledge overload anxiety and knowledge deficiency anxiety all had significant positive effects on willingness to knowledge payment, the perceived value played a completely mediating role between the willingness to pay for knowledge and knowledge quality anxiety, knowledge overload anxiety, had a partial mediating effect between it and knowledge deficiency anxiety, user stickiness could positively regulate the relationship between knowledge quality anxiety, knowledge overload anxiety, knowledge deficiency anxiety and knowledge payment willingness.Based on the above research conclusion, platform enterprises in the development of the user, could guide users to focus on emotion management to carry out the knowledge marketing.Specially, they could create the knowledge contents matching to user knowledge quality anxiety caused by anxiety marketing, at the same time, strengthen improving platform functions for the user perceived value increasing, and establish effective actions to develop the user cognitive surplus to push knowledge reconstruction efforts, enhance the user experience.

[Purpose/Significance] There are three kinds of weights in the evaluation index data of science and technology.One is reality weight, the second is natural weight, the third is data distribution weight.To analyze their relationship is very important.[Method/Process] The paper taking the JCR2016 Mathematical Journal as example, compared and analyzed principal component analysis and TOPSIS as the representatives of nonlinear evaluation methods and propose standardization method of approximate dynamic maximum mean to eliminate the natural weights.[Result/Conclusion] It showed that the natural weights of evaluation indexes affected the results of linear evaluation;The weights of evaluation index data mainly affected the nonlinear evaluation results;Different nonlinear evaluation methods also affected the actual weights of nonlinear evaluation;the weight characteristics of index data affected the choice of evaluation methods;The linear evaluation can eliminate the natural weights, and there is no problem of the weight of the data distribution;The weight of nonlinear evaluation data distribution can not be eliminated, so the linear evaluation is more reasonable in the view of evaluation method.

[Purpose/Significance] The subjectivity in the process from the setting of evaluation indicators to the distribution of weights has always affected the evaluation of universities and also affected the ranking results.[Method/Process] This paper designed a new measurement model of indicator weights based on the panel data model and analytic hierarchy process model, and implements the two improved evaluation model including TOPSIS model and gray correlation model based on our proposed new measurement model of indicator weights.Then this paper utilized two improved TOPSIS model and gray correlation model to evaluate and compare the differences among the ranking results from the 2017 Best Chinese Universities Ranking and from two traditional evaluation model including the TOPSIS model and TOPSIS model based on the entropy weights.[Result/Conclusion] As a result, it is found that the evaluation model of universities based on our proposed new measurement model of indicator weights was of the better effect of evaluation, and keep the higher coincidence degree with the 2017 Best Chinese Universities Ranking results.However, the ranking results of universities were easily influenced by the differences of indicator weights and evaluation methods that can be reflected by the high average value of the difference between the minimum and maximum values of each university obtained by different kinds of ranking methods.

[Purpose/Significance] This paper aims to explore the research status of artificial intelligence ethics in library and information science, and promote the healthy development of information management technology in the synchronous development of ethical research.[Method/Process] 39 academic papers discussing the ethics of artificial intelligence were collected under the discipline category of library and information in the core collection of Web of Science in the past five years.The content analysis method was used to analyze from the perspectives of problem levels, application fields and solutions.[Result/Conclusion] The findings indicated that the papers published in library and information science journals studied similar issues as the whole field of artificial ethics.The research perspective covered both humanism and technicism.The research content covered data, algorithm and artificial intelligence system at different levels, involving medical and health, political and military fields.In addition, the research discussed the solutions from the social and technical perspectives.The future research can further focus on the information life cycle, strengthen the diversity of the level and field of ethical issues, and promote the research of artificial intelligence ethical norms and guidelines.

[Purpose/Significance] Academic social networks are important platforms for managing academic resources, maintaining scholarship relationships, and accelerating the efficiency of knowledge transfer.By combing the research status and future research direction of knowledge organization, it is helpful to optimize the knowledge organization mode and promote the knowledge exchange and innovation of scholars.[Methods/Process] Firstly, it sorted out the theoretical development of knowledge organization in academic social networks, and then summarized the knowledge organization methods of academic social networks into four types:scholar relationship, tag clustering, topic association and recommendation system.[Results/Conclusions] The future research aims to reveal the relationship between user relationship and knowledge demand based on the exploration of user interaction behavior and content characteristics, to realize multi-dimensional knowledge association organization based on user interaction relationship, and to expand knowledge service mode based on user interaction demand.