[Purpose/Significance] Since the arrival of the era of big data, research on higher education evaluation at home and abroad has developed rapidly. This article aims to explore the development status of international higher education evaluation in order to provide enlightenment to my country's higher education evaluation. [Method/Process] This paper used the Web of Science core data set as the data source and used SATI and SPSS tools to conduct text mining, cluster analysis and visual analysis of the current situation of international higher education evaluation. [Result/Conclusion] It is found that the research hotspots focus on the evaluation method problems with feedback as the core. The evaluation index problem is the center, and the evaluation quality problem is the reliability and validity as the core. In the future, domestic higher education evaluation should make full use of the characteristics of big data, establish a scientific evaluation system, and improve the quality of evaluation.

[Purpose/Significance] Research on the overall and dissemination characteristics of hot papers is of great significance to the promotion of academic influence and knowledge sharing. [Method/Process] The article was based on the Altmetric TOP list literature for frequency statistics and analysis, and discussed the overall characteristics of the literature from the subject, journal, acquisition method, the first author's country and the distribution of the first two authors' colleges and universities;K-Means clustering was used to divide the different indicators Level, and used the association rules of the Apriori algorithm to explore the characteristics of document dissemination in different situations. [Result/Conclusion] It can be found that the literature in the field of medicine and health accounts for a relatively large proportion, the social influence of the humanities is low, and the single AAS and the number of citations in the field of information and computer science are relatively high. The linear forecasts of the proportions of open and free access both show an increasing trend, which is in line with the development trend of knowledge sharing. At the same time, under different citation intervals, AAS value intervals, disciplines, countries, and acquisition methods, there are big differences in the channels of document dissemination.

[Purpose/Significance] The paper reveales the hot topics in domestic research in the field of big data. [Method/Process] The paper collected papers on big data from 2008 to 2020 included in CNKI, extracted research topics through the LDA topic model, and identified hot topics. The paper selected representative topics, performed LDA topic clustering on them again, and used LDAvis to visualize the topics. [Result/Conclusion] The LDA model can more accurately extract the research topics of the literature in the big data field, which helps researchers understand the development status of the field, grasp the future research direction, and explore emerging topics.

[Purpose/Significance] The purpose of this paper is to mine academic social network data, recommend personalized paper search results for researchers, and provide new ideas for enhancing and optimizing the functions of the knowledge service platform. [Method/Process] Based on the academic social network, this paper constructed a following relationship network among academic users, and proposes recommendation indicators by combining the degree of prestige with the shortest distance. The research was carried out from the perspective of a certain academic user. After the user's user-generated content was used for the first search content expansion, a list of candidate researchers was calculated and used for the second search content expansion to more accurately match the researcher's search intent and interest. [Result/Conclusion] This paper chosen ResearchGate as the source of the academic social network data, crawled user relationships and user-generated content data related to the topic of "Artificial Intelligence". The academic field involved in the final recommendation result is more detailed, which is more consistent with the academic interest expressed by the certain user.

[Purpose/Significance] In response to the call of Data Security Law to build a data security system, this paper explores the construction of China's data security system under a holistic approach to national security, in order to provide possible reference for related work. [Method/Process] Through literature analysis and comparative analysis, this study combed through the development and research status of data security institution at home and abroad, combined the idea of system theory, a holistic approach to national security and the Data Security Law to propose a data security institution construction model containing basic elements as well as basic sub-systems, expounded the components of data security institution, analyzed the seven basic sub-systems in the light of domestic and foreign experience, and finally made suggestions from the perspective of intelligence professionals. [Result/Conclusion] Data security institution under a holistic approach to national security is a complex system including target value, rule system, adjustment object and implementation guarantee. Its starting point and foothold are the holistic approach to national security. Its construction principles are people-oriented, collaborative development, incentive compatibility, harmony and keeping pace with the times. Its sub-systems are the data classification protection system, the data classification protection system and the data classification system Data security risk management system, data security emergency response system, data security review system, data export control system, anti discrimination system and system evaluation supervision and feedback system. It can be used for reference from three aspects:the development goal of Information Science in the new era, the characteristics of information science and the talents of information science.

[Purpose/Significance] Aiming at the singularity of the existing weak signal identification methods for enterprises, this paper proposes an automatic weak signal identification method based on the LDA-BERT fusion model. [Method/Process] This method first classifies the text data set through the LDA topic model, and secondly, constructs a two-layer filter function for topics and terms to extract early warning signals from the results of topic classification, and uses three major metrics:close centrality, topic weight, and topic autocorrelation to filter the subject, and weak signals are extracted based on the normalized frequency and probability of the terms in the subject. Finally, use the BERT deep learning model to semantically expand weak signals. [Result/Conclusion] This article uses the corporate social media news data set to verify the built system model, effectively detecting related weak signals, and digging out the evolution characteristics of weak signals that gradually increase over time. This model not only solves the problems of the existing weak signal research with more manual participation and low interpretability of the detection results, but also use the fusion model to make up the lack of the LDA bag-of-words model, making it more effective in identifying weak signals. While providing reference information for corporate crisis early warning and strategic decision management, it also provides new ideas and new methods for weak signal identification research.

[Purpose/Significance] The RippleNet recommendation model is optimized by correcting noise and changing preference propagation rules to improve recommendation accuracy. [Methods/Process] The idea of "the same-level participating users and evaluated items have the same score" was borrowed to detect and correct the noise in the scoring data set;formulated preference propagation rules for preferential selection of knowledge graph nodes connected by high word frequency semantic relations and constructed user preference data sets. [Results/Conclusion] The optimized model was 5% higher than the average precision of different K values under the Top-K scenario of the original RippleNet model;the average AUC value of the test set under the CTR hit rate prediction scenario was increased from 90. 7% of the original model to 92. 3%. Modifying the noise in the data set and improving the preference propagation rules could improve the performance of the RippleNet recommendation model.

[Purpose/Significance] According to the objective characteristics of books as the basis for book acquisition, compared with based on the needs of readers as the basis for book acquisition, it has a more objective basis. [Methods/Process] On the basis of the classification and division of whether the collection of books has been circulated, the description information of the collection of books characterized by titles and publishers was formed through text classification technology to form a book feature classification method, and the statistical results of whether the collection has been circulated, it can be transformed into probabilistic events that can be classified and learned using techniques such as Naive Bayesian classification of artificial intelligence. [Results/Conclusion] The results of the operation of the model show that the Naive Bayesian classification technology based on text segmentation can classify the trend of book circulation according to the book representation, which can make the book acquisition get rid of the interference of artificial subjective factors, and provide objective and reliable quantitative evidence.

[Purpose/Meaning] Finance currently occupies an important position in the field of social development, and the personal financial information security crisis cannot be ignored. This article, from the perspective of policy tools, conducts a quantitative analysis of my country's policy text on personal financial information protection, and explores current relevant issues. The characteristics of the selection and use of policy tools, and suggestions for optimization are put forward to help my country's personal financial information protection work be carried out. [Methods/Process] Based on the 28 collected policy texts of personal financial information protection with national effectiveness, it designed an X-Y two-dimensional analysis framework, based on the code results of the policy provisions, supplemented by NVivo qualitative data analysis software, from personal financial information protection Analyze the two dimensions of basic policy tools and policy elements. [Results/Conclusions] The research finds that my country's personal financial information protection policy texts mostly use less effective policy tools at the regulatory level;the structure of policy tools is unbalanced, and there is a phenomenon of overuse of individual tools;the operability of the personal financial information protection policy Need to be strengthened and other issues, and put forward suggestions for optimizing my country's financial privacy protection policy.

[Purpose/Significance] COVID-19 has a huge impact on the subject service of university libraries, libraries and users are unable to adapt to this change for a time. Under this situation, how to make full use of the subject service context to adapt to this situation in a short time and achieve good performance is the focus of academic and industry attention. [Method/Process] Therefore, through reviewing the existing practice of subject service in University Library and combing the existing achievements of subject service under the epidemic situation, the paper abstracted and summarized the difficulties of subject service under the epidemic situation according to the characteristics of the discipline service of University Library under the epidemic situation. On this basis, from the perspective of the coupling of the two subjects of "University Library-subject service-teachers and students", the subject service mechanism of university library was explored by building the subject service scene for users through the library, and by virtue of the subject service context owned by the library, and based on the scenario contextual adaptation theory, The mechanism model of subject service was formed, which was "subject service scene-subject service expectation-subject service context", and the path to realize the subject service in university library is designed. [Result/Conclusion] Taking the dissertation guidance as an example, this paper designs its subject service scheme. The results show that university library should start from the subject service expectation of teachers and students, and take the subject service experience of teachers and students as the end result, so as to create a new contact point of subject service in University Library and solve the pain point of subject service in university library, To provide reference for subject service of University Library under epidemic situation and other emergencies.

[Purpose/Significance] This study aims to find the influencing factors of social media users' personal information security and privacy protection behavior, by proposing hypotheses, constructing a theoretical model, and validating the model. [Method/Process] We conducted a questionnaire survey(n=360)on a representative sample of social media sites users in China. The research model was validated by using the Structural Equation Model(SEM). [Results/Conclusion] Privacy concern, response efficacy, self-efficacy, response cost, and perceived threats are found to have significant direct impacts on users' protection intentions of security and privacy, whereas response cost has a negative effect. The above five variables affect user's protection behavior indirectly through the variable "protection intention".

[Purpose/Significance] With the increasing influence of the Internet in the society, weibo is the most important platform of Internet information communication. When an emergency occurs, public opinions on Weibo have a huge impact on the Internet and offline. In this paper, the law of public opinion communication on weibo of emergencies will be studied, the public opinion communication on Weibo will be segmented, and the index system influencing the public opinion communication on Weibo will be built. Then, the influence direction and degree of factor indicators on the public opinion communication on Weibo will be analyzed to provide the direction for the government to guide and control public opinion. [Method/Process] Logistic model was used to divide the communication process of public opinion on Weibo of emergencies into stages. Then, based on SEM(Structural Equation Model), an influence mechanism model of public opinion dissemination on Weibo of emergencies was built to analyze the direction and degree of influence of factors and indicators at each stage on public opinion dissemination on Weibo. [Result/Conclusion] The law of emergency public opinion communication on Weibo conforms to the information life cycle theory. Based on the differentiation of logistic model, this paper divides the five stages of public opinion communication on Weibo. The "communication user", "blog content" and "text characteristics" in the index system play different roles in different stages of public opinion communication on Weibo.

[Purpose/Significance] From the perspective of privacy risk prevention control, this paper investigates the privacy protection practices of government open data in Brazil, in order to provide a reference for the development of policy of open data in China. [Method/Process] Using the research methods of literature review and case analysis, this paper took Brazil as an example, accessed to the first-handed materials from literature and discussed the experience of government open data privacy risk prevention in Brazil, analyzed and summarized the judgment criteria of personal privacy in government open data, data protection impact assessment(DPIA)and risk assessment, data processing standards which respond to privacy risk, and organizations and positions for privacy risk response. [Results/Conclusion] This paper puts forward to revise the open data policy, improves the privacy risk assessment, and formulates detailed data processing specifications.

[Purpose/Significance] In order to solve the network security problem of social network information exchange, [Method/Process] This paper analyzed the components and operation process of social network information exchange based on the collaborative theory with the guidance of cyberspace governance, and pointed out that social network information exchange presents the characteristics of global information, value driven, unbalanced interaction, sparse aggregation of information groups, collaborative information behavior, etc at the micro level, [Result/Conclusion] Aiming at the problems of excessive democracy, information preference and unbalanced information interaction in social network information exchange, this paper puts forward the network governance objectives of orderly management, production of high-quality and effective information and coordination of relationship conflicts, and designs a collaborative innovation mechanism of social network information exchange integrating multi-agent collaborative management mechanism, full interactive dynamic mechanism and adaptive operation mechanism.

[Purpose/Significance] Virtual Teams(VT)is a dynamic, flexible and effective organizational form, which attracts more and more scholars at domestic and abroad. [Method/Process] To facilitate the understanding of VT research, this study described the distribution characteristics, analyzed the evolutionary topics, and revealed possible future convergence trends of VT research by using bibliometrics and visualization tools. [Result/Conclusion] The results shows that the three research topics of communication, collaboration, and technology have always been the research hotspots in this field, while knowledge, trust, performance, and creativity have shown different research enthusiasm at different development stages. Moreover, related technologies, project management, telemedicine, creativity, information and knowledge are the potential research frontiers with a convergence trend.

[Purpose/Significance] The research of blockchain has been paid more and more attention. The analysis of blockchain research topics and their evolution is helpful for Chinese scholars to grasp its development trend and do research facing the international frontier. [Method/Process] Focusing on the topic and evolution of blockchain research, this paper analyzed from two aspects:hot topics and topic evolution by using the LDA topic model, with the web of science core collection SCI-EXPANDED and SSCI papers from 2008 to 2020 as data sources. [Result/Conclusion] The hot topics in the field of blockchain included "health care", "data privacy protection", "energy transaction and consensus algorithm" and "Internet of things security". The two topics of "business smart contract" and "cryptocurrency" have obvious evolutionary characteristics, which were shown in the rising popularity of the two topics after the publication of influential works. The research of blockchain extended and penetrated from its technology itself to many fields including medical treatment and energy. At present, the proportion of research on blockchain 3. 0, which is characterized by integrating and solving problems in other fields, has exceeded that of traditional blockchain 1. 0 and blockchain 2. 0 represented by smart contracts.

[Purpose/Significance] Argument mining is the most promising research topic in the areas of artificial intelligence and text mining. It may be seen as the latest research direction of opinion mining and it aims to recognize, extract and analyze the argument components and the argument structure within the textual information. The technologies and methods of argument mining could help us to measure the claims and evidences which argumentative text carries, and to evaluate of the credibility of claims. Such works are also of fundamental importance to realize the fine-grained analysis and deep understanding of the text. [Methods/Process] In this paper, 121 papers published were collected on argument mining workshop from 2014 to 2020, and were sorted out in terms of the relevant theories, methods and application scenarios. [Results/Conclusions] We find that the study of argumentation mining is still in the initial stage, and the overall researches are undermined by the shortage of holism, the absence of annotation standard, and the insufficiency of study depth. Therefore, more attention should be paid to the application research and the quality of the corpus construction in the future. Besides, it is necessarg to realize the leap from the understanding of argumentative information to the automatic generation of argumentation.