[Purpose/Significance] Scientific knowledge is the main source of technological innovation, and absorbing scientific knowledge in technological innovation is an important way of innovation.Scientific knowledge applies in the process of technological innovation act as a bridge between science and technology, the characteristics of translation knowledge play an important role in improving the efficiency of scientific knowledge adoption.Previous studies proves that the source-related characteristics of knowledge, as a convenient method to evaluate impact, influences the user's attention distribution and information screening process.However, whether this impact exists and how it manifests in the process of cross-border adoption, remains to be explored. [Methods/Process] This study explored the adoption of scientific knowledge by technology from the perspective of patent-paper citation and constructed a dataset of patents adapting scientific knowledge in different fields from 1800 to 2018.Based on the extraction and measurement of the source-related characteristics of scientific knowledge, the study analyzed its distribution and trends and further extracted a sub-dataset for the regression models to explore the influence of source-related characteristics on adoption intensity, adoption lag, and adoption scope. [Results/Conclusion] The results show that: ①S-T integrational innovation becomes a trend, and innovation activities becomes more refined.It accelerates the S-T adaption but reduces the scope and intensity.②The source-related characteristics of scientific knowledge significantly affect the adoption effect and the better knowledge producer and disseminators often bring better performance.③The S-T adoption performs differently in fields.There are indeed more and faster adoptions in strong-practical fields, but such knowledge is hard to produce strong technical influence.

[Purpose/Significance] Terminology serves as important media in scientific communication, helping scientists reach a consensus on the same concepts.In-depth study of the movement mechanism of terminology in scientific communication is helpful to reveal the stage characteristics of scientific development, and it's of great significance to guide scientific development rationally. [Method/Process] Through theoretical research and descriptive statistical analysis of practical disciplines, the key elements of the movement of terminology were summarized and their movement model was designed.Control variables were simulated by Vensim to observe the movement mechanism of terminology. [Result/Conclusion] The development of terminology usually has three stages: reserve terms, standardization of terms and promotion of scientific exchange environment.In addition, it is greatly influenced by the structure of terms and the interaction between terms.Therefore, we should pay attention to the monitoring of terminology movement, provide appropriate development guidance for terminology in time, accelerate the process of terminology integration and the formation of scientific communication context, promote the steady progress of scientific communication, and promote the development of science.

[Purpose/Significance] The study starts from the dynamic migration of users' interests in online health communities, temporal features, which are integrated into social relationships and personal preferences, so as to improve the personalized recommendation algorithm for online health communities and further improve the accuracy of users' access to health information. [Methods/Process] Firstly, starting from users' social relationships, the paper constructed a users' influence relationships network fusing temporal features; Secondly, based on users' personal preferences, it built a matching matrix of user topic posts which integrated temporal features; Finally, an interest rating matrix of user topic post was fused, from which the TOP-N recommendation list for each user was extracted. [Results/Conclusion] The constructed personalized recommendation algorithm with fused temporal features can improve the accuracy of recommendations and enhance the performance of the personalized recommendation algorithm for online health communities.

[Purpose/Significance] Utilizing online physician review(OPR)information is important for online medical services.Exploring how the review features affect users' adoption will help to promote the optimization of online medical platforms and the development of services. [Methods/Process] Based on the ELM model and trust transfer theory, a 2*2*2 inter-group situational decision-making experiment design was used.This study collected 539 valid questionnaires, and used Smart PLS to carry out direct effect, mediation and moderation effect analysis. [Results/Conclusion] Factual(vs evaluative)reviews and reviewer identifiers with real surnames(vs nicknames)can improve perceived review objectivity and reviewer credibility, and further enhance users' perceived usefulness of reviews and their willingness to adopt reviews subsequently.Trust to the website can be transferred through general reviewer group to specific reviewer.Gender and disease severity characteristics negatively and positively moderated the influence of review features on user cognition and behavioral reaction, respectively.Several measures proposed in the study can help to optimize the design of online medical platform and promote users' adoption and utilization of OPR information.

[Purpose/Meaning] In the era of mobile Internet, it has become common for users to search through multiple APPs, but the intention pattern that drives users' APP transfer behavior is still unclear.The purpose of this paper is to investigate the cross-APP intent transfer patterns in mobile search in real-life contexts. [Method/Process] The study collected users' daily cross-APP search experiences through interview method, obtained users' cross-APP intention chains through open coding, mined frequent sequences using Markov chains, summarized cross-APP intention transfer patterns, and then analyzed the application laws of the patterns in combination with task types.The paper condensed seven intention patterns of supplementary exploration, contrast decision, contrast merit, step-by-step advance, supplementary preparation, sharing and communication, and frustrated exploration; and found that users preferred different intention patterns in different task types—the supplementary exploration pattern was most preferred in simple tasks, while the contrast decision pattern was most preferred in hierarchical tasks. [Results/Conclusion] This study helps to clarify the cross-APP behavior pattern of mobile search from the perspective of users' subjective cognition, and provides a basis for optimizing APP system service support and improving users' mobile search experience.

[Purpose/Significance] This study explores how to identify depression symptoms from online medical consultation, which profoundly enhances health big-data utilization and adds data value. [Method/Process] Using the patient consultation text on"HaoDaiFu", an online medical platform, the study employed unsupervised machine learning, phrase recognition, and deep-learning modeling to identify depression symptoms.The algorithm was evaluated on test data and tested further in two tasks: depression symptom analysis and depression patient identification. [Results/Conclusion] The model's accuracy is 73.85% in the symptom extraction task, indicating it performs well.In the task of analyzing depressed patients' features, the conclusion is consistent with the clinical psychological tests, and the accuracy in the task of recognizing depresses patients can reach 78.81%, which verifies the effectiveness of the algorithm.Among the three semantic models describing symptoms, Sentence-BERT performs the best, confirming that strengthening the semantics of symptom phrases and using unsupervised machine learning can extract disease symptoms swiftly and effectively improve the efficiency of processing large-scale textual information.

[Purpose/Significance] Understanding policy intentions is a critical goal of quantitative analysis of political texts.Using degree lexicon constructed based on the feature of semantic intensity of Chinese political texts contributes to identify policy intentions implied in the political texts efficiently.However, the existing degree lexicon suffers from a small number of words, a single weight design, so it is necessary to carry out optimization and application research of degree lexicon. [Method/Process] Following the previous methods about the construction of the degree lexicon, the paper used different types of political texts to expanding the lexicon words; then, used the expert survey method to construct the graded degree lexicon; finally, combined the graded degree lexicon to analyze the main content of the China's Science and Technology and Innovation(STI)plans in the past 20 years. [Result/Conclusion] The optimised graded degree lexicon has reliability and validity.The political text analysis combined with the degree lexicon is helpful for in-depth exploration and analysis of complex and comprehensive STI plan, and reflects many advantages.It can also choose appropriate analysis granularity according to actual needs, and help carry out problem-oriented policy research.

[Purpose/Significance] As the core engine for the deepening development of the current digital economy, data security has become a major issue related to national security and economic and social development.The security governance of data can promote the true and complete record of data, safe and real-time open sharing, and efficient transaction application.Government data is the core resource for the construction of digital government and the opening of government data.Currently, there are many security problems and demands in the process of using and sharing government data.As a new information technology, blockchain can provide a new path for government data security governance with its decentralized, tamper proof, traceable and timestamp, cryptography, intelligent contract and other functional characteristics. [Method/Process] In view of the demand and technical fit of government data security governance, based on the blockchain cross chain mechanism, a government data security governance system based on the blockchain cross chain mechanism was built from the perspective of system elements, system processes, etc., and based on six core scenarios, the application practice was carried out in the field of transportation government data security governance. [Result/Conclusion] The system constructed in this paper can provide theoretical and practical value for improving the security governance of government data in the context of the overall national security concept strategy, promote the value of government data in the digital economy era, make government data live, use, move, and improve the digital government governance capability.

[Purpose/Significance] Algorithm plays an important role in the development of economy and society in the era of digital intelligence, and algorithm governance gradually enters the public eye.The realization of algorithm governance goals must rely on a combination of various policies and regulations to match, so the analysis of the current situation of algorithm governance policies and regulations in China will help promote the healthy, orderly and prosperous development of algorithm. [Method/Process] In this paper, with the help of the Nvivo12 qualitative analysis software, the study coded and analyzed the contents of 12 algorithmic governance policies and regulations as a sample, while extracting and outlining the framework of algorithmic governance policies and regulations in China. [Results/Conclusion] At this stage, there are three problems of algorithm governance in China: "the lack of general legal guidance, uneven distribution of topics, non-detailed industry standards and norms, no local special policy".Based on this, the paper proposes to start from improving the"strength-depth-precision-breadth"of algorithm governance, and gradually aggregate the fragmented algorithm governance policies and regulations into a system of the algorithm governance policy and regulatory system with Chinese characteristics.

[Purpose/Significance] To predict user behavior in online social network emergencies, so as to explore user behavior rules in public opinion dissemination, the study effectively controls and guides public opinion information, especially to grade users who may do different behaviors, and achieve differentiated guidance and response. [Method/Process] First of all, the study collected the data of users who engaged in interactive behavior of two different kinds of emergency events in Sina Weibo.Then, on the basis of theoretical research and statistical analysis, the study took the effective attributes of users who click like, comment and forward as feature indicators.Finally, the study used Random Forest algorithm and Back Propagation Neural Network algorithm to predict user interaction behavior. [Results/Conclusion] The prediction effect of neural network is better, which indicates that the prediction model proposed in this paper has certain feasibility.At the same time, this paper also makes a regular summary of the attributes of each interaction behavior of users in emergencies.

[Purpose/Significance] The essence of infodemic is the process of competitive propagation between true and false information.Clarifying the mechanism of competitive propagation between true and false information in infodemic is of great significance to the formulation of infodemic intervention strategies. [Method/Process] FT-SIR model was constructed based on epidemic model and user creative behavior characteristics.It solved the equilibrium point of the dynamical model and conducted stability analysis.Using NetLogo to simulate the competitive propagation process between true and false information, and verify the effectiveness of the model against the typical case data of Weibo. [Results/Conclusion] The experiment shows that under the influence of user's creative behavior, information will change in the process of propagation and promote the competitive propagation between true and false information.The"information cocoons"effect and the existence of opinion leader enable users to have different role characteristics, thus having significant effects on the competitive propagation between true and false information.It provides specific intervention strategies and suggestions for the government to take measures to reduce the harm of infodemic.

[Purpose/Significance] The new research universities' international openness and influence index is an important topic to guide the construction of high-level research universities with international influence and the establishment of a scientific, comprehensive, and international diversified evaluation mechanism. [Method/Process] According to the background and important characteristics of the construction of new research universities, from the dimensions of international openness and international influence, the index set three first-level indicators of Global Leadership of Scientific Research(GLSR), International Openness of Education(IOE), and Global Leadership of Society(GLS).There were 8 second-level indicators including Influence of Scientific Research(ISR), Influence of Platform(IOP), Social Leadership Force(SIF), Influence of Teaching(IOT), Influence of Faculty(IOF), Influence of Student(IOS), International Communication and Services(ICS), and International Exchange and Cooperation(IEC).21 third-level indicators such as CNCI, percentage of highly cited papers, h index, number of international teaching and research platforms, etc.In terms of the weight determination method, the subjective weighting method ANP was combined with the objective weighting method CRITIC to construct a new type of research university international openness and influence index.Combined with the public data of 36"Double First-Class"A-level universities for empirical analysis, Chinese universities' international openness and influence rankings were obtained. [Result/Conclusion] The research results show that the score distribution of the first-level indicators of 36 universities presents an"inverse J-shaped", indicating that top universities have strong driving forces but a small number.Most research universities have similar internationalization processes and paths.In the end, this study proposes corresponding construction strategies from global leadership of scientific research, international openness of education, and global leadership of society.

[Purpose/Significance] CiteSpace is a visualization-based analysis tool for identifying and displaying new trends and developments in science in the scientific literature.Co-citation analysis of CSSCI and WOS database literature using CiteSpace has been a popular research method in the academic community.Previous studies have shown that when using CiteSpace to analyze CNKI database literature, only very basic analysis can be performed, and co-citation analysis cannot be performed as in the case of WOS database and CSSCI database.To address this limitation, this study proposes a method to achieve co-citation analysis of literature based on CNKI databases by populating reference data. [Method/Process] The method crawled the references of"Wise Information Technology of med"literature in CNKI database through a self-coded crawler program, and wrote the crawled references into the converted CNKI literature through a self-coded python program, thus realizing the co-citation analysis, author co-citation analysis and journal co-citation analysis of CNKI database literature. [Result/Conclusion] Through the empirical analysis, it is clear that the co-citation method proposed in this paper is effective and can provide a new idea of co-citation analysis for the researchers, which has some practical significance.

[Purpose/Significance] Analyzing the interdisciplinary nature of social science datasets helps to understand their diffusion patterns across different disciplines and promote open sharing among them. [Method/Process] Used CHARLS and CGSS datasets as examples, this paper first measured the disciplinary diversity and balance of the two datasets.Then, constructed a cross-disciplinary collaboration network for CHARLS and CGSS datasets, and used the Louvain algorithm to cluster the network and detect different research communities.Next, used BERTopic to model the topics of the papers using the datasets.Finally, constructed cross-disciplinary collaboration networks at different stages to reveal the evolutionary characteristics of cross-disciplinary collaboration for CHARLS and CGSS datasets. [Results/Conclusion] The cross-disciplinary diversity and balance of CHARLS and CGSS datasets have continued to grow, and the disciplines that use CHARLS and CGSS datasets have formed a pattern in which a few disciplines dominate while multiple disciplines participate together.The research topics using the CGSS dataset are relatively more scattered than those using the CHARLS dataset.The number of nodes, edges, and communities in the cross-disciplinary collaboration networks of CHARLS and CGSS datasets has been continuously increasing, while the network density has decreased.Additionally, the dominant disciplines in different stages of cross-disciplinary collaboration have been constantly changing.