345 Construction of Information System Oriented to Identifying the Frontier of Science and Technology in Key Areas Hot! Liu Qiyan, Zeng Wen, Che Yao DOI: 10.3772/j.issn.1000-0135.2020.04.001 Recently, the development trend of science and technology in China has entered the stage of “following, merging, and leading.” To adapt to the new situation of science and technology, we need to develop frontier knowledge of science and technology in key areas, follow up the new developments of science and technology in major foreign countries in an all-round way, and perceive and judge the future development trend of science and technology. Additionally, we need to play a better role in ensuring the development of information systems. Based on the analysis of the current research situation all around the world, this study proposes and expounds the framework and methods for building an information system to identify the frontier of science and technology in key areas, and introduces the information practice in relevant research fields. The results show that the intelligence system and method of using multi-dimensional data to realize intelligence perception for identifying the science and technology frontier in key fields are operable. 2020 Vol. 39 (4): 345-356 [Abstract] ( 591 ) HTML (92 KB)  PDF (4942 KB)  ( 1069 )

357 Evaluation of Academic Papers Impact Based on Scientific Communication Path: A Case Study of Chinese International Academic Papers in Social Sciences Hot! Guo Fengjiao, Zhao Rongying, Sun Shaomin DOI: 10.3772/j.issn.1000-0135.2020.04.002 Based on the scientific communication path of academic papers, the academic papers impact is divided into native impact, offline impact, Web 1.0 impact, and Web 2.0 impact, and the generation processes of various types of impact are discussed. An evaluation model of the academic papers impact is constructed, and two sets of data (including a control group) obtained from the Chinese international academic papers in social sciences are empirically studied. This study has obtained the following research results andfirstly, the score of the academic papers impact demonstrates the highest correlation with relative impact factor and citation frequency. Secondly, the impact score of data group 1 is still lower than that of the overall control group, and the trend line of the academic papers impact score in both groups shows a right-leaning trend. Thirdly, the impact score of the 50 research institutes with the highest total score of academic papers decreases exponentially, and universities are the backbone of international social science research in China. Lastly, among the 79 subjects covered by the two data groups, the average score of data group 1 with 24 subjects is higher than that of the control group, and the subjects covered by academic papers are classified according to their relative influence. 2020 Vol. 39 (4): 357-366 [Abstract] ( 299 ) HTML (141 KB)  PDF (1234 KB)  ( 831 )

367 Application of Network Representation Learning in the Prediction of Scholar Academic Cooperation Hot! Lin Yuan, Wang Kaiqiao, Liu Haifeng, Xu Kan, Ding Kun, Sun Xiaoling DOI: 10.3772/j.issn.1000-0135.2020.04.003 In the context of Big Data, scientific cooperation has become an important means to improve the level of scientific research and output. A research focus in recent years includes accurately identifying the cooperation objects that are suitable for scholars, institutions, and fields among the vast numbers of these entities. This study constructs a co-occurrence network of author-author, institution-institution, author-institution, author-keyword, and institution-keyword through the recorded data of scientific literature in the field of science of science. The network representation method is used to learn the context information of authors, institutions, and keywords in a network, and the information entity is represented as a low-dimensional dense vector of the same space. Finally, the mining of the cooperation object is achieved based on the similarity calculation of representation vector. The network representation learning method can realize a variety of heterogeneous information fusion, quantitatively calculate the correlation strength between each information entity, capture the relationship between scholars-scholars, scholars-institutions, and scholars-keywords in the research network, accurately explore potential collaborators, and partner institutions and keywords for scholars. 2020 Vol. 39 (4): 367-373 [Abstract] ( 279 ) HTML (71 KB)  PDF (2280 KB)  ( 924 )

374 Query-oriented Opinion Summarization Model Using Debatepedia as Datasource Hot! Yu Chuanming, Zheng Zhiliang, Zhu Xingyu, An Lu DOI: 10.3772/j.issn.1000-0135.2020.04.004 This study systematically studies query-oriented opinion summarization, aiming to construct a query-oriented opinion summarization framework to explore the impact of different text summarization methods on the opinion summarization. Considering sentiment orientation and the similarity between the sentence and query, we extract the sentences from the original documents and employ neural networks and word embeddings to achieve abstractive summarization. The query-oriented opinion summarization framework is then constructed upon this. We crawl the topics and arguments to build the experimental opinion summarization dataset from the Debatepedia websites and apply the proposed method to the dataset to validate its effect. The experimental results show that on this dataset, the summaries generated by the extractive method are of higher quality; the highest average ROUGE score, deep semantic similarity score, and emotional score are 6.58%, 1.79%, and 11.52% higher than the generative method, and 8.33%, 2.80%, and 13.86% higher than the combined method, respectively. Furthermore, the evaluation indicator deep sentence similarity and the sentiment score proposed in this study can better evaluate the effects of the query-oriented opinion summarization model. The research results are of great significance for improving the effects of query-oriented opinion summarization and promoting the application of the opinion summarization model in the field of information science. 2020 Vol. 39 (4): 374-386 [Abstract] ( 249 ) HTML (191 KB)  PDF (1580 KB)  ( 539 )

387 Learning Concept Hierarchies from Chinese Academic Literature for Domain Ontology Construction Hot! Tang Lin, Guo Chonghui, Chen Jingfeng, Sun Leilei DOI: 10.3772/j.issn.1000-0135.2020.04.005 Constructing domain ontology from academic literature has great significance in promoting discipline development. Taking Chinese academic literature as a data source, this study proposed a semi-automatic method for extracting concept hierarchy. First, a fine-grained universal research framework for constructing hierarchical relations of the domain ontology was proposed. Then, a novel concept representation fusion method was developed, considering concepts semantic features based on deep learning and concept frequency in time series. Combined with an affinity propagation (AP) clustering algorithm, Prim s algorithm, and data from a Web search engine, the ontology concept hierarchy extraction algorithm was proposed via rule-based reasoning (RROCHE). Concept hierarchy relations are learned semi-automatically. The algorithm was then applied to the academic literature on Chinese word segmentation. Numerical experiments examined the feasibility and effectiveness of the proposed methods. The proposed method can also be applied effectively and widely to other domains. 2020 Vol. 39 (4): 387-398 [Abstract] ( 322 ) HTML (159 KB)  PDF (3549 KB)  ( 870 )

399 Impact of Source Selection on Health Information Credibility Judgment: A Heuristic Information Processing Experiment among Digital Natives Hot! Song Shijie, Zhao Yuxiang, Song Xiaokang, Zhu Qinghua DOI: 10.3772/j.issn.1000-0135.2020.04.006 The Internet has become one of the most important channels to seek consumer health information. The morphing of information production and distribution in the 2.0 Web era has sharply increased the complexity of the online health information environment and made consumers credibility judgments more difficult. This study is therefore based on a heuristic processing perspective. We propose a conceptual model addressing the relationships between health heuristic processing and credibility judgments. In the employed experimental method, several pieces of stimuli are manipulated by combining controlled text and heuristic cues. The “digital natives,” namely, adolescents are recruited from a local high school. The results suggest that heuristic cues have a general impact on participants. Especially, institutional heuristic cues can significantly improve adolescents credibility evaluations compared with commercial and social media related cues. The difference in credibility judgments between commercial and social media cues are not statistically significant. The study may yield theoretical implications to help adolescence health information seeking and pattern evaluation and generate practical implications to guide health information design for public health purposes. 2020 Vol. 39 (4): 399-408 [Abstract] ( 198 ) HTML (136 KB)  PDF (803 KB)  ( 1407 )

409 Automatic Extraction of Chinese Terminology Based on BERT Embedding and BiLSTM-CRF Model Hot! Wu Jun, Cheng Yao, Hao Han, Ailiyaer·Aizezi, Liu Feixue, Su Yipo DOI: 10.3772/j.issn.1000-0135.2020.04.007 High quality professional term recognition and its extraction play an important role in the fields of domain information retrieval and knowledge graph building. To improve the precision and recall rate of terminology recognition, we propose a Chinese terminology recognition and extraction approach that does not rely on specific domain knowledge or artificial features. Using the latest developments in representation learning, this study introduces BERT embedding as a character-based pre-trained model and incorporates it with a bi-directional long short-term memory (BiLSTM) and a conditional random field (CRF) to extract deep learning terminologies from 1278 annotated datasets collected from domain e-books. The experimental results show that the proposed model reaches 92.96% in F-score and outperforms other competing algorithms, such as left and right entropy, mutual information, a word2vec based similar terminology recognition algorithm, and a BiLSTM-CRF model. The best practices and related procedures for the implementation of the proposed model are also provided to guide its users in its further improvement. 2020 Vol. 39 (4): 409-418 [Abstract] ( 261 ) HTML (106 KB)  PDF (2255 KB)  ( 1733 )

419 Analysis of Users Passive Use Behavior Patterns on Social Network Platforms Based on Time Characteristics Hot! Lu Xinyuan, Huang Mengmei, Lu Quan, Wang Xuelin DOI: 10.3772/j.issn.1000-0135.2020.04.008 This study considers the Zhihu platform and uses the Python software to obtain the data related to users information behaviors, such as publishing articles, answering questions, asking questions, and participating in live activities. The time characteristics, frequency, and time interval of users behaviors are then analyzed, and the time characteristics and rules of users negative behaviors on social network platforms are examined. The results show that some users may reduce their use behavior, intermittently drop out, or interrupt during their use of social network platforms, thereby presenting different degrees of negative behavior characteristics. For identifying users passive use behavior, obtaining an insight into the underlying causes and formulating targeted measures is necessary. This analysis can guide the users to continuously use the social network platforms. 2020 Vol. 39 (4): 419-426 [Abstract] ( 471 ) HTML (81 KB)  PDF (2115 KB)  ( 795 )

427 The Quantitative Algorithm of Sentiment Divergence Based on Web User Reviews Hot! Xu Jian, Wu Siyang DOI: 10.3772/j.issn.1000-0135.2020.04.009 The main aim of this paper is to provide a new method and perspective to analyze the sentiment of web user reviews from the standpoint of sentiment divergence. Drawing on the five existing formulas for calculating the degrees of difference and dispersion, integrating the emotional elements and applying them to the scene for calculating sentiment divergence, five algorithms for measuring sentiment divergence based on emotional value difference, standard deviation, coefficient of variation, information entropy, and probability of emotional distribution are obtained. An algorithm for measuring sentiment divergence based on the frequency of positive and negative emotional values is proposed, which takes advantage of the fact that emotional values can be positive or negative. It is based on the sentiment analysis of text. This paper puts forward six quantitative algorithms of sentiment divergence to quantify the sentiment divergence of user reviews and analyzes the quantified results of sentiment divergence. The results show that the model can achieve the quantification of the sentiment divergence of web user reviews. However, there are differences in applicability and discrimination of different quantitative algorithms of sentiment divergence. 2020 Vol. 39 (4): 427-435 [Abstract] ( 320 ) HTML (157 KB)  PDF (2465 KB)  ( 770 )

436 Modeling Research of Users Dynamic Interests Based on Ontology and Folksonomy Hot! Li Yuanyuan, Li Xuhui DOI: 10.3772/j.issn.1000-0135.2020.04.010 Web 2.0 application technologies, such as blog, instant messaging, social network, community sharing, and Folksonomy, enable the combination of users, information, and resources to form a closely related web network. In recent times, there have been limited research studies that apply ontology to analyze the dynamic interests of the users of social tags. Therefore, a platform’s requirements related to an accurate understanding of its users’ interests are not fulfilled. Our study discusses the construction method of interest model from this standpoint. The ontology of interest labels for Douban readers is constructed in this study on the basis of restriction rules and relationship definitions of words in the Chinese Classification Thesaurus and Chinese Library Classification. According to the index of reproduction, coverage, and calorific rate, the interest intensity and stability of tags are predicted, and the expression form of interest is determined. The initial interest model is then constructed, and the corresponding renewal process of interest nodes is proposed. The user’s interest model combined with ontology and its update process have improved the expression depth and width of user interests to a certain extent. Thus, this work demonstrates a higher scientific reference significance in the applicability of resource recommendation, retrieval, and other applications. 2020 Vol. 39 (4): 436-449 [Abstract] ( 250 ) HTML (127 KB)  PDF (3965 KB)  ( 656 )

450 Research on Peer Effects of Knowledge Contribution Behavior on Answerers in Zhihu Q&A Community Hot! Chen Xiaohui, Hu Ping, Zhou Yicen DOI: 10.3772/j.issn.1000-0135.2020.04.011 The emergence of Q&A community has transformed the original Q&A mode and expanded the means of knowledge acquisition. As the main contributor of knowledge, answerers are the core elements to promote knowledge transfer in Q&A communities. The existing research on Q&A community users is generally based on users attribute data with much less consideration for the network relationship data. This study considers the Chinese Q&A community “Zhihu” as the research object and crawls large-scale data under the two topics of “Pop Music” and “English Learning”. By constructing answerers following network, we can define and quantify the direct and indirect peer relationships. Moreover, we employ the network auto-regressive model to explore the impact of peer effect on answerers knowledge contribution behavior. It is discovered from this evaluation that even though answerers knowledge contribution behavior is positively influenced by direct and indirect peer effects, the indirect peer effect gradually weakens or even disappears with an increase in the answerers network density. Moreover, the clustering coefficient is observed to exhibit a negative impact on answerers knowledge contribution behavior. 2020 Vol. 39 (4): 450-458 [Abstract] ( 293 ) HTML (140 KB)  PDF (833 KB)  ( 1098 )