997 Research on Scientific and Technological Interaction Patterns Based on Topic Relevance Analysis Hot! Liu Ziqiang, Xu Haiyun, Luo Rui, Dong Kun, Zhu Lijun DOI: 10.3772/j.issn.1000-0135.2019.10.001 Analyzing the internal mechanism of science and technology interaction at the micro level and revealing the modes of science and technology interaction quantitatively, automatically, and visually are of great significance to remedying the deficiency in current research on the internal relationship between science and technology and to revealing the development law and evolution characteristics of the collaborative innovation between science and technology. First, by constructing a co-occurrence matrix of multi-relationship fusion, research topics in papers and patents are identified on the basis of a community detection algorithm. Then, the structured data on science and technology topic association are constructed by synthesizing co-words, authors, and citation association degrees. Finally, the visualization map of science and technology topic evolution is drawn using the visualization method of topic evolution to assist in the analysis of science and technology interaction patterns. Empirical research was carried out with papers and patent data in the field of genetic engineering vaccines. The results showed that the main scientific and technological interaction modes in the field of genetic engineering vaccine were the S pattern, the T pattern, the S-T pattern, and the T-S pattern. Among them, the S pattern and the T pattern increased synergistically with the passage of time; the T-S pattern and the S-T pattern increased cross-wise with the passage of time. 2019 Vol. 38 (10): 997-1011 [Abstract] ( 277 ) HTML (142 KB)  PDF (4981 KB)  ( 766 )

1012 A Method Considering Local and Global Information for Constructing Stereoscopic and Accurate Portraits of Scientific Researchers Hot! Zhang Yanan, Huang Jingli, Wang Gang DOI: 10.3772/j.issn.1000-0135.2019.10.002 By constructing scientific research behavior portraits, researchers can easily use various research services for efficiency. Existing research often abstracts the portrait problem into a multi-classification problem without considering the full use of information and the problem of updating portraits. Accordingly, this study proposes a scientific research behavioral portrait method for researchers considering local and global information and introduces a deep learning method. Deep learning can extract highly abstract features for sequence modeling, extracting partial portraits, and combining global information to build stereoscopic and accurate portraits. Finally, based on the actual scientific research behavior data, the method proposed in this study is verified, and its effectiveness is proven. 2019 Vol. 38 (10): 1012-1021 [Abstract] ( 274 ) HTML (103 KB)  PDF (2622 KB)  ( 639 )

1022 Scholar Evaluation Research Based on an Improved h-index Hot! Xiong Huixiang, Ye Jiaxin, Ding Ling, Zeng Ting DOI: 10.3772/j.issn.1000-0135.2019.10.003 The h-index is one of the common scientific indicators that are used to measure the academic ability of scholars. In view of the shortcomings of the h-index, which ignores often-cited papers, this paper analyzes the frequency of citation of papers with different h-indexes and calculates the average number of citations of papers with a certain h-index. On this basis, a new index is proposed: the hc index, which consists of an hc1 index and an hc2 index. The hc1 index is used to measure whether the scholar’s academic ability is higher or lower than the average value of the current h index; the hc2 index is used to measure the author’s academic activity in recent years. Scholars in the fields of digital libraries, digital archives, and digital museums have been selected for empirical research. The research shows that the hc index can better measure the author’s academic ability and that it can solve the problem of the h index ignoring often-cited papers. It can also be used to find scholars who have been active in recent years, and it can evaluate scholars according to the characteristics of different fields. 2019 Vol. 38 (10): 1022-1029 [Abstract] ( 231 ) HTML (108 KB)  PDF (1019 KB)  ( 474 )

1030 Research on the Influence Evaluation Method of Interdisciplinary Journals Hot! Zhang Huiling, Xu Haiyun, Yue Zenghui, Liu Chunjiang DOI: 10.3772/j.issn.1000-0135.2019.10.004 Although there is much discussion on the influence evaluation methods of interdisciplinary journals, there is no perfect evaluation index and evaluation system for operability. First, this paper systematically combs the research progress of domestic and foreign journalsinfluence evaluation methods and analyzes the research status of the field. Then, based on the characteristics of interdisciplinary journals, through the standardization of an evaluation index based on citation distribution and the normalization of citations between disciplines, this paper proposes an evaluation index of interdisciplinary journals that considers both citation skewness and interdisciplinary citation diversity—the CS[citation normalization (Cn) and subject normalization (Sn)] index based on the journal articles—and the model and operational steps of the interdisciplinary journal impact evaluation are constructed. Finally, considering the differences in citations in the field, this paper selects “comprehensive,” “physical chemistry,” and “biology” as the empirical fields and demonstrates the utility of CS series indicators and the applicability of indicators. The empirical results show that the CS index has applicability in three subject areas. It can integrate the CS correction value of the traditional evaluation index, improve the accuracy of the evaluation of interdisciplinary journals, and, finally, summarize research gaps and possible future improvement directions. 2019 Vol. 38 (10): 1030-1040 [Abstract] ( 239 ) HTML (155 KB)  PDF (1391 KB)  ( 676 )

1041 Research on Topic Discovery and Evolution Based on Time Series Clustering Hot! Li Hailin and Wu Xianli DOI: 10.3772/j.issn.1000-0135.2019.10.005 In view of the uniqueness of the existing methods of topic discovery and evolutionary analysis in literature, this paper proposes a method of topic discovery and evolutionary analysis based on time series clustering. The co-occurrence matrix of high-frequency keywords in document datasets is found by co-word analysis. The co-occurrence matrix is transformed into a similarity matrix by the Ochiia coefficient calculation method, and then the topic of the document is found by using the nearest neighbor propagation clustering algorithm. At the same time, the research heat of each topic during a certain period is analyzed and transformed into time series data reflecting the heat of each topic, and the time series clustering method is used to classify and analyze the evolution trend of each topic. The experimental results show that the proposed method can effectively discover the research topics of journals and better analyze the evolution trends of these topics through data processing and mining of the journal literature related to innovation management in CNKI from 2000 to 2018. 2019 Vol. 38 (10): 1041-1050 [Abstract] ( 309 ) HTML (152 KB)  PDF (3138 KB)  ( 981 )

1051 Research on Construction and Implementation of a Corporate Bankruptcy Prediction System Model Hot! Tang Xiaobo, Tan Mingliang, Li Shixuan, Zheng Du DOI: 10.3772/j.issn.1000-0135.2019.10.006 As an important research topic in the field of finance and management science, corporate bankruptcy prediction plays a significant role in financial decision-making. This study proposes a corporate bankruptcy prediction system model from the perspective of knowledge engineering. The system constructs a bankruptcy prediction ontology by acquiring domain knowledge from domain experts and the literature, and it exploits sentiment analysis and knowledge discovery to mine interpretable rules for bankruptcy prediction from quantitative and qualitative information. Consequently, a corporate bankruptcy prediction ontological knowledge base can be constructed in a semi-automatic way. The system utilizes an inference engine to reason over the ontological knowledge base to predict corporate bankruptcy and interpret the prediction results. We validated the constructed system model by using the data of U.S. companies that had been listed as bankrupt in the past decade, and we implemented a corporate bankruptcy prediction prototype system. This system can not only improve the accuracy of predictions, but also support corporate stakeholders in predicting bankruptcy effectively. 2019 Vol. 38 (10): 1051-1065 [Abstract] ( 296 ) HTML (136 KB)  PDF (6236 KB)  ( 974 )

1066 Research on Query Expansion Based on Deep Learning Hot! Yu Chuanming, Cai Lin, Hu Shasha, An Lu DOI: 10.3772/j.issn.1000-0135.2019.10.007 By introducing a deep learning framework into query expansion and combining both local and global query expansion, the query drift problem caused by pseudo relevance feedback in the query expansion is solved. Query phrases and product names posted on eBay in 2017 are selected as experimental data. A deep learning query expansion model (DLQEM) is proposed to achieve more accurate and effective query expansion based on pseudo relevance feedback and is applied to information retrieval. The experimental results show that the precision@10 value of the DLQEM is improved by 3.5% and 3.7% on the basis of pseudo relevance feedback (PRF), which validates the hypothesis proposed in this study (i.e., that the intersection of concept-related extended words and feedback information extended words can effectively control the query drift caused by feedback-related extended words). Deep learning can solve the problem of the difficulty of supervised learning in obtaining good classification results on short-text corpus. Combining deep learning with the traditional query expansion model can solve the two disadvantages of the traditional query expansion model that requires user participation and low retrieval speed, and it controls the query drift. 2019 Vol. 38 (10): 1066-1077 [Abstract] ( 213 ) HTML (155 KB)  PDF (1388 KB)  ( 681 )

1078 A Study on the Measurement Methods of Term Discriminative Capacity for Academic Resources Hot! Wang Hao, Tang Huihui, Zhang Haichao, Zhang Jin, Zhang Zixuan DOI: 10.3772/j.issn.1000-0135.2019.10.008 Improving the quality of indexing terms can effectively improve the retrieval efficiency of the IR system, but the inherent properties of the term are susceptible to the length of the document, making it difficult to fully measure the quality of the term. In this regard, this paper starts from the intrinsic property of the term’s discrimination and proposes the theory of term discriminative capacity (TDC) and three different calculation methods based on the idea of the bag-of-words model. In this paper, 900 records containing 4 entries from three sub-databases of Web of Science were collected as experimental data to realize large-scale calculation of TDC and observe the differences between the three algorithms in practice. Through experimental analysis, the best method for calculating the term discriminative capacity is determined to be TDC-T. Its algorithm is stable in many respects and is not affected by the DF value. Therefore, as a new indicator to measure the quality of the term, it is recorded as TDC. However, the A&HCI database selected in this study has fewer records, which may cause an imbalance in the calculation results of the other two fields. 2019 Vol. 38 (10): 1078-1091 [Abstract] ( 140 ) HTML (136 KB)  PDF (3077 KB)  ( 617 )

1092 An Empirical Study on Mobile Social Media Fatigue User Portraits in the New Media Environment：A Causality Perspective Based on SSO Theory Hot! Zhang Yanfeng, Peng Lihui, Liu Jincheng, Hong Chuang DOI: 10.3772/j.issn.1000-0135.2019.10.009 This paper mainly explores the causal and outcome-based factors contributing to mobile social media fatigue by constructing a mobile social media burnout theory model and analyzing user portraits, providing guidance for companies to understand the development of the trend of mobile social media fatigue. This research was conducted by mining the psychological and behavioral characteristics of different types of social media fatigue among users, extracting the tags of fatigued users in mobile social media based on grounded theory and SSO theory, and taking farmers, students, and teachers as the survey subjects. Through the K-medoids clustering method, this paper makes an empirical analysis of four groups of user portraits with significant differences. According to the characteristics of the user's portrait tag, the types of fatigue user portraits in mobile social media can be divided into four categories, namely the diving neglected type, patient used type, platform transferred type, and the behavior substituted type. Next, this paper specifically analyzes the key characteristics of each type of user portrait to provide a more comprehensive explanation of the tag types of user portraits for mobile social media fatigue. 2019 Vol. 38 (10): 1092-1101 [Abstract] ( 190 ) HTML (106 KB)  PDF (1731 KB)  ( 1384 )

1102 Abstractive Summarization Based on Sequence to Sequence Models: A Review Hot! Shi Lei1, Ruan Xuanmin2, Wei Ruibin1 and Cheng Ying2,3 DOI: 10.3772/j.issn.1000-0135.2019.10.010 ive Summarization Based on Sequence to Sequence Models: A ReviewShi Lei1， Ruan Xuanmin2， Wei Ruibin1 and Cheng Ying2,3(1. School of Management Science and Engineering, Anhui University of Finance and Economics, Bengbu 233030; 2. School of Information Management, Nanjing University, Nanjing 210023; 3. School of Chinese Language and Literature, Shandong Normal University, Jinan 250014)Compared with the early abstractive summarization method, the text summarization method based on Sequence to Sequence models is much closer to the process of human-written summaries, and the quality of the generated summary has also been significantly improved, which has attracted increasing attention from the academic community. This paper reviews the research related to abstractive summarization based on Sequence to Sequence models in recent years. According to the structure of the model, this paper summarizes the research on the model in terms of encoding, decoding, training, and so on, and it compares and discusses these works. On this basis, some technical routes and development directions for future research in this field are put forward. 2019 Vol. 38 (10): 1102-1116 [Abstract] ( 371 ) HTML (290 KB)  PDF (979 KB)  ( 1287 )

1117 An Overseas Review of the Spread of Opinions on Social Networks and Information Distortion Based on Information Cascade Hot! Wei Jianliang, Zhu Qinghua DOI: 10.3772/j.issn.1000-0135.2019.10.011 With the prevalence of false information and propaganda on social networks, more researchers are paying attention to information cascade. Many individual imitation behaviors enlarge the influence of information diffusion in social networks rapidly, but they also trigger great fluctuation and uncertainty, and distorted opinions are becoming increasingly frequent. Therefore, many researchers focus on this topic, from feature and structure discussions of information cascade, to modeling and its optimization based on real data, then to cascade forecasting and maximizing influence. Nevertheless, most of this research employs technique analysis, and seldom is it concerned with the effect, especially the suboptimum effect and opinion distortion caused by cascade. Based on this context, this paper proposed several research topics from the perspectives of the user, information, and structure. 2019 Vol. 38 (10): 1117-1128 [Abstract] ( 266 ) HTML (153 KB)  PDF (772 KB)  ( 1064 )