667 UsersPrivacy Protection in Academic Social Networks： A Case Study of Blogs from Sciencenet.cn Hot! Hu Changping, Qiu Rongrong, and Wang Lili DOI: 10.3772/j.issn.1000-0135.2019.07.001 The user’s setting status regarding privacy permissions can reflect the conditions of the user s privacy protection; therefore, this paper analyzes the privacy preferences of academic social network users and the impact of usersattributes on their privacy protection behaviors by analyzing the privacy permission setting behaviors of users in academic social networks. Taking Sciencenet.cn as an example, this paper investigates the privacy protection of 1032 bloggers who utilized libraries, information, and archives. The research includes three aspects. We obtain the information on privacy permission settings restricting the bloggerspersonal information and analyze the main types of privacy information that users are concerned about. According to the registration date of the bloggers of Sciencenet.cn, we analyze the impact of registration date on usersprivacy protection behaviors. According to the educational level of the bloggers on Sciencenet.cn, we analyze the impact of educational level on usersprivacy protection behaviors. The findings suggest that there are differences between academic social network users and regular social network users regarding privacy protection. The Sciencenet.cn bloggers pay attention to privacy protection. They are especially concerned about the privacy and security of personally identifiable information, and they tend to disclose information related to their academics in order to promote academic communication. Over time, the degree of privacy protection of Sciencenet.cn bloggers has been rising. Recently registered bloggers on Sciencenet.cn pay more attention to the privacy of basic personal information than the established bloggers. The recently registered bloggers tend to value academic communication. Their educational level has no effect on the privacy protection behavior of users in academic social networks. 2019 Vol. 38 (7): 667-674 [Abstract] ( 255 ) HTML (98 KB)  PDF (808 KB)  ( 805 )

675 Research on the Relationship between the Interdisciplinarity of Scholars and Its Impact on Citation in the Humanities and Social Sciences Hot! Zhang Pei, Ruan Xuanmin, Lyu Dongqing, Cheng Ying, and Ke DOI: 10.3772/j.issn.1000-0135.2019.07.002 Based on an analysis of the shortcomings of existing research methods, this paper studies the influence of interdisciplinarity on citation from the author s perspective. Considering that most of the existing research focuses on the natural sciences, this paper selects the humanities and social sciences to construct data sets. In order to control the quality of the thesis, this paper only utilizes source papers from first-class journals on various subjects from the CSSCI database. The paper uses the author's specialization and the Euclidean distance of the voting vector to quantify the interdisciplinarity of individual authors and co-authors, supplemented by interdisciplinary numbers to improve interdisciplinary measurements. The empirical results show that overall, interdisciplinarity is conducive to improving citations in papers; for monographs, the interdisciplinarity of an individual author using both measures has a significant positive impact on the quality of the paper s citations. In terms of the Euclidean distance, there is a significant positive correlation between the two, but interdisciplinary numbers display a curvilinear (inverted U shape) relationship with citation impact. In addition, the study also finds a significant effect stemming from the number of authors, their disciplines, and the quality of papers on the relationship between the interdisciplinarity of scholars and its citation impact. 2019 Vol. 38 (7): 675-687 [Abstract] ( 158 ) HTML (182 KB)  PDF (938 KB)  ( 949 )

688 Direct Measurement of the Degree of Interdisciplinarity Hot! Ma Ruimin, Yan Xiaohui, and Shen DOI: 10.3772/j.issn.1000-0135.2019.07.003 At present, interdisciplinary integration is becoming more pronounced, thus requiring additional collaborative innovation of scholars from different disciplines. Accordingly, the measurement of the degree of interdisciplinarity becomes an important research task. After elaborating the theoretical basis of model construction, a comprehensive model of direct interdisciplinary measurement is constructed from three aspects: direct citation, bibliographic coupling, and co-keywords. Based on Web of Science data, this paper discusses the interdisciplinary relation between information science and library science, as well as six other subjects related to management, computer science, and information systems. In addition, the study conducts empirical research from two aspects: internal comparisons (comprehensive vs. single-indicator models) and external comparisons with current mainstream indicator models. The results prove that the model proposed in this paper has comparative advantages: it is more scientific in principle, easier to manipulate, better aligned with the current situation, stronger discrimination, and the results are easier to interpret. Therefore, it provides an effective method to detect the degree of interdisciplinarity. 2019 Vol. 38 (7): 688-696 [Abstract] ( 253 ) HTML (128 KB)  PDF (908 KB)  ( 733 )

697 On the Quantification and Distribution of Citation Peaks Hot! Li Lingying, Min Chao, and Sun DOI: 10.3772/j.issn.1000-0135.2019.07.004 Citations are important carriers of scientific knowledge, and a citation peak reflects the most influential stage in the process of citation diffusion. In this study, the distribution of citation peaks was analyzed to provide a better understanding of the dynamic diffusion of scientific knowledge. A citation peak identification method was applied to the citation curves of articles in publications of the American Physical Society (APS 2013). Definition and quantification were included in the analysis, and six types of articles have been summarized. In the study, peak distribution was used to distinguish citation patterns based on the number of peaks, peak position, and peak interval. This article demonstrates that the most influential stage can be explained by a simple peak model, which allows us to probe differences in peak distribution quantitatively. Citation peaks of articles showed a high degree of temporal regularity, with most articles having only one peak. The first peak and the highest peak were generally reached within a few years after publication (most within five years, especially within the first or second year). The results also show that the first peak position has a positive and significant correlation with the distribution of the highest peak. We also found that highly cited papers are more likely to reach the first peak in the early phase of publication. 2019 Vol. 38 (7): 697-708 [Abstract] ( 155 ) HTML (158 KB)  PDF (4438 KB)  ( 718 )

709 Evaluation of Authors Influence Based on Principal Component Analysis and a Neural Network Hot! Li Qinmin, and Guo DOI: 10.3772/j.issn.1000-0135.2019.07.005 To better analyze the influence of scientific researchers, a comprehensive evaluation of influence is performed by combining six factors of researchers’ influence with the method of multivariate statistics. First, the initial H index is obtained and improved. A comprehensive index based on WRSR (weighted rank sum ratio) and the principal component from principal component analysis is then established. Finally, a prediction model is obtained by training a neural network. The empirical results show that, compared with other traditional indicators, the method has good distinction, correlation, and comprehensiveness, enabling the precise evaluation of the influence of scientific researchers. 2019 Vol. 38 (7): 709-715 [Abstract] ( 193 ) HTML (179 KB)  PDF (800 KB)  ( 735 )

716 Auto-Identification of Authors Affiliation Based on Class-Center Vectors Hot! He Tao, Wang Guifang, and Ma DOI: 10.3772/j.issn.1000-0135.2019.07.006 When analyzing a large amount of scientific and technical literature, identification of the author s affiliation is always necessary. A key step in this task is matching the author s address to the corresponding institution. Authors from one institution often state their affiliations in various forms in English. This causes string-matching methods to yield unsatisfactory results. In this paper, a machine learning method known as “class-center vectors” has been proposed to solve this problem according to the characteristics of the author s address. Compared with traditional methods, our method does not require matching rules to be written manually. The experimental results of Chinese Academy of Sciences (CAS) author s address data sets illustrate the feasibility of our method. 2019 Vol. 38 (7): 716-721 [Abstract] ( 216 ) HTML (74 KB)  PDF (720 KB)  ( 711 )

722 Research on the Methods of Information Science and Artificial Intelligence Fusion Innovation Hot! Wen Youkui, Wen Hao, and Qiao Xiaodong DOI: 10.3772/j.issn.1000-0135.2019.07.007 The improvement of computer hardware performance and the development of cloud computing technology have increased the speed of scientific literature information retrieval and multi-type data clustering problems. However, the objects of retrieval cannot directly enter the factual knowledge of a document s content; thus, it is difficult to realize intelligent technology. Literature Big Data Knowledge quickly answers questions and recommends service functions. The browsing of scientific and technological literature information in the context of big data continually increases the time spent by scientific and technological personnel to obtain innovative knowledge and the burden that is placed on them. There are two reasons for this. One is that the data model of scientific literature is an unstructured text data structure, and the other is that the database of traditional information retrieval systems does not support unstructured text data structures. These two points have constrained the development of scientific and technological literature on big data results and led to user issues with artificial intelligence and automated answering services. In response to this problem, this paper proposes intelligent mining and knowledge service research based on the achievements of big data innovation in scientific literature. Firstly, it utilizes the idea of ??artificial intelligence to uncover innovative results in scientific and technological literature, then it establishes a semantic knowledge base for innovative results, and it finally establishes a semantic knowledge base intelligent inference engine problem-answer service system. This study explores the research methods for the intelligent and automated development of the browsing model of a big data service for scientific literature. 2019 Vol. 38 (7): 722-730 [Abstract] ( 348 ) HTML (61 KB)  PDF (1584 KB)  ( 1185 )

731 Interoperability between Ontological Word Lists of Persons and Construction of Classification Systems Hot! Junzhi Jia, Xiyan Cui DOI: 10.3772/j.issn.1000-0135.2019.07.008 With the evolving construction of human ontology vocabularies, it is difficult for users to query and use the appropriate ontology vocabulary. Therefore, by reaggregating the attributes of human ontology vocabularies based on natural and social attributes of the characters, a human-centered multidimensional list of character attributes can be constructed to provide users with classified navigation for using the ontology vocabulary. Ontology mapping is the most extensive and effective technique for constructing semantic bridges between ontology vocabularies. In this paper, the string similarity algorithm is used to calculate the similarity of attributes based on attribute name, domain, value domain, and upper attribute. At the same time, the ontology vocabulary attributes with mapping relationships are summarized. Lastly, eleven different types of attribute classification systems are obtained, such as human relationships, locations, contact methods, and events, and the aggregation of multiple ontology vocabularies is realized. 2019 Vol. 38 (7): 731-741 [Abstract] ( 193 ) HTML (91 KB)  PDF (3138 KB)  ( 586 )

742 A Semantic Query Extension Method for Enterprise Information Retrieval Hot! Geng Shuang, Yang Chen, Niu Ben, Yi Wenjie, and Liu DOI: 10.3772/j.issn.1000-0135.2019.07.009 Conventional information retrieval methods usually attain relatively low accuracy in obtaining inner enterprise information retrieval solutions. This is partially because of the limited amount of training data available. To overcome these difficulties, this study proposed a query expansion approach based on enterprise knowledge domain categories and semantic relevance. The proposed method first makes use of a topic model and the expertise of professionals to create enterprise knowledge domain categories with weighted description terms, then classifies queries using semantic similarity into knowledge domain categories and selects terms for expansion from category description terms. This research used an electronic manufacturing company as case for experimental study. The experiment s results proved that the query expansion method effectively improves the enterprise information retrieval accuracy. 2019 Vol. 38 (7): 742-749 [Abstract] ( 157 ) HTML (127 KB)  PDF (803 KB)  ( 645 )

750 A Study on the Evolution of Knowledge Network Structure Features Based on Patent Intelligence: Empirical Analysis of the Automobile Industry Hot! Xu Luyun, Zeng Deming, and Chen Jing DOI: 10.3772/j.issn.1000-0135.2019.07.010 The key to technological innovation is to realize knowledge integration and combination configuration. Different strategies of knowledge integration and combination configuration lead to heterogeneous structural features of knowledge networks, which reflects the differences in firms’ innovation capabilities and decisions. Using the invention and utility types of patents in China's automobile industry from 2001 to 2014, this paper employed data from 961 firms and constructed their knowledge networks using the International Patent Classification and focused on the evolution analysis of relational and cohesive features. Through analysis of the structural features of 2821 knowledge networks, we found that the general level of relational features in China’s automobile industry is low, but that it still shows an upward trend. The automobile industry has formed a certain level of local and global knowledge cohesion, but the level of local cohesion is obviously higher than that of global cohesion. Furthermore, the sample firms were classified based on whether the firms had participated in R&D collaborations or setting technological standards, and the knowledge network structure features of different types of firms exhibited distinguishing characteristics. The research findings are helpful in revealing the law of knowledge assimilation of China s automobile industry and providing some theoretical guidance for promoting the cross-border integration of technology to enhance the capability for independent innovation in the future. 2019 Vol. 38 (7): 750-759 [Abstract] ( 275 ) HTML (93 KB)  PDF (1096 KB)  ( 865 )

760 Change of Gaze Behavior in Information Preparation and Resumption of Cross-Device Search Based on Query Lists Hot! Wu Dan, Liang Shaobo, and Dong DOI: 10.3772/j.issn.1000-0135.2019.07.011 The visual gaze behavior of SERP (search engine result pages) users is important in the field of information retrieval. Specifically, as the popularity of cross-screen interaction and cross-device web searching increases, the adjustments to a user’s gaze behavior during device transition have become a hot research topic. Users will submit a series of queries when facing a complex search task, but research is lacking about the usersgaze behavior based on their perspectives of those queries. This paper collects eye movement data while performing cross-device web searches during user experiments, including basic eye movement, time dimension, and spatial dimension phenomena. The results reveal that both the gaze duration and gaze count in “information resumption” are lower than in “information preparation.” The evolution of the gaze region also changed, and both the saccade and regression counts are generally reduced. In addition, tools that support a user’s cross-device web search can effectively reduce eye movement load. 2019 Vol. 38 (7): 760-770 [Abstract] ( 217 ) HTML (99 KB)  PDF (2899 KB)  ( 673 )