ScholarGraph:a Chinese Knowledge Graph of Chinese Scholars

ScholarGraph :a Chinese Knowledge Graph of Chinese Scholars Shuo Wang, Zehui Hao, Xiaofeng Meng, Qiuyue Wang Renmin University of China, Hebei University Beijing, Baoding in Hebei shuowang@ruc.edu.cn, jane0331@126.com, xfmeng@ruc.edu.cn, qiuyuew@ruc.edu.cn Abstract Scholars and their academic information are widely distributed on the Web. Integrating these information and making association between them can play a catalytic role in academic evaluation and research. Since 2008, Web and Mobile Data Management Laboratory (WAMDM) in Renmin University of China began to collect Chinese literatures in more than 20 academic domains, and build a data integration system called ScholarSpace to automatically integrate the relevant chinese academic information from the chinese scholars and science. Focusing on the chinese scholars, ScholarSpace can give you an academic portrait about a chinese scholar with the form of knowledge graph. So the ScholarSpace can be tranformed into a knowledge graph called ScholarGraph. It includes the scholar information such as the affiliation, publications, teacher-student relationship, etc. ScholarGraph is a subset of the whole knowledge graph generated from the ScholarSpace and is published on the web page of WAMDM. ScholarGraph consists of more than 10,000,000 triples, including more than 9,000,000 entities and 6 relations. It can support the search and query about portrait of Chinese scholars and other relevant applications. Keywords: knowledge graph construction,chinese scholars, entity, entity relation 1. Introduction On the Web, many information about any person such as working place, living addresss, friends, personal journey can be searched online. The same things happened in the academia. The academic information related to a scholar are already available on the Web, but they are not effectively linked together. As shown in Figure 1, the relations are actually existing among scholars, journals and academic institutions. In fact, these relations are a small part of the reality. So it is necessary to build an integration system for the academic field. The main character of the academia information is centered on scholars. Because the academic actions are completed by scholars. For example, a professor can publish a paper in a journal or a coference, supervise postgraduate students, cooperate with other scholars, apply for fundings, etc. The relations corresponding to these actions are generated by scholars. Figure 1: Example of scholars’ information. So the data integration system needs to be focused on scholars. According to DBLP, we firstly built a Chinese data integration system – CDBLP – to collect Chinese literature information of the computer science community in China. Based on CDBLP1, we expanded the collection from a single domain of computer science (Chen Wei et al., 2011) to 25 domains including education, management, archaeology, economics, etc. On this basis, we propose ScholarSpace system which was constructed for Chinese scholars to support Chinese search, Chinese query, Chinese journals’ evaluation, etc. There is a wealth of knowledge in these integrated data. Many entities (Professor /Ph.D. /Journal /Paper) and relations (Author /PublishedIN /Affiliation /Advisor) are shown in Figure 1. So a knowledge graph – ScholarGraph – is generated from ScholarSpace. We select a partial data from ScholarSpace and transform …

Content truncated to 500 words