摘要
[目的/意义]Alpha Go战胜李世石后,人工智能的研究与发展备受关注。在此之前不久,基于深度学习的Semantic Scholar免费学术搜索引擎的问世,也为科研工作者们搜索和筛选学术文献资源带来了新的体验。[方法 /过程]在介绍人工智能、机器学习和深度学习之间关系的基础上,介绍了Semantic Scholar的检索功能,重点就该引擎基于系统在理解文献内容基础上的学术影响力评价功能作了分析,并将Semantic Scholar与现行主流学术搜索引擎Google Scholar、Microsoft Academic、必应学术和百度学术进行比较研究。[结果/结论]Semantic Scholar通过机器学习可以使系统理解不同引用之间的影响力差异,提出了基于引用内容分析的学术影响力评价指标,但在信息来源、学科范围、检索功能和个性化服务功能方面还有待进一步完善。最后提出今后学术搜索引擎的发展展望。
[ Purpose/Significance ] It had triggered more concern about artificial intelligence （AI） research and development after Google＇s AlphaGo computer program outperformed the human world champion, Lee Sedol, with a winning result of four to one in the ancient game of Go. Shortly before that, Semantic Scholar, a free scholarly search engine, which was based on deep learning, brought a new experience for researchers to search and select scholarly literature information. [ Method/Process] Based on a brief introduction of the relationship of AI, machine learning and deep learning, this paper introduces the search function of Semantic Scholar, focusing on its evaluation function of academic influence which is based on the understanding of the literature content. Then, the paper compares Semantic Scholar with Google Scholar, Microsoft Academic, Bing Scholar and Baidu Scholar from the aspects of information sources, search functions, academic influence evaluation and personalized service. [ Result/Conclusion ] Semantic Scholar can understand the influence differences among different citations via machine learning, which helps to develop an evaluation index of academic influence based on the citation content analysis. However, it still needs to be further improved in information source, subject scope, retrieve function, and personalized services. The future development prospects of the academic search engine are pointed out at last.
作者
谢智敏
郭倩玲
Xie Zhimin Guo Qianling(Library of Beijing University of Chemical Technology, Beijing 100029)
出处
《情报杂志》
CSSCI
北大核心
2017年第8期175-182,共8页
Journal of Intelligence
