多源数据情感分析在图书荐购工作中的应用研究

Research on the Application of Multi-source Data Emotion Analysis in Book Recommendation

编写独立于图书管理系统之外的Windows应用程序,爬取多个图书类网站评论并使用SnowNLP开发库执行情感分析为图书赋分,供采访馆员决策参考。结果表明,软件功能确切,多源数据接口支持插件扩展。缺点为无法甄别炒作与恶评。程序可有效节省采访馆员机械性重复劳动,在一定程度上保障了不同采访馆员处理不同批次荐购需求时采访原则的统一性。

This paper programs a Windows application independent of the library management system to crawl the re⁃view informations of multiple book websites,and uses SnowNPL to carry out emotional analysis to assign scores to books for the reference of the acquisition librarians.The result shows that the software has exact functions and good expansibility in collecting multi-source data.The disadvantage is that it can not distin⁃guish hype and bad reviews.The program can effectively save the mechanical repetitive work of the inter⁃view librarians,and to a certain extent ensure the unity of acquisition principles when different interview li⁃brarians handle different batches of purchase recommendation needs.