刘爱琴 贾一帆 冷长青

摘 要：[目的]为了实现知识的快速聚类和关联分类，由传统的以资源检索为目标的高校图书馆系统转变为完全面向用户需求的、主动发现和推送知识的图书馆知识发现系统。[过程]本文融合网络爬虫技术和学术资源网站结构化数据的特征，构建了基于随机游走模型，依据摘要词频对文献资料进行主题词的提取、聚类;随后在标签信息标注的基础上，根据相似性对游走过程进行加权处理;最终完成了知识关联分类的知识发现系统。[结果]本文实现了用高效的知识提取手段，基于智慧云、物联网构建更加准确和更具关联性的知识发现系统，提高了高校图书馆知识检索系统的查全率和查准率。

关键词：高校图书馆;摘要信息标注;加权处理;随机游走模型;知识发现系统

DOI：10.3969/j.issn.1008-0821.2020.05.012

〔中图分类号〕G250.74 〔文献标识码〕A 〔文章编号〕1008-0821（2020）05-0096-08

Construction of Knowledge Discovery System in

University Library Based on Random Walking Model

Liu Aiqin Jia Yifan Leng Changqing

（School of Economics and Management，Shanxi University，Taiyuan 030006，China）

Abstract：[Purpose]In order to realize the fast association classification of knowledge，the traditional university library system with the goal of resource retrieval should transform into a library knowledge discovery system that completely faces the needs of users and actively discovers and pushes knowledge.[Process]In this paper，based on the combination of web crawler technology and the characteristics of structured data of academic resources websites，a random walk model was constructed，and the subject words were extracted and clustered according to the abstract word frequency;then，on the basis of label information tagging，the walking process was weighted according to similarity;Finally， the knowledge discovery system of knowledge association classification was completed.[Result]This paper realized the use of efficient knowledge extraction means to build a more accurate and relevant knowledge discovery system based on the cloud of wisdom and the Internet of things，which improved the recall and precision of the university library knowledge retrieval system.

Key words：university library;information tagging of the abstract;weighted processing;random walk model;knowledge discovery system

1 研究背景

用戶知识需求的不断增加，要求获取信息的准确性和关联性越来越高[1]，高校图书馆利用高效的知识提取手段和知识发现技术，基于智慧云、物联网构建更加准确和更具关联性的知识发现系统，实现短时间内准确获取知识[2]，解决用户的特定问题和满足用户日益增长的个性需求，要比高校图书馆提供传统数据库更吸引用户的使用[3]。由传统的以资源检索为目标转变为完全面向用户需求的、主动发现和推送知识的图书馆知识发现系统的工作已迫在眉睫。

根据研究对象的数据特点，国内外学者正在展开积极的研究。……