基于标签相似度计算的学术圈构建方法  

Construction Method of Academic Circle Based on Label Similarity Computation

作　　者：傅城州[1] 汤庸[1] 贺超波[2] 王津凌 袁成哲 

机构地区：[1]华南师范大学计算机学院,广州510631 [2]仲恺农业工程学院信息科学与技术学院,广州510225

出　　处：《计算机科学》2016年第9期52-56,76,共6页Computer Science

基　　金：863高技术研究发展计划项目基金(2013AA01A212);国家自然科学基金(61272067;61502180);广东省公益研究与能力建设专项(2015A020209178)资助

摘　　要：为面向学者的社交网络系统中的用户构建学术圈,对促进学者之间的交流具有重要的应用价值。根据学者之间的共同属性进行相似度计算,形成学术领域相似和研究课题相近的学术圈,能让学者们更加紧密和频繁地协同合作。提出了利用学者的学术信息提取代表个人特征的学术标签,并对不同类别标签的权重进行衡量,再通过相似度计算和聚类算法构建学术圈的方法。通过抓取学者社交网络平台SCHOLAT公开的学者信息进行实验,进而验证所提方法的可靠性和实用性。Constructing academic circles for users in the scholar-oriented social network system has important applica- tion values for promoting exchanges among scholars. Similarity computation is done based on the common properties of scholars, constituting academic circles with similar academic field and research subject, allowing scholars to collaborate more closely and frequently. This paper proposed a method which uses scholars of the academic information to extract the personal characteristics of academic labels, and measures the weight of different class labels. Then through the simi- larity computation and clustering algorithm, the academic circle can be constructed. By crawling the public information on the academic social network platform named SCHOLAT to perform experiments, the reliability and usefulness of the proposed method are verified.

关 键 词：社交网络 标签 相似度计算 聚类算法 学术圈 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]