Due to the problems of traditional paper reviewing method,this paper proposed a paper recommendation system on the basis of a graph algorithm called,TextRank algorithm to achieve the automation of the distribution in paper reviewing process.The system firstly extracted keywords from text by adding the computation of the influences among terms and the in-verse document frequency.Then it made papers match reviewers’interests through computing the cosine similarity of each vec-tor.Finally,it computed the influence of reviewers in each field in order to recommend accurately.When verifying the validity of the recommendation algorithm,the article used precision,recall and F-measure as evaluation indicator.As a result,the sys-tem is proved highly effective and reliable.%针对传统审稿方式所存在的问题,设计了基于 TextRank 图算法思想的论文推荐系统,以实现论文审稿分配过程的自动化。系统通过加入词与词之间的影响力计算以及多文档集中逆文档频率 IDF,实现关键词抽取部分,并使用基于余弦向量值的计算对抽取出的关键词向量进行相似度匹配,最后计算审稿人在各研究领域的影响力,实现论文的推荐。采用了综合考察准确率、召回率的 F 值作为评测指标,验证了该方法的有效性。在实际使用环境中,该系统具有较高的准确性与可靠性。
展开▼