中文摘要：

基于搜索学术论文对研究者造成的困扰问题,在学术论文聚合平台基础上提出一种学术社交平台相似论文推荐方法,给出了推荐方法的总体架构及各部分的详细设计方案。该方法首先使用ANSJ对论文数据集中的论文进行分词并统计词条的TF-IDF,使用这些词条表示该论文的关键信息。其次使用Word2Vec把每一篇论文映射到一个高维向量,使用余弦相似度公式计算其与用户查询论文间的相似度,根据相似度结果高低生成论文推荐列表。最后在SCHOLAT论文数据集上通过应用实例以及量化指标分析验证了该推荐方法的有效性。

英文摘要：

According to the defects that researchers search academic papers.This paper proposed a similar paper recommendation method in scholar social platform that includes several popular search engine,and explainedthe framework of recommendation method and detailed design of the system.This recommendation method executes word-segmentation with ANSJ,calculate the TF-IDF of lemma and extract paper key words in initialization.Next,read the segmentation result to get the word-vectors by Word2 Vec,calculate its similarity with querypaperfrom users according to cosine similarity formula.And further,the paper recommendation list will be generated.In the end,the efficacy will be proof by an application instance and quantitative index analysison SCHOLAT paper dataset.