基于主题模型和文本相似度计算的专利推荐研究

如何利用数量庞大的专利并从中找到用户感兴趣的专利进行推荐是很多专利数据库迫切需要解决的问题。文中从专利文本的标题和摘要入手,提出一种基于文本挖掘的专利推荐方法。首先,利用词袋模型将专利文本转化成计算机能够识别的数据;其次,利用文本聚类算法完成专利数据集进行领域划分;再次,结合词频-逆文档频率特征权重计算和余弦相似度来选择合适的发明人进行专利的推荐;最后,以我国物流产业下的专利数据作为数据集完成文中所提方法的验证与分析。实验结果表明,基于文本挖掘的专利推荐研究能够实现对发明人的个性化推荐。

Patent recommendation research based on topic model and text similarity calculation

How to use a large number of patents and find patents of interest from relevant inventors is a problem that many patent databases urgently need to solve.Starting with the title and abstract of the patent text,a patent recommendation method based on text mining is proposed.Firstly,the word bag model is used to convert patent text into computer-recognizable data.Then,the text clustering algorithm is used to complete the patent data set for domain division.The word frequency-inverse document frequency feature weight calculation and cosine similarity are used to select the appropriate inventor for patent recommendation.Finally,the verification and analysis of the proposed method is completed by using the patent data under the logistics industry in China as the data set.The experimental results show that the patent recommendation research based on text mining can realize the personalized recommendation to the inventor.