基于游记主题挖掘与表达的旅游信息推荐研究

Research on Tourism Information Recommendation—Based on Topic Mining and Presentation of Travel Notes

[目的/意义]针对在线旅游平台,提出一种挖掘游记主题标签,以代表性游记以及其中相关内容进行旅游信息推荐的新策略。[方法/过程]在利用文本挖掘技术,构建LDA主题模型,形成游记文本主题标签的基础上,通过游记代表度算法,筛选出针对相应标签的高描述度与高忠诚度游记进行旅游信息推荐,以客观表达文本聚类结果以及主题词之间的语义关系,并以蚂蜂窝旅游网中的"杭州游记"为例,加以验证。[结果/结论]结果表明,这种方式能挖掘出旅游者在历史旅游经历中真实的旅游热点及重点信息需求,针对高相似度游记的识别与聚类具有良好效果,对旅游信息细粒度推荐具有指导意义与实践意义。

[Purpose/Significance] The paper presented a new strategy for online tourism platform in information service that finding the labels for core topics in travel notes and using related content of representative travel notes to recommend tourism infor- matiorr [ Method/Process] Using text mining technology, this paper conducted the construction of Latent Dirichlet Allocation model for travel notes data. On the basis of finding the labels for core topics in travel notes, the paper computed travel notes＇ de- gree of description and loyalty in accordance with corresponding labels to get representative notes. Then, the selected results were used for information recommendation, which could objectively express and explain the content of text clustering and semantic rela- tion between subject terms. The paper took the travel notes data of ＂Hangzhou＂ as an example, which were extracted from Mafengwo, for testifying this approach. [Resuh/Conclusion] From the experimental results, this method could discover real hot spots and key tourism demands of tourists in the previous tourism experience. Besides, travel notes with high Similarity had been distinguished and clustered. It had guiding meaning and practical significance for smaller granular recommendation of tourism infor- mation.