作者投稿和查稿 主编审稿 专家审稿 编委审稿 远程编辑

计算机工程 ›› 2011, Vol. 37 ›› Issue (19): 62-64,67. doi: 10.3969/j.issn.1000-3428.2011.19.019

• 软件技术与数据库 • 上一篇    下一篇

基于LDA模型的餐馆评论排序

吕韶华,杨 亮,林鸿飞   

  1. (大连理工大学计算机科学与技术学院,辽宁 大连 116024)
  • 收稿日期:2011-03-23 出版日期:2011-10-05 发布日期:2011-10-05
  • 作者简介:吕韶华(1986-),男,硕士研究生,主研方向:情感计算; 杨 亮,博士研究生;林鸿飞,教授、博士、博士生导师
  • 基金资助:
    国家自然科学基金资助项目(60673039, 60973068);国家“863”计划基金资助项目(2006AA01Z151);教育部博士点基金资助项目(20090041110002)

Ranks of Restaurant Reviews Based on LDA Model

LV Shao-hua, YANG Liang, LIN Hong-fei   

  1. (School of Computer Science and Technology, Dalian University of Technology, Dalian 116024, China)
  • Received:2011-03-23 Online:2011-10-05 Published:2011-10-05

摘要: 在餐馆评论中,存在评论文本未明确指出评价等级及评论文本不一致等问题。为此,提出一种基于LDA模型的餐馆评论排序方法。利用LDA模型对评论文本进行主题抽取,过滤掉不相关评论,基于过滤后的用户评论和用户给出的评论等级计算餐馆评论若干方面的得分,在该得分的基础上,利用逻辑回归进行训练,得到餐馆评论排序模型。实验结果表明,该方法的排序效果较优。

关键词: LDA模型, 餐馆评论, 排序, 观点挖掘, 逻辑回归

Abstract: In order to solve the problems of implicit aspects in review text and the inconsistency between rank of review and review text, this paper makes use of LDA on restaurant reviews to get the useful topics and discard unrelated ones, then gets the scores of some aspects based on them, and at last a model, which can predict ranks of restaurants based on new reviews, is trained with logistic regression using these scores. Experimental results show that the effectiveness of this method is better.

Key words: Latent Dirichlet Allocation(LDA) model, restaurant reviews, rank, opinion mining, logistic regression

中图分类号: