| 基于日志和知网的查询推荐研究 |
| |
| 引用本文： | 周德志.基于日志和知网的查询推荐研究[J].现代情报,2013,33(10):65-69. |
| |
| 作者姓名： | 周德志 |
| |
| 作者单位： | 西安电子科技大学经济管理学院, 陕西 西安710071 |
| |
| 摘 要： | 考虑到传统的基于日志的查询推荐算法受到数据稀疏问题的影响,本文在分析查询日志的基础上,构建查询词与点击URL之间的双向图,计算查询词与候选词之间的相似度。然后基于知网计算查询词与候选词之间的相似度,考虑词性和同义词因素对相似度的影响。最后将两个相似度分别赋予权重计算查询词与推荐词的相关度。实验结果表明,该方法不易受数据稀疏问题的影响,稳定性较好。
|
| 关 键 词： | 查询日志 查询推荐 双向图 |
Study on Query Recommendation Based on Log and HowNet |
| |
| Authors: | Zhou Dezhi |
| |
| Institution: | School of Economics and Management, Xidian University, Xi'an 710071, China |
| |
| Abstract: | Considering the traditional recommendation algorithm based on log query affected by the problem of data sparse- ness, on the basis of analyzing the query log, the paper will construct a two- dimemional plot between word and clicked URL, it is used to calculate the similarity between the query words and the candidate words. Then we will calculate the similarity between the query words and the candidate words based on HowNet, considering the influence on similarity as the part of speech and syn- onym. Finally, the two similarity were given weight that we can calculate the correlation between the query words and the recom- mend words. The experimental results show that this method can overcome the problem of data sparseness, has more stability. |
| |
| Keywords: | query log query recommendation two- dimensional plot |
| 本文献已被 维普 等数据库收录！ |
| 点击此处可从《现代情报》浏览原始摘要信息 |
| 点击此处可从《现代情报》下载免费的PDF全文 |
|
