论文部分内容阅读
【目的】通过对国内4大微博平台中特征词质量的测度,探讨其质量指标对检索效果的影响。【方法】将权重计算指标TF-IDF从特征词角度提升为特征的研究,并通过描述能力和辨别能力两个质量测度指标对国内4个主流微博平台中各特征的质量进行评估。【结果】微博中文本特征的描述能力和辨别能力对检索效果产生正向影响;各平台不同特征的质量对分类有着不同程度的影响,两种测度指标综合考虑时得到的分类效果最好。【局限】微博中的对话回复、粉丝数、关注数等特征并没有被考虑在内;对于语义研究中的特征词一词多义或者同义词的讨论并未涉猎。【结论】本研究可更好地揭示微博中各种特征影响检索效果好坏的重要程度,有助于研究者对各平台特征作用的深入理解,从而从根本上提高社会化媒体平台的检索质量。
【Objective】 By measuring the quality of feature words in the four Weibo platforms in China, the impact of quality indicators on the search results is discussed. 【Method】 Research on the feature of weighting index TF-IDF was promoted from feature point to feature, and the quality of each feature in 4 domestic mainstream microblogging platforms was evaluated by describing two quality measures indexes of ability and distinguishing ability. [Results] The descriptive ability and discriminating ability of the Chinese text in the Weibo had a positive impact on the retrieval performance. The quality of the different features of each platform had different degrees of impact on the classification, and the best classification results were obtained when the two measures were taken into account. [Limitations] The features such as the number of conversations, the number of fans and the number of followers in Weibo are not taken into account. The discussion of polysemy or synonym for feature words in semantic research is not covered. [Conclusion] This study can better reveal the importance of various features in Weibo in influencing the retrieval quality, and help researchers to understand deeply the features of each platform so as to fundamentally improve the retrieval of social media platforms quality.