针对传统聚类算法在处理大规模和高维文本聚类时存在的不足和局限性,提出了新的以LDA(latent dirichlet allocation)模型为基础的聚类方法 。通过LDA主题模型挖掘得到文本之中的潜在主题分布以及不同主题内的词语分布,分别计算文本在“文本-主题”特征空间和“主题-词语”特征空间的相似度,然后对两者线性加权,获得最终的文本相似度。利用经典的K-Means算法,在中英文语料库上进行的实验表明,与单纯地利用VSM结合K-Means相比,具有较好的聚类效果。
In order to overcome the shortcomings and limitations of traditional clutering algorithms in dealing with largescale and high dimension text clustering, a text clustering method is presented based on weighted LDA (latent dirichlet allocation) model. Two distributions are obtained by LDA: the topic distribution and the word distribution of different topics hidden in the text, which are then combined as the text feature to obtain the final text similarity. Using the classic K-Means algorithm in both English and Chinese corpus, the experimental results show that compared with the pure VSM combined with K-Means, this algorithm has better clustering effect.