TTM DMM Topical Translation Model for Short Text Keyphrase Extraction
WANG Rui
QIN Yongbin
YAN Yingying
Abstract：The keyphrase extraction methods for short texts based on topical translation model all use LDA(Latent Dirichlet Allocation model)as the topical discovery method. However,when utilizing LDA topic model deals with short texts which are of sparse features,the effect of topical discovery is poor,which leads to the imperfection of the current topical translation model.By us-ing DMM(Dirichlet Multinomial Mixture)as topic discovery model and combining statistical machine translation,a new keyphrase extraction method for short texts,namely TTM DMM(Topical Translation Model based on Dirichlet Multinomial Mixture),is pro-posed in this paper.In this model,DMM model is used to find topical information of short texts,and then topic-specific translation probability is estimated between words and keyphrases,thus improve the effect of keyphrase extraction for short texts.Experimental results on the real dataset demonstrate that the proposed TTM DMM model is superior to the existing keyphrase extraction methods for short text in evaluation indicator Precious,Recall and F-measure.
Keywords：TTM DMM topical translation modelkeyphrase extraction for short textsDMMLDAstatistical machine translation
Publication Date：2018-01-01
Online Publishing Date：2025-08-15(First online date of this platform, not the publication date of the document)
Pages：6( 945-949,955 )