The Research On Semantic Field Model For Clustering Based On Domain Ontology With Topic Model

With the arrival of the era of social network,the comment is rapidly becoming a crucial expression method in the Internet.The emergence of comments makes people express their opinionsmore freely and convenient.And it also becomes a new form of information acquisition.However,with the comments flood in the Internet,how to organize,induce and improve acquisition efficiency from such huge amount of information resource has become a problem.Document clustering is used to solve the problem of information organization and induction.But the traditional document clustering method often has the problem of high dimensions,data sparse and lack of information semantics.Aiming at the shortage of traditional text clustering method,this thesis presents a semantic field model for clustering based on domain ontology with topic model.And the model is used to tourism domain.The research work mentioned in the paper is as following:(1)The domain ontology is constructed.Taking the case of tourism domain,this paper analyses tourism document and determines the scope and important terms of tourism ontology.C lasses,relationship and property of tourism ontology are defined.Tourism ontology construction is completed by creating the tourism instances.Tourism ontology represents the tourism domain knowledge.(2)Considering the incomplete ontology,this paper studies the extension of domain ontology based on topic model.The topics of comment are identified with topic model and used for extend domain o ntology.The similarity between domain ontology and topics are calculated by Word2 vec technology.The similarity between domain ontology and topics not only reveals the means of topics,but also establish the link between the topics and domain ontology.O n the basis of the similarity between topic and ontology,topics can be selected and then are used for extend the ontology.Taking case of tourism,this paper designs multiple sets of domain ontology expansion based on topics.Through multiple sets of contrast experiments,the optimal topics are selected to supplement the tourism ontology.The experimental results shows that method for the extension of domain ontology based on topic model can effectively improve the describing capability of domain knowledge.(3)To solve the problem of lack of guideline for feature selection,domain ontology is used to guide the feature selection,transform and reduction.Domain ontology represents the scope of specific domain.Besides,it is the collection of important concept s for specific domain.So using the domain ontology to extract the domain feature,turning the instances into classes and selecting features within ontology can solve the problem of lack of guideline for feature selection.(4)Considering the problem of lacking semantic information for traditional text clustering,the extension of domain ontology based on topic model is used to construct the semantic filed and it is introduced into tourism domain.The semantic field depicts the important concepts,the relation of concepts and semantic distribution of specific domain.The document clustering is transformed into calculating the value of the interaction forces among documents.Experimental result shows that semantic field with domain ontology extended by topic model can enhance the effect of text clustering.