The Study Of Agricultural Scientific Information User Modeling System Based-on Ontology

The fast developing Internet has brought great convenience to us in our routine, in the meanwhile the explosive increase of the resources of the internet information also inevitably contributes some problems. Besides, as the internet information space is open and distributive, its resulted inherent characters, such as heterogeneity, multiplicity and distributivity, also cause some more and more serious consequences, known as"Information Overload"and"resources wilder". As we know, the search engine has made the information search much more easy, however currently the common search engines are still staying on the"Keyword"level, which lacks of semantic meaning and pertinence and the information also flows away to some extent. On the aspect of pertinence, this study describes the user information with user model to supply personalized services. On the aspect of semantic meaning, this study drawing ontology to supply understanding in the semantic level, to analyze user requests and information resources.The user modeling technology can be traced back to the late of 1970s. The research of user modeling began very late in China, and the user model based on ontology has just emerged in the recent years. From the existing references, we can learn that the work is still desolate in the field of agriculture. This study takes some users of agricultural science and technology for example, to discuss how to access the users'knowledge structure, how to express users'information requests and perferences, then obtain user model from these analysis.The main issues of this study include:(1) Agricultural Thesaurus is used as primary Agricultural ontology, and concepts are extracted from literatures,which are expressed as the knowledge structure of a user.(2) Agricultural Thesaurus is added to general vocabulary, which posses larger weight so that the agricultural words could be preferentially segmented. The ontology is the base for semantic annotating of the literatures to matching the words of the documents and the concepts of the ontology and modeling the user ontology.(3)The frequencies of the knowledge elements in the literatures are calculated by TF-IDF algorithm as vectors of user's concepts, which is the record of the user's preference.(4) The User model consists of User ontology and vectors of user's concepts.(5)The similar degree of the retrieval results of literatures and the user model is calculated by cosine similarity algorithm. The retrieval results of literatures could be reordered in perspective of semantic.(6)As the tools of word segmentation and semantic annotation and a part of user models, ontology is used to match the information resources and the user's knowledge.The user model is expressed as user ontology and vectors of user's concepts. The user ontology is made up of the concepts abstracted from literatures and the words in the literatures are normalized by the ontology so that the vectors of the concepts the user interests in the literatures.The experiment shows this study improved the order of the retrieval results. This study is not only suitable for technological users modeling, but also to the general uers modeling in the Inernet.