Impact of linguistic and ontology information on improvement of Persian text clustering

In Persian language, words have various writing forms and it is impossible to cover all the grammatical points of words by applying a series of certain rules. For this reason, automatic key words extraction from Persian texts is complex and difficult. In this paper, we try to provide more meaningful keywords by using linguistic information and thesaurus. By using thesaurus that has a structured system, we can complete and improve the equivalent, hierarchical and dependent words. So we can increase adaptation between users' search and textual key words and also comprehensiveness of the search.At the first stage, unimportant and general words are removed. Then the roots of words are extracted and in the following, a numerical weight is assigned to each word to define the importance of words by using weighting methods that shows the amount of importance of word in relation to text subject and in comparison with other used words in the text. The above operations specially using thesaurus would cause a more accurate text classification and also somehow hierarchical category of texts in information retrieval context is specified. Test results on some text in different subjects show accurately and ability of proposed method in extracting key words in adaptation with user's demand and lead to more accurate text clustering.

کلیدواژه ها:

extracting key words ، thesaurus ، linguistic ، information retrieval