| Summary: | 碩士 === 國立成功大學 === 資訊工程學系碩博士班 === 91 === Nowadays, we can retrieve information from search engine, but traditional search engine can’t perform the search with high quality and low quantity of information. That’s because of traditional search engine mainly use pure keyword and some statistic data to process the similarity between query words and documents. In the thesis, we first propose an infrastructure of object-oriented ontology to be the knowledge base of fuzzy inference model. Then we apply Chinese grammar to do syntax processing for separating different parts of sentences, and we put kernel of sentences to gradually infer the resulted instance with domain ontology. We collect various linguistic messages in documents, such as passive, negative voice and semantic degree to store in index at this stage. At last we apply Extended Boolean Model to be the personalized ranking mechanism. On the other hand, we apply Genetic Algorithm on tuning the parameters of TSK model to make our system more robust. Besides, we take the CKIP as the part-of-speech tagging tool, and it is the base of Chinese grammar analysis.
|
|---|
