網際網路的蓬勃發展造成資訊迅速膨脹，資訊檢索系統為了幫助使用者取得所需的資訊亦隨之發展。本研究提出了一套方法，能夠擷取並利用顯性相關回饋的資訊在概念萃取的機制當中，並利用萃取出的概念來表達文件，建立一個以概念為維度的向量空間模型，最後應用此模型在文件分類上，以提升文件分類的效能。實驗結果顯示本研究所提出的方法，分類效果都較傳統以字詞作為文件特徵時來得好；本研究中也實驗將亂數的相關回饋資訊應用在概念萃取的機制當中，結果發現其分類效果都較傳統以字詞作為文件特徵時來得差上許多，因此本研究證實，顯性相關回饋中，確實有資訊可應用於概念化文件建立以促進文件分類之效能。 The rapid development of the Internet causes the problem of information explosion. To solve this problem, information retrieval system is developed to help user in the finding of the information of their needs. This study proposes an approach that extracts information from relevance feedback to construct a concept extraction algorithm. At first, extracts concepts from the document set, and uses these concepts as document’s attributes. Then, creates a vector space model with the extracted concepts as dimension value for the document. Finally, uses the proposed model to improve the performance of document classification. The result of experiments show that the proposed approach can perform better than term based vector model. This study also apply the information of random relevance feedback to construct the concept extraction algorithm. The results of the experiments show that the application of the information of random relevance feedback performs much worse than term based vector model. This study confirms that the application of the information of explicit relevance feedback to create a vector space model with the extracted concepts as dimension value for the document can improve the performance of document classification.
