Document Retrieval Based on Knowledge Acquisition and Merging: A Methodology

摘要

Most part of the document retrieval systems use defined keywords by the user to find documents. In the information retrieval, for some application, is important to have a system that, instead of putting a keyword, takes a complete document and gives us as output a set of documents more similar to the input document. This paper presents an approach to compare documents and proposes - through the use of MFS to represent the tokens - two heuristics that can be used together with TF-IDF algorithm in order to improve the results in the document retrieval. The results showed that the use of these heuristics improves the accuracy of the TF-IDF algorithm and generates smaller indexes for document representation.