Relevance Feature Discovery for Text Mining

Abstract

With so many phrases, patterns, and noise in text, ensuring the quality of features identified for defining user performance that are relevant is a difficult task. Most of the popular text mining and categorization algorithms now in use utilize term-based techniques. Polysemy has afflicted all of them, though. Most Internet search results are based on patterns rather than meaning. when user searches for a word, only specific keywords will be considered for results and includes noise. Instead of looking for patterns, the focus of this study is to determine the meaning of provided words and then provide comparable material because of that inquiry and that can be implemented using Sematic search algorithm. In the end, the user receives the accurate information they requested, along with context. Here, it gives around 80% - 90% of accuracy in results with cancelled noise. Positive, negative, and neutral facts are all readily apparent in this search.