Ensemble Approach for Keyword Extraction

Abstract

Text mining likewise referred to as content information mining, generally comparable to content investigation, is the way toward getting astounding data from content. The data is normally determined through the conceiving of examples and patterns through means, for example, factual example learning. Statistical analysis based techniques has done a broad analysis of the text in perspective of the predicted execution of classification and storage strategies. Analysis of text and techniques occurred when keyword based representation is used to classification of logical content documents. The presentations of content documents with keyword in a concise way can be greatly valuable, since content documents are described by the high dimensionality of highlight space. In this thesis the main emphasis is on the natural language processing based extraction of quality keywords from content. These keywords are used to search the multiple documents in the database or dataset. For the keywords extracted ranking and scoring is given to documents based upon the similarity matching in the documents, similarity matching is based upon the cosine similarity. In the last ensembling of the whole procedure is to be done using the proposed approach. From the results and discussion part it is clear that the accuracy, precision, recall and f-measure parameters are better in case of the proposed approach as compared to the existing system that is statistical analysis based keyword extraction.