Text mining defines generally the process of extracting interesting features (non-trivial) and knowledge from unstructured text documents. Text mining is an interdisciplinary field which depends on information retrieval, data mining, machine learning, parameter statistics and computational linguistics. Standard text mining and retrieval information techniques of text document usually rely on similar categories. An alternative method of retrieving information is clustering documents to preprocess text. The preprocessing steps have a huge effect on the success to extract knowledge. This study implements TF-IDF and singular value decomposition (SVD) dimensionality reduction techniques. The proposed system presents an effective preprocessing and dimensionality reduction techniques which help the document clustering by using k-means algorithm. Finally, the experimental results show that the proposed method enhances the performance of English text document clustering. Simulation results on BBC news and BBC sport datasets show the superiority of the proposed algorithm.

A major challenge in topic classification (TC) is the high dimensionality of the feature space. Therefore, feature extraction (FE) plays a vital role in topic classification in particular and text mining in general. FE based on cosine similarity score is commonly used to reduce the dimensionality of datasets with tens or hundreds of thousands of features, which can be impossible to process further. In this study, TF-IDF term weighting is used to extract features. Selecting relevant features and determining how to encode them for a learning machine method have a vast impact on the learning machine methods ability to extract a good model. Two different weighting methods (TF-IDF and TF-IDF Global) were used and tested on the Reuters-21578 text categorization test collection. The obtained results emerged a good candidate for enhancing the performance of English topics FE. Simulation results the Reuters-21578 text categorization show the superiority of the proposed algorithm.

Currently, the dynamic website has increased with more than thousands of documents associated to a category topic available. Most of the website documents are unstructured and not in an arranged method and thereby the user suffer to obtain the related documents. A more helpful and efficiency technique by combining document clustering with ranking, where document clustering can collection the similar documents in one category and document ranking can be carried out to each cluster for selecting the best documents in the initial categorization. Besides the specific clustering technique, the different types of term weighting functions implemented to select the features that it represents website document is a chief part in clustering mission. Moreover, document clustering indicates to unsupervised categorization of text documents into clusters in such a method that the text documents in a specific cluster are similar. Therefore, this study proposed a new technique combined chi-square with k-means for clustering the website documents. Furthermore, this study implements information gain and chi-square combined with k-means for document clustering. It helps the user to obtain the whole related documents in one cluster. For experimental objective, it has selected the BBC sport and BBC news datasets to show the superiority of the proposed technique. The experimental findings show that the chi-square with combined with k-means clustering improves the performance of document clustering.