Variance-based features for keyword extraction in Persian and English text documents

Abstract

This paper address automatic keyword extraction in Persian and English text documents. Generally, for keyword extraction in a text, a weight is assigned to each token and words having higher weights are selected as the keywords. We have proposed four methods for weighting the words and have compared these methods with five previous weighting techniques. The previous methods used in this paper are term frequency (TF), term frequency inverse document frequency (TF-IDF), variance, discriminative feature selection (DFS), and document length normalization based on unit words (LNU). The proposed weighting methods are based on using variance features and include variance to TF-IDF ratio, variance to TF ratio, the intersection of TF and variance, and the intersection of variance and IDF. For evaluation, the documents are clustered using the extracted keywords as feature vectors, and K-means, expectation maximization (EM), and Ward hierarchical clustering methods. The entropy of the clusters and pre-defined classes of the documents are used as the evaluation metric. For the evaluations, we have collected and labelled Persian documents. Results show that our proposed weighting method, variance to TF ratio, has the best performance for Persian. Also, the best entropy is resulted by variance to TD-IDF ratio for English.