Comparative Study of Clustering Techniques for Extractive Text Summarization

Text summarization aims to generate concise summaries of lengthy text for faster consumption of relevant information. Summaries can be generated either by abstraction or extraction methods. This paper focuses on creating extractive text summaries using different clustering techniques, namely k-means, Hierarchical Agglomerative, Birch, and Affinity propagation using Term Frequency—Inverse Document Frequency (TF-IDF). The optimal number of clusters is determined using the silhouette score, while other statistical measures such as precision, recall, and F1 score are used to evaluate the accuracy of the generated summaries. This paper compares various clustering techniques based on their computational efficiency and their ability to determine the optimal number of clusters. Additionally, it evaluates the accuracy of news article clustering across various sizes while considering the respective advantages and challenges associated with each algorithm.