Bangla Text Summarization Analysis Using Machine Learning: An Extractive Approach

A notable expansion of information in digital space provides quick access to these massive amounts of news, literature, and so on. Although maximum news is available in English, people are interested in reading in their language. Almost all Bangla newspapers now have their online version, so people prefer to read newspapers online. Without reading the whole news, it is not possible to obtain significant information and a summary from a news article. A news summary is a process of extracting the most significant information from a news article in a precise manner. In this case, we require an effective Bangla text summarizer tool to get insight from the news easily. Text summarizers use linguistic methods to generate and interpret text before uncovering natural concepts and phrases to represent the content by compressing the original text document into a shorter text that contains the most relevant information. Summarizing lengthy and important text by enforcing specific guidelines is a challenging task. Therefore, a method based on intelligent machine learning is widely anticipated in this content. This paper uses extractive automatic text summarization techniques to condense the source text into a shorter version by preserving the significant information of the original text. Here we have adopted two machine learning-based techniques (i.e., word2vec, and similarity matrix) and a rule-based technique (word count) for Bangla text summarization. The results show that the word2vec technique performs very well compared to the summary performed by other techniques.