Automatic Key-Phrase Extraction: Empirical Study of Graph-Based Methods

Abstract

Key-phrases in a document are phrases that provide a high-level description of its content without reading it completely. In some research articles, authors specify key-phrases in the articles they have written. However, the vast majority of books, articles, and web pages published every day, lack key-phrases. The manual extraction of these phrases is a tedious task and takes a long time. For this reason, automatic key-phrase extraction (AKE), which is an area of Text Mining, remains the best solution to overcome these difficulties. Because they are used in many Natural Language Processing (NLP) applications, such as text summarization and text classification. This article presents a comparison of some methods of extracting key-phrases from documents. Especially the graph-based approaches. These approaches are evaluated by their abilities to extract key-phrases. Our work focuses on the study of the performance of these methods in extracting key-phrases, whether from short or long texts, with the aim of providing information that contributes to improving their efficiency.