Toward a New Process for Candidate Key-Phrases Extraction

Key-phrases in a document are the terms that allow us to have or know an idea of its content without having to read it. They can be used in many Natural Language Processing (NLP) applications, such as text summarization, machine translation, and text classification. These phrases are selected from a set of terms in the document called candidate key-phrases. Thus, any flaws that may arise during the selection of candidate phrases may affect the automatic key-phrase extraction (AKE). Despite the importance of identifying candidate key-phrases in the AKE process, we found a very limited number of researchers interested in identifying their features in the document. In this paper, we will present the features that will allow the identification of candidate key-phrases, based on the study and analysis of the features of 60,000 key-phrases manually selected from five different datasets. To improve the performance of AKE approaches. © 2022, The Author(s), under exclusive license to Springer Nature Switzerland AG.