Classification of Keyphrases using Random Forest

Keyphrases are words or phrases from a document that can describe its meaning. A keyphrase integrates the general idea of a document and implicitly contains the resources that the author used during the development of its research to achieve his goal. Therefore, there is a need to create classification models that allow the clustering of keyphrases according to their content for simplify reading. In this paper, keyphrases classification from scientific publications based on LSA and some classifying techniques is proposed and implemented. The aim is to create a classification model based on the extraction of features from the input corpus, without enriching it using external resources such as Wikipedia or online resources. Process, task, and material are the classes considered from Computer Science, Material Sciences, and Physics publications domains. Results show that Random Forest was found to be the best classification technique of keyphrases with 60% of measure-F-1.