New Approaches for Extracting Arabic Keyphrases

Keyphrases extraction has a considerable importance in many applications such as search engine optimization, clustering, summarization, and sentiment analysis. The importance of keyphrases comes from the semantic meaning they provide as they can be used as descriptors for the documents. In this paper we compare four approaches for extracting keyphrases from Arabic documents. The first method uses the KP-Miner keyphrase extraction system. The second method uses Arabic natural language processing tools (stemmer and part of speech tagger) in order to filter some patterns that can be weighted by token frequency inverse document frequency (TF-IDF) algorithm. The third method uses Google'sWord2Vec library to calculate the weighting of the resulting patterns by measuring the similarity of the candidate pattern and the document title. The fourth method combines the weightings result from the second and the third method.