Unsupervised Approach for Automatic Keyword Extraction from Arabic Documents

The 2014 Conference on Computational Linguistics and Speech Processing ROCLING 2014, pp. 175-184  The Association for Computational Linguistics and Chinese Language Processing Unsupervised Approach for Automatic Keyword Extraction from Arabic Documents Arafat Awajan  Abstract In this paper, we present an unsupervised two-phase approach to extract keywords from Arabic documents that combines statistical analysis and linguistic information. The first phase detects all the N-grams that may be considered keywords. In the second phase, the N-grams are analyzed using a morphological analyzer to replace the words of the N-grams with their base forms that are the roots for the derived words and the stems for the non-derivative words. The N-grams that have the same base forms are regrouped and their counts accumulated. The ones that appear more frequently are then selected as keywords. An experiment is conducted to evaluate the proposed approach by comparing the extracted keywords with those manually selected. The results show that the proposed approach achieved an average precision of 0.51. Keywords: Keyword extraction, Keyphrase extraction, Arabic Language, N-gram. 1. Introduction Keyword extraction is the process of identifying a short list of words or noun phrases that capture the most important ideas or topics covered in a document. Keyword extraction has been used in a variety of natural language processing applications, such as information retrieval systems, digital library searching, web content management, document clustering, and text summarization (Rose et al. 2010). Although keywords are very useful for a large spectrum of applications, only a limited number of documents with keywords are available on-line. Therefore, appropriate tools that can automatically extract keywords from text are increasingly needed with the continually growing amount of electronic textual content available online. In this paper, an unsupervised two-phase approach for keyword extraction from Arabic  Princess Sumaya University for Technology – Department of Computer Science, Amman – Jordan E-mail: awajan@psut.edu.jo 175 documents is described. The proposed method combines the document’s statistics and the linguistic features of the Arabic language to automatically extract keywords from a single document in a domain-independent way. In the first phase, all the N-grams are extracted and those considered as potential candidate keywords are retained. In the second phase, the candidate keywords are analyzed linguistically by a morphological analyzer that replaces each term with its base form, which are the roots of the derived words and the stems of the non-derivative words. The candidate keywords are then grouped in such a way that the keywords extracted from similar roots and stems are put together and their counts accumulated. This paper is organized as follows. In section 2, we present related works and the main approaches to keyword extraction. Section 3 highlights the main Arabic language features used in our technique. A detailed description of the proposed technique and its two phases provided in Section 4 and Section 5. Section 6 consists of the experimental results and the main findings of the evaluation of the proposed method. 2. Related Work Existing automatic keyword extraction methods can be divided into …

Content truncated to 500 words