TY - JOUR ID - TI - Proposed aspect extraction algorithm for Arabic text reviews AU - Ahmed bahaa aldeen abdul wahhab AU - aliaa kareem abdul Hassan PY - 2018 VL - 10 IS - 3 SP - Comp Page 79 EP - 90 JO - Journal of Al-Qadisiyah for Computer Science and Mathematics مجلة القادسية لعلوم الحاسوب والرياضيات SN - 20740204 25213504 AB - Opinion mining from reviews is a very crucial area in NLP. This area has many applications in social networks, business intelligence, and decision making. Aspect extraction is the main step to achieve opinion mining. This paper proposed an algorithm for aspect extraction from reviews in the Arabic language, to determine the aspects that the reviewers are described in their comments. The proposed algorithm begins with analyzing the comments dataset using latent Dirichlet analysis (LDA) to identify the aspects and itsessential representative words, then extracting nouns and its' adjectives as a possible aspect phrase in a review. After that the categorizing process to categorize the extractedphrases according to the words specified from LDA analysis. The proposed approach has been tested by using two standard Arabic reviews datasets. The result was auspicious inspite of the difficulties if the Arabic natural language processing.

ER -