بناء موارد أساسية لتحليل الآراء و المشاعر الواردة في التعليقات باللغة العربية

Abstract EN

This paper aims at highlighting the procedures of building some linguistic resources that would help to make and assess software and applications for opinion and sentiment analysis.

Among the resources that we have built are a specialized dictionary that includes opinions and sentiments lexicon, a list of words and prepositions that orient the Arabic discourse, and the corpus adopted during the assessment of ‘opinion mining’ and ‘sentiment analysis’ software.

And in order to build these linguistic resources, we basically used the descriptive approach so as to provide the tools required for opinion and sentiment analysis.

We also adopted the manual and automated approach for preparing the specialized dictionary, the list of words and prepositions, and the corpus.

So we were able to create a dictionary which is containing 21453 lexical entries, prepare a list which contains 9784 words and prepositions, and collect about 100630 comments which have 25% of manually tagging.