• Rafi Ahmad Khan
• Khalid Bashir Bajwa
• Waqas Nawaz
• Qaiser Abbas
• AbdulAziz Almuzaini

Classification and analysis of Arabic text in terms of social media analysis is presented here in this work. This work, keeping in view the morphological and syntactical difficulties in Arabic language, focuses on negative and positive sentiment candidates for polarity as a first step towards a major goal of building a complete framework for political sentiment analysis. There are many works for sentiment classification and analysis in literature, which can be categorized into supervised, unsupervised, and hybrid. However, majority of these methods either lack in availability of sufficient size dataset for learning or suffer from Arabic language complications in terms of processing diverse nature of text. To overcome this issue, we introduce a semi-supervised approach for sentiment classification and analysis in Arabic language. Our approach is based on the concept of word embedding model to improve the performance of classification even with small size seed data. It has the capability to model the words in a large vector space where similar words are expected to occur in close proximity. Once the data is mapped to vector space model, then we utilize various classifiers to learn the patterns in the lexicon and predict the classification as positive or negative for unknown similar words. Classifiers such as Stochastic Gradient Descent and SVM are trained and tested with the specified 80% and 20% data ratio. Our approach yields around 80% accuracy through intermediate experiments. It has been observed that besides the common problem of lexicon size, this is attributed mainly to the quality of word embedding available for the Arabic language.

Sentiment Analysis, Classification, Tweets, Polarity, Arabic Language, Lexicon, Text Processing

"Sentiment Classification and Analysis on Arabic using Semi-Supervised Approach", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.5, Issue 8, page no.233-237, August-2018, Available :http://www.jetir.org/papers/JETIR1808930.pdf

"Sentiment Classification and Analysis on Arabic using Semi-Supervised Approach", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.5, Issue 8, page no. pp233-237, August-2018, Available at : http://www.jetir.org/papers/JETIR1808930.pdf