BUILDING A LARGE SCALE ARABIC SENTIMENT ANALYSIS
Afnan Atiah Alsolamy · 2018
The dramatic amount of posts expressing Arabic opinion on the Internet has raised a huge need for sentiment analysis. Sentiment analysis is a challenging task whose accuracy depends on the availability of a number of resources and tools, including sentiment lexicon. This work builds an Arabic sentiment lexicon using a corpus-based approach. The lexicon has been built using different similarity measurements and propagation method. The use of similarity measurements depends on the fact that the words are appearing in the same context will have similar polarity. Also, propagation method helps to distribute a polarity value between lexicon entries. Cosine similarity has outperformed other similarities in building the Arabic lexicon.