Please use this identifier to cite or link to this item:
http://hdl.handle.net/20.500.12358/25202
Title | Arabic opinion mining using combined classification approach |
---|---|
Untitled | |
Abstract |
In this paper, we present a combined approach that automatically extracts opinions from Arabic documents. Most research efforts in the area of opinion mining deal with English texts and little work with Arabic text. Unlike English, from our experiments, we found that using only one method on Arabic opinioned documents produce a poor performance. So, we used a combined approach that consists of three methods. At the beginning, lexicon based method is used to classify as much documents as possible. The resultant classified documents used as training set for maximum entropy method which subsequently classifies some other documents. Finally, k-nearest method used the classified documents from lexicon based method and maximum entropy as training set and classifies the rest of the documents. Our experiments showed that in average, the accuracy moved (almost) from 50% when using only lexicon based method to 60% when used lexicon based method and maximum entropy together, to 80% when using the three combined methods. |
Authors | |
Type | Journal Article |
Date | 2011 |
Publisher | Naif Arab University for Security Sciences |
Citation | |
License | ![]() |
Collections | |
Files in this item | ||
---|---|---|
Arabic Opinion Mining Using Combined Classification Approach.pdf | 158.8Kb |