Opinion Based Word Extraction of Supervised & Unsupervised Models for Alignment |
Author(s): |
| Shalaka Dinkar Gaidhani , MET BKC, Adgaon, Nashik; Dr. M. U. Kharat, MET BKC , Adgaon, Nashik |
Keywords: |
| Opinion, targets, opinion words, alignment models |
Abstract |
|
In today’s electronic era, consumers are often forced to go through many on-line reviews in order to make product choice. The Web provides ocean of opinions about products. So, the user needs to do proper analysis of the reviews, which is done by aligning the reviews efficiently. Previously, alignment was carried out using syntactic pattern matching in an unsupervised environment. Moreover, this alignment did not prove to be efficient and the alignments which were made propagated error. So, to overcome such supervised word alignment model is introduced. Opinion relations are identified as an alignment process. In order to estimate the confidence of each candidate, a graph-based co-ranking algorithm is exploited. Finally, candidates with higher confidence are extracted as opinion targets or opinion words. |
Other Details |
|
Paper ID: IJSRDV4I50841 Published in: Volume : 4, Issue : 5 Publication Date: 01/08/2016 Page(s): 1641-1644 |
Article Preview |
|
|
|
|
