Difference between revisions of "Opinion mining"

From Cohen Courses
Jump to navigationJump to search
Line 8: Line 8:
  
 
* Document level
 
* Document level
** [[Turney,2002]] presented an approach to calculate the opinion orientation using the Web as a corpus. The input review is classified based on the average semantic orientation of the phrases in the review. They have used PMI-IR technique to measure the semantic orientation of each phrase in the review.
+
** [[Turney,2002]] [[category::paper]] presented an approach to calculate the opinion orientation using the Web as a corpus. The input review is classified based on the average semantic orientation of the phrases in the review. They have used PMI-IR technique to measure the semantic orientation of each phrase in the review.
  
 
** [[Turney and Littman, 2003]] expanded [[Turney,2002]] work using cosine distance in latent semantic analysis as the distance measure.
 
** [[Turney and Littman, 2003]] expanded [[Turney,2002]] work using cosine distance in latent semantic analysis as the distance measure.

Revision as of 15:44, 1 December 2010

Summary

Opinion mining is a problem in the field of information extraction that which aims to automatically extract opinion expressions from product reviews. Also one of the goal of the opinion mining techniques is to determine the opinion direction of a review.

Common Approaches

Generally there are two approaches for opinion mining: 1- document level and 2- feature level opinion mining.

  • Document level
    • Turney,2002 paper presented an approach to calculate the opinion orientation using the Web as a corpus. The input review is classified based on the average semantic orientation of the phrases in the review. They have used PMI-IR technique to measure the semantic orientation of each phrase in the review.
    • Dave et al.,2003 introduced a novel approach to classify reviews in Amazon.com using normalized term frequency on uni-gram, bi-gram and tri-gram.
  • Feature level

Some common models for named entity recognition include the following:

  • Lexicons
    • Checks if a token is part of a predefined set
  • Classifying pre-segmented candidates
    • Manually select candidates, then use YFCL on a piece of text to deterimine what type of entity it is
  • Sliding Window
    • Try all reasonable token windows (different lengths and positions), train a Naive Bayes classifier or YFCL, then extract text if Pr(class=+|prefix, contents, suffix) > some threshold
  • Token Tagging / Sequential

Example Systems

References / Links

  • BBN Named Entity Types - [1]
  • Satoshi Sekine's Extended Named Entity Hierarchy - [2]
  • Wikipedia page on Named entity recognition - [3]

Relevant Papers