×

LEARNING OPINION-RELATED PATTERNS FOR CONTEXTUAL AND DOMAIN-DEPENDENT OPINION DETECTION

  • US 20140067370A1
  • Filed: 08/31/2012
  • Published: 03/06/2014
  • Est. Priority Date: 08/31/2012
  • Status: Abandoned Application
First Claim
Patent Images

1. A method for extracting opinion-related patterns, comprising:

  • receiving a corpus of reviews, each of the reviews in the corpus including an explicit rating of a topic;

    partitioning at least a portion of the reviews among a predefined plurality of classes, based on the explicit ranking;

    identifying syntactic relations in a text portion of each of the reviews in the portion of the reviews, each of the identified syntactic relations including a first term comprising an adjective and a second term comprising a noun, the adjective serving as a modifier or attribute of the respective noun;

    generating a set of patterns, each of the patterns having at least one of the identified syntactic relations as an instance;

    with a processor, clustering the patterns into a set of clusters based on a set of features, at least one of the features in the set of features being based on occurrences, in the predefined classes, of the instances of the patterns;

    selecting a subset of the clusters and assigning a polarity to patterns in the respective clusters in the subset.

View all claims
  • 1 Assignment
Timeline View
Assignment View
    ×
    ×