Opinion Mining

  • English
  • Italiano

Sentiment analysis is the discipline that studies natural language text in order to detect and recognize the presence of expressed opinions, emotions, sentiments, and other cognitive states of a subjective nature. In the last ten years sentiment analysis has grown from a small niche concern to being one of the most actively investigated subdisciplines of natural language processing.

The reason of this success is its importance within the broader field of opinion mining, a subfield of text mining where the main goal is extracting useful knowledge from large quantities of sentiment-laden texts. Opinion mining has come to play a key role in customer relationship management, consumer attitude detection, brand / product positioning and management,  market research, and the social sciences in general. Interest in these applications has spawned a new generation of companies and products devoted to online reputation management, market perception, and online content monitoring in general.

Within opinion mining, sentiment analysis is mostly used at the aggregate level rather than at the individual level, since in most applications the interest is in gauging collective (rather than individual) sentiment. This is determining a gradual shift of focus from the traditional paradigm of "classification" to the new paradigm of "quantification", where the final goal is estimating sentiment prevalence rather than in determining the opinion of specific individuals.