Sentiment Analysis, the New Search and Retrieval Tool
March 24, 2011
Computer World has a concept for us to ponder: “Sentiment Analysis Comes of Age.” Sentiment analysis has been around for some time, but with the rapid proliferation of up-to-date social media and semantic-software vendors that offer commentary and relevant components, its benefits are coming to the light. Sentiment analysis has many potential paybacks, including new data sources never before tapped. Many semantic platforms already analyze material from social networking sites.
“Sentiment analysis platforms use two main methodologies. One involves a statistical or model-based approach wherein the system learns to assess sentiment by analyzing large quantities of pre-scored material. The other method utilizes a large dictionary of pre-scored phrases.”
Computerworld discovers sentiment analysis the way a father realizes his 15 year old daughter is ageing quickly. Can sentiment analysis improve search and retrieval? We’re not sure, but it makes advertisers perceive an advantage in explaining why consumers dislike a product or brand.
We anticipate search vendors will pile into this market as well.
Whitney Grace, March 24, 2011
Freebie
Comments
One Response to “Sentiment Analysis, the New Search and Retrieval Tool”
-
- Subscribe to Beyond Search
Feature archive
News archive
-
Whitney, that Computer World statement about methodologies is a bit simplistic: “The other method utilizes a large dictionary of pre-scored phrases.” First, a better tool might have several dictionaries (or more complex lexical structures such as taxonomies or ontologies), suited to different business domains. Think of “grass” in the contexts of a hardware store and of law enforcement: different senses. Secondly, a tool that uses a linguistic approach may also have a set of language rules — patterns — to handle matters such as negation and, if the tool’s really good, co-reference including anaphora.
Anyway, folks (including Boss Man Stephen?) can learn more on the topic at the Sentiment Analysis symposium, April 11 (tutorial) and April 12 (symposium) in NY, http://sentimentsymposium.com .