Available Knowhow
Enables more accurate understanding of reviews and comments
|
Categories |
Data Mining, Web Technologies, Text Mining, Sentiment Analysis |
|
Development Stage |
Proof of concept, initial work completed |
|
Knowhow |
Available Knowhow |
Highlights
-
For many natural language processing systems, such as review summarization and public media analysis, it is often important to understand the sentiment being expressed in very short texts, such as Tweets.
-
This new algorithm accurately identifies the sentiments behind comments even if they are not explicitly expressed.
-
For example, if feedback received just says that something is small, it could be good (in the case of a cellphone, for instance) or bad (a room in a hotel).
-
The algorithm can also accurately identify moods and feelings (bored, stressed, angry, excited) in short messages or comments.
-
Successful demonstration of accuracy using Amazon product reviews.
Our Innovation
Automated, supervised sentiment classification framework that allows identification and classification of diverse sentiments expressed in short texts.
Key Features
-
Avoids the need for labor-intensive manual annotation.
-
Successfully identifies sentiment types of untagged sentences.
The Opportunity
-
May be used to offer mood-appropriate advertising
-
May be used to understand consumers' feelings about brands and products
-
Security applications in gauging changes in crowd moods
-
Application in review analysis
1586128900.png)