Salience 4.0 With Improved Entity Extraction/Sentiment Scoring

Lexalytics, Inc. (, a software and services company specializing in text analytics and social media monitoring, announced today it has released Salience 4.0. This latest version of the popular Salience Engine marks the first major software release for Lexalytics since the merger with Infonic’s Text Analytics Division. The result is significant improvements to entity extraction, sentiment scoring and thematic extraction, as well as several new features.

Salience technology is widely used by professionals in the marketing, public relations, enterprise search and financial services industries to manage brand reputation and improve the business value of information. The Salience 4.0 software is the first in the industry to deliver a new inferred entity engine which will allow companies to identify products and brands from within text without the use of massive product lists or dictionaries.

Salience 4.0 also allows for better entity recognition, with initial test results showing up-to 90 percent accuracy for people and over 80 percent for companies. Improvements also include the ability to detect entity relationships. For example, “John Smith is CEO of XYZ Company which is located in Amherst, MA”, would show relationships between John Smith, his company, the location and his job title.

Salience 4.0 includes two new enhancements to sentiment scoring: adjective amplifiers and co-reference. Adjective amplifiers solve problems in text such as “perfectly awful,” where the two descriptors would normally negate each other for sentiment. The co-reference enhancement will draw correlations between a person and a pronoun such as, “John Smith is being promoted. He has taken a new position as CEO.” It will recognize that “he” is “John Smith.” Lastly, Salience 4.0 has dramatically improved the theme extraction capabilities and will return longer themes or concepts from a document with more accuracy to the subject matter, and can role those themes up to a parent theme more easily.

The result is delivery of data to the user that immediately provides relevant context around entities and the themes with which they are associated, arming business leaders with the ability to monitor and react quickly. “Reputation monitoring and opinion tracking are some of the hottest areas today in software applications,” said Sue Feldman, Vice President, Search and Digital Marketplace Technologies, IDC. “Firms need to find out what is being said in the media and on the Web about their brands or their executives. But the accuracy is often not as high as companies need. The enhancements Lexalytics incorporated into Salience 4.0 improves accuracy and allows companies to understand opinions not just on the sentence, but on the entire document or blog level. For adoption of sentiment and text analytics to continue to grow, it is imperative to develop the high level of accuracy that companies demand.”

“Our goal when we began the development cycle for Salience 4.0 was to build off of the strong architecture that already existed and to enhance it to be more accurate and robust. We also included new features that we believe addressed some of the analytic needs in the marketplace that have been overlooked until now,” said Mike Marshall, CTO of Lexalytics, Inc. “Ultimately, with the merger with Infonic’s Text Analytics Division, we were able to draw upon certain strengths from both technologies and produce a version that exceeds those development goals. I’m confident we are the leaders in the text analytics space, particularly when it comes to sentiment.”