Home

Web Demo

Paste a block of text in the window below and click the "Process Text" button.  Note that these results are the "absolute worst case" results - the Lexalytics Salience Engine allows for a tremendous amount of configuration and tuning; none of which has been applied here - this is a completely stock "out of the box" installation. 

You'll get good results from this demo - but your results in your installation will be even better, and can be completely differentiated for your application.

  • Document Sentiment:  The overall tone of the document 
  • Entities:  Includes people, places, dates, companies and other entity types; with the number of times that entity appeared in the text as well as the sentiment for that entity in this text. 
  • Summary:  The most important sentences in the document 
  • Entites and Sentiment Phrases:  A marked up version of the text showing positive and negative phrases as well as the entities that appear in the text.
  • Themes:  The most important concepts in the document, ordered by the importance of that theme to the text.
  • Concept Topics:  These are classifiers based on wikipedia content.   Here's the configuration of the classifiers that this demo knows about - so, you can type something like "I like chicken.".  That will be classified as food, and you can see from the configuration file that there's no "chicken" in the food Concept Topic.

Web Demo FAQ

At a document level, you`ll see the overall sentiment for the document and a highlighted version of the text. An important consideration is than often times the sentiment is negated for a particular phrase - so if you see something positive that you think should be negative, then look for the negation phrases. Also, remember that this is a statistical process. It`s almost certain that you`ll find a phrase or two that you disagree with - what is important is that numbers work out in the correct direction.

In addition, entity sentiment is generally more interesting than document sentiment for longer documents. The shorter the document, the more they`ll tend to converge.