OpenCalais classifier
| Line 12: | Line 12: | ||
== OpenCalais classification example == | == OpenCalais classification example == | ||
| − | In our example user has started Wandora application and Wandora embedded HTTP server. Then user browses The New York Times web site and finds interesting text fragment discussing Senator Obama's speech. User selects the text fragment as shown below and starts Wandora Firefox plugin with option Classify with OpenCalais. | + | In our example user has started Wandora application and [[Embedded HTTP server|Wandora embedded HTTP server]]. Then user browses The New York Times web site and finds interesting text fragment discussing Senator Obama's speech. User selects the text fragment as shown below and starts Wandora Firefox plugin with option Classify with OpenCalais. |
Revision as of 15:00, 6 September 2008
|
|
OpenCalais classifier takes a text fragment and creates a topic for it, and associates created text topic with keywords given by OpenCalais web service. The effect is that given text is tagged with certain topics found in the text. Wandora's OpenCalais classifier starts with File > Extract > OpenCalais classifier.... Classifier can read given file, URL, or plain text. Wandora's Firefox plugin can also access the OpenCalais classifier. You can send complete WWW page or just WWW page fragments to Wandora for OpenCalais classification while browsing the web. |
OpenCalais classification example
In our example user has started Wandora application and Wandora embedded HTTP server. Then user browses The New York Times web site and finds interesting text fragment discussing Senator Obama's speech. User selects the text fragment as shown below and starts Wandora Firefox plugin with option Classify with OpenCalais.
Plugin sends the text to Wandora's OpenCalais classifier. Classifier passes the text to OpenCalais web service and receives keywords found in the text. Finally Wandora creates topics for
- The text fragment
- WWW page the text fragment originates
- Each keyword provided by the OpenCalais
and links everything with reasonable associations. Below is a screenshot of Wandora after successful classification. It appears OpenCalais recognized keywords Obama and McCain in the given text fragment.
Additional notes
- The quality of keywords provided by OpenCalais varies. Sometimes OpenCalais doesn't recognize obvious keywords in given text. Sometimes provided keyword set is very comprehensive.
- OpenCalais classifier accepts only limited size texts.
- OpenCalais classifies properly only English language texts.
