UClassify integration

From WandoraWiki
Jump to: navigation, search

UClassify is a free text classifier web service. With the web service user can not only use preset classifiers but also teach a text classifier her self. Preset classifiers include language detection, sentiment detection, topic classifier etc. Wandora's uClassify integration enables Wandora user to perform uClassify classifications directly in Wandora application. Wandora application sends user specified text to the web service of uClassify, receives some data and transforms that data to topics and associations. Wandora's uClassifier starts with a menu option File > Extract > Classify > uClassify. uClassification requires a valid api key from uClassify. Wandora user can request her api key here.

By default Wandora integration of uClassify support classifiers:

  • Text Language by uClassify
  • Sentiment by uClassify
  • Topics by uClassify
  • Mood by prfekt
  • GenderAnalyzer_v5 by uClassify
  • Ageanalyzer by uClassify
  • classics by uClassify
  • Society Topics by uClassify
  • News Classifier by mvazquez
  • Tonality by prfekt
  • Myers Briggs Judging Function by prfekt
  • Myers Briggs Lifestyle by prfekt
  • Myers Briggs Attitude by prfekt
  • Home Topics by uClassify
  • Art Topics by uClassify
  • Game Topics by uClassify
  • Health Topics by uClassify
  • Computer Topics by uClassify
  • Business Topics by uClassify
  • Recreation Topics by uClassify
  • Sport Topics by uClassify
  • Science Topics by uClassify

Wandora user can change the classifier set by configuring the uClassify tool. Configuring dialog opens when user starts uClassifier while holding CTRL key down. In configuration dialog, Wandora user can also forget the used api key. Wandora doesn't store api key between use sessions. Below is a window capture of configuration dialog window.

Uclassify conf.gif


In this example Wandora user first starts uClassify classifier.

Uclassify 01.gif

User inputs Wandora application descriptions to the raw tab. User selects Text language classifier and threshold value 0.01. Threshold value is used to filter out weak classifications. It is usual that uClassify returns also classes with a very small probability. Wandora user might want to filter out such classes.

Uclassify 02.gif

After user presses Classify button, Wandora either asks api key or performs classification depending on the current api key status. After classification Wandora user should see some additional topics in topic tree. Classified text can be found under Document and all classifier topics under uClassify. There is an association with four members for each classification received. Associations has a player for

  • Classified text with role Topic.
  • Classification probability with role uClassify probability.
  • Classification class with role uClassify class.
  • Used classifier with role uClassify classifier.

Uclassify 03.gif

In our example Wandora user mouse clicks next the occurrence document-text. An occurrence editor dialog opens. User selects menu option Refine > uClassify > uClassify Topics. Wandora performs classifications instantly using the api key entered earlier.

Uclassify 04.gif

After closing the occurrence editor dialog window Wandora user sees yet another 4-ary association that classifies current topic with a class Computers.

Uclassify 05.gif

See also

Personal tools