Package org.apache.uima.fit.examples.experiment.pos

Class RunExperiment

java.lang.Object
org.apache.uima.fit.examples.experiment.pos.RunExperiment
public class RunExperiment extends Object
This class demonstrates a very common (though simplified) experimental setup in which gold standard data is available for some task and you want to evaluate how well your analysis engine works against that data. Here we are evaluating "BaselineTagger" which is a (ridiculously) simple part-of-speech tagger against the part-of-speech tags found in "src/main/resources/org/apache/uima/fit/examples/pos/sample-gold.txt".

The basic strategy is as follows:

  • post the data "as is" into the default view
  • parse the gold-standard tokens and part-of-speech tags and put the results into another view we will call GOLD_VIEW
  • create another view called SYSTEM_VIEW and copy the text and Token annotations from the GOLD_VIEW into this view
  • Run the BaselineTagger on the SYSTEM_VIEW over the copied Token annoations
  • Evaluate the part-of-speech tags found in the SYSTEM_VIEW with those in the GOLD_VIEW
Please see comments in the code for details on how the UIMA pipeline is set up and run for this task.