Financial Sentiment Analyzer using GATE - gsi-upm/SAGA GitHub Wiki

Welcome to this step by step example of how to create a Spanish Financial Sentiment Analyzer application using GATE Developer, without touching a line of code, just applying what you already know about GATE.

  1. Clone or download SAGA repository into your system.
  2. Open GATE Developer application. My version is 7.1.
  3. Now we are going to load into GATE the plugins we need:
    • Go to menu File -> Manage CREOLE Plugins...
    • A list of Installed Plugins is shown.
    • Click on + button -> Select a directory -> Go to the downloaded SAGA repository -> gate.sa.modules -> bin -> Single click on processingResources -> Choose -> OK
    • A new folder of plugins called processingResources should appear in the list of installed plugins.
    • Mark processingResources and ANNIE plugins as Load Now -> Apply all
  4. Go to the left column -> Right click on Processing Resources:
    • Select Document Reset PR -> Name = Delete PR -> OK
    • Select ANNIE English Tokenizer -> Name = Tokenizator -> OK
    • Select ANNIE English Gazetteer ->
      • Name = Gazetteer
      • listsURL -> Select a directory -> Go to the downloaded SAGA repository -> gate.sa.modules -> src -> resources -> gazetteer -> finances -> lists.def -> Choose -> OK
    • Select ANNIE NE Transducer ->
      • Name = NE Transducer
      • gramarURL -> Select a directory -> Go to the downloaded SAGA repository -> gate.sa.modules -> src -> resources -> jape -> main.jape -> Choose -> OK
    • Select Text Value and Polarity Generator PR.
  5. Go to the left column -> Right click on Applications -> New -> Create New Applicaiton -> Corpus Pipeline -> Name = Example -> OK
  6. Double click the application -> Add them in te following order:
    • Document Reset PR
    • Tokenizator -> and set the annotationSetName as Annotations.
    • Gazetter -> and set the annotationSetName as Annotations.
    • NE Transducer -> set the inputASName as Annotations and set the opuputASName as Sentiment (it is really important that the output annotation set is called Sentiment in order to the next processing resource to work)
    • Text Value and Polarity Generator PR.
  7. Add the example corpus from gate.sa.modules/src/resources/data/input/
  8. Run it!