Feature Extraction using WEKA & UIMA

This project extracts features around the "it" pronoun from text. The features can then be fed to WEKA binary classification models to predict whether a specific instance of "it" is clause (referring to a clause) or nominal (referring to a specific noun phrase) anaphoric.

Setup and execution

1. This project was built and run using Java™ SE Runtime Environment (build 13+33)
2. Build pom.xml using Maven to get all dependencies.
3. Run main function in src/main/java/ca/uottawa/csi5137b/pipelines/FeatureExtractionPipeline.java
4. The output file will be placed in the same io folder along with the input.txt.

Paths

Output:

src/main/resources/io/output.csv

Input:

Without labels, both classes merged into one file after removing duplicates and cleanup: src/main/resources/io/input.txt

Split by label: src/main/resources/data/clause.txt and src/main/resources/data/nominal.txt

Acknowledgment

CSI 5137B Course at U Ottawa