Improving Anaphora Resolution Through Corpus Mined Gender Information

The whole thesis is available here

Anaphora Resolution describes the process of finding the most recent preceding phrase regarding to the same entity in a text and is needed in multiple Natural Language Processing (NLP) tasks (for instance Information extraction). Besides contextual information, a frequency distribution of gender information http://www.clsp.jhu.edu/~sbergsma/Gender/Data/ was used to increase the performance.

Installation

git clone https://github.com/HenryvanderVegte/henryvdv.BA
cd henryvdv.BA/de.unidue.henryvdv.ba

Import as maven project. Note that this project uses the DKPro Core software components.