An automatic RST parser implemented using neural networks. This team-project is done as part of the Masters course ANLP at the University of Potsdam. The parser is trained and tested on wall-street-journal corpus.
The whole project is divided into 2 sections :
- Segmentation
- Parsing
Files :
- RST-00-corpus-reader : Contains Feature extractions for the Segmentor Neural Network
- RST-01-EDU-segmenter : Segmentor implementation and results.
- RST-02-parser : Feature extraction for Parsing and Parser results.
- rst.py : Parser implementation
The guide to use each notebook is provided in the notebook itself.