Skip to content

Latest commit

 

History

History
13 lines (7 loc) · 1.21 KB

File metadata and controls

13 lines (7 loc) · 1.21 KB

Neuro-Symbolic Sentiment Analysis with Dynamic Word Sense Disambiguation

This repository contains an implementation of the paper "Neuro-Symbolic Sentiment Analysis with Dynamic Word Sense Disambiguation," written by Xulang Zhang et al. This work was accepted at EMNLP 2023, and you can access the paper here.

Current Status

  • Implementation: Due to resource constraints, I could only perform simple experiments to validate the basic functionality of the implementation. The models have not yet been trained on publicly available datasets, and therefore, pre-trained weights are not available at this time. I plan to train the models on standard datasets as soon as I have access to a proper GPU.

  • Lexical Substitution Task: The implementation uses RoBERTa instead of the pretrained ALM model introduced in the paper "MetaPro 2.0: Computational Metaphor Processing on the Effectiveness of Anomalous Language Modeling," as the latter has not been published yet.

  • Dynamic Rewarding Strategy: The dynamic rewarding strategy proposed in the paper is not implemented yet.

For any questions or feedback, feel free to open an issue in this repository.