This repository contains the SenticVec knowledge base described in SenticVec: Toward Robust and Human-Centric Neurosymbolic Sentiment Analysis.
To download SenticVec, please click here.
In SenticVec, the sentiment-pragmatic representation of each lexical unit is formulated as a 19-dimensional vector. SenticVec is currently available in English, Spanish, French, and Italian. To use the knowledge base, simply load the corresponding json file as dictionary. Please refer to the paper above for ways it could be used.
If you use this knowledge base in your work, please cite the paper - SenticVec: Toward Robust and Human-Centric Neurosymbolic Sentiment Analysis with the following:
@article{zhang2024senticvec,
title={SenticVec: Toward Robust and Human-Centric Neurosymbolic Sentiment Analysis},
author={Zhang, Xulang and Mao, Rui and Cambria, Erik}
conference={Proceedings of ACL}
year={2024}