Skip to content

A neural-symbolic joint reasoning approach for Natural Language Inference (NLI). Modeling NLI as inference path planning through a search engine. Sequence chunking and neural paraphrase detection for syntactic variation. SOTA result on SICK and MED.

License

Notifications You must be signed in to change notification settings

eric11eca/NeuralLog

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

NeuralLog: Natural Language Inference with Joint Neural and Logical Reasoning

PWC PWC

Deep learning (DL) based language models achieve high performance on various benchmarks for Natural Language Inference (NLI). And at this time, symbolic approaches to NLI are receiving less attention. Both approaches (symbolic and DL) have their advantages and weaknesses. However, currently, no method combines them in a system to solve the task of NLI. To merge symbolic and deep learning methods, we propose an inference framework called NeuralLog, which utilizes both a monotonicity-based logical inference engine and a neural network language model for phrase alignment. Our framework models the NLI task as a classic search problem and uses the beam search algorithm to search for optimal inference paths. Experiments show that our joint logic and neural inference system improves accuracy on the NLI task and can achieve state-of-art accuracy on the SICK and MED datasets.

The following publications are integrated in this framework:

Installation

The recoomanded environment include Python 3.6 or higher , Stanza v1.2.0 or higher, and **ImageMagick v7.0.11. The code does not work with Python 2.7.

Clone the repository

git clone https://github.com/eric11eca/NeuralLog.git

Getting Started

First download a pretrained model from Google Drive. Replace the Stanza defalut depparse model with this pretrained version. The Stanza model path is:

C:\Users\$your_user_name$\stanza_resources\en\

Then open UdeoLog.ipynb

Pre-Trained UD Parser Models

We provide two UD Parser Models for English. Some models are general purpose models, while others produce embeddings for specific use cases. Pre-trained models can be loaded by just passing the model name: SentenceTransformer('model_name').

Training

For training new UD parser models, see Stanza's training dcumentation for an introduction how to train your own UD parser.

Citing & Authors

If you find this repository helpful, feel free to cite our publication NeuralLog: Natural Language Inference with Joint Neural and Logical Reasoning:

@misc{chen2021neurallog,
      title={NeuralLog: Natural Language Inference with Joint Neural and Logical Reasoning}, 
      author={Zeming Chen and Qiyue Gao and Lawrence S. Moss},
      year={2021},
      eprint={2105.14167},
      archivePrefix={arXiv},
      primaryClass={cs.CL}
}

Contact person: Zeming Chen, chenz16@rose-hulman.edu Don't hesitate to send us an e-mail or report an issue, if something is broken or if you have further questions.

This repository contains experimental software and is published for the sole purpose of giving additional background details on the respective publication.

About

A neural-symbolic joint reasoning approach for Natural Language Inference (NLI). Modeling NLI as inference path planning through a search engine. Sequence chunking and neural paraphrase detection for syntactic variation. SOTA result on SICK and MED.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published