This repository curates a comprehensive collection of research papers on neural-symbolic Artificial Intelligence, with a strong emphasis on Natural Language Processing (NLP) applications. It serves as a valuable resource for researchers and engineers seeking to integrate symbolic logic and reasoning capabilities into deep neural networks. The collection highlights advancements in areas like knowledge regularization, weak supervision, and inductive biases, aiming to bridge the gap between data-driven deep learning and knowledge-driven symbolic AI for more robust and interpretable AI systems.

How It Works

The collection showcases diverse methodologies for integrating logic into deep learning. Key approaches include using logic as a form of knowledge regularization to enhance model performance and consistency, employing logic as weak supervision for tasks like named entity recognition, and incorporating logical inductive biases into neural network architectures. Papers explore graph neural networks, relational learning, and specialized "Logical Neural Networks" to imbue models with structured reasoning capabilities, enabling them to handle complex relational data and perform more sophisticated inference.

Quick Start & Requirements

This repository is a curated list of research papers and associated resources, not a software project with installation instructions. Users are directed to individual paper links for specific implementation details, code, and requirements.

Highlighted Details

• Features papers on "LogicBench," a systematic evaluation framework for LLM logical reasoning abilities.
• Includes research on enhancing zero-shot Chain-of-Thought reasoning in LLMs through logic.
• Covers applications such as clinical temporal relation extraction, structured semantic role labeling, and fine-grained propaganda detection.
• Highlights work on integrating first-order logic with transformers and explainable recommendation systems via neural logic reasoning.

Maintenance & Community

Information regarding project maintenance, contributors, community channels (e.g., Discord, Slack), or roadmaps is not provided in the README.

Licensing & Compatibility

The README does not specify a software license for the collection itself, nor does it detail licensing or compatibility for the individual research papers or their associated code.

Limitations & Caveats

As a curated list, this repository does not offer a unified framework or codebase. Access to runnable code and specific experimental setups depends entirely on the availability and completeness of the linked research papers. The breadth of topics covered means that practical implementation details vary significantly across the included resources.