Neuro-Symbolic Artificial Intelligence: A Task-Directed Survey in the Black-Box Models Era

TL;DR

This survey examines task-specific advancements in the NeSy domain to explore how incorporating symbolic systems can enhance explainability and reasoning capabilities.

Abstract

The integration of symbolic computing with neural networks has intrigued researchers since the first theorizations of Artificial intelligence (AI). The ability of Neuro-Symbolic (NeSy) methods to infer or exploit behavioral schema has been widely considered as one of the possible proxies for human-level intelligence. However, the limited semantic generalizability and the challenges in declining complex domains with pre-defined patterns and rules hinder their practical implementation in real-world scenarios. The unprecedented results achieved by connectionist systems since the last AI breakthrough in 2017 have raised questions about the competitiveness of NeSy solutions, with particular emphasis on the Natural Language Processing and Computer Vision fields. This survey examines task-specific advancements in the NeSy domain to explore how incorporating symbolic systems can enhance explainability and reasoning capabilities. Our findings are meant to serve as a resource for researchers exploring explainable NeSy methodologies for real-life tasks and applications. Reproducibility details and in-depth comments on each surveyed research work are made available at https://github.com/disi-unibo-nlp/task-oriented-neuro-symbolic.git.

Figures (3)

• Figure 1: Distribution of NeSy peer-reviewed papers over the period 2017-2024. The innermost ring delineates the inclusion criteria. Greyscale slices denote studies excluded for relying on black‑box methods, exclusively logical approaches, or brevity, while the colored slice comprises the surveyed ones. The outermost ring represents the number of research works from each selected venue, with exact counts reported in the legend. The number of papers considered for each track is enclosed in parentheses.
• Figure 2: Task-Directed NeSy Taxonomy. We organize the most relevant task families by classifying each according to applicable NeSy techniques and grouping them into three macro-categories: (1) Rule Mining, (2) Rule Enforcement, and (3) Program Synthesis. We also label the most commonly used datasets to highlight their role in real-world benchmarking and model evaluation. Tasks are organized from bottom to top and from left to right within each category, mirroring the survey's conceptual progression.
• Figure 3: Overview of the intermediate formal languages employed by each taxonomy method.