A Review on Neuro-symbolic AI Improvements to Natural Language Processing

Symbolic artificial intelligence (AI) reflects the domain knowledge of experts and adheres to the logic of the subject area, rules, or any relations between entities. Connectionist (neuro) approaches based on artificial neural networks are excellent for extracting abstract features, contextualizing, and embedding interactions between features. When connectionist and symbolic approaches are properly aligned in a model, they benefit from complementary strengths; the combination is referred to as a hybrid or neuro-symbolic artificial intelligence (NSAI) model. The advantages that NSAI brings to the field of natural language processing (NLP) have received little attention from researchers in recent years. Therefore, in this review, we focus on the impact of neuro-symbolic approaches for NLP tasks, i.e. text classification, information extraction, machine translation, and language understanding. Relevant research articles from Scopus, Web of Science, and Google Scholar were carefully examined using appropriate keywords in the period from 2019 to 2024. The review aims to show the types of NSAI systems, identify the motivation for using NSAI, evaluate the use of additional annotations for content description, and briefly describe how the neuro-symbolic connection improves the methodology and enables trustworthy and explainable AI systems in current NLP research. The review also highlights areas of application and improvements achieved by NSAI approaches in benchmarks.