Logic programming is becoming a core symbolic component in neurosymbolic AI systems. This survey focuses on three dominant rule‑based languages—Datalog, Answer Set Programming (ASP), and Probabilistic Logic Programs (PLP)—and evaluates them along four axes: semantics, expressiveness, neural integration, and evaluation mechanisms. Over 50 recent systems are examined, with usage statistics broken down across four research domains: databases & programming languages, machine learning, vision, and robotics. A decision matrix is provided to map application scenarios (e.g., deterministic reasoning, probabilistic inference, interpretability) to the required language features, enabling practitioners to select the most suitable formalism. The paper concludes by highlighting open challenges such as unified learning‑inference interfaces, probabilistic modeling of multimodal data, and the lack of large‑scale evaluation benchmarks.
Review