ABSTRACT. In this presentation we will expand upon the claim that logic programming (LP) is especially well suited for teaching logical thinking (LT) because it is simpler than other logics and in many respects more powerful. We will also present three ways in which LP can be made more accessible than conventional academic logics: (i) by employing a controlled natural language syntax, such as Logical English, (ii) by treating logical thinking and problem solving as an argumentation game, and (iii) by animating change of state in such extensions of LP as LPS (logic-based production systems). Arguably, these features help to close the gap between STEM and non-STEM subject areas, and make STEM subjects more accessible to a wider range of students.

ABSTRACT. In the winter semester 2025/26, I began to explore the question of to what extent a Bento Box could serve as a useful mental model for providing art and design students in my introductory course of logic programming with an accessible entry point and for enabling an intuitive understanding of its fundamental concepts in relation to concepts of composing an aesthetic artefact. I started to examine the Bento Box with regard to its capacity to embody these qualities and made a series of interesting discoveries, which I discuss in this paper.

ABSTRACT. This article examines the role of symbolic artificial intelligence in preserving and operationalising linguistic and cultural diversity under low-resource conditions. It compares a Prolog-based intelligent system for the Rhodope dialects of Bulgarian with two broader symbolic models, Womb Grammar and Doughnut Computing. The paper argues that these approaches are complementary rather than competing: the Rhodope system captures fine-grained lexical, morphological and phonetic particularities, while Womb Grammar supports translingual grammar induction and Doughnut Computing extends constraint-based reasoning toward socially and ecologically oriented planning. By combining related work, a theoretical framework and a comparative analysis, the study shows that symbolic AI remains especially valuable where interpretability, modularity, community participation and data scarcity are decisive constraints.

ABSTRACT. Game-based learning (GBL) is a common approach to teach computational thinking skills to children, as it increases engagement and student motivation. In the literature, the video game Minecraft is steadily gaining popularity as a teaching environment in this context. In this paper, we look at how Minecraft can be used to teach logical thinking instead. We outline a few ideas for pedagogical applications within Minecraft, and discuss two of these in detail. In general, we think Minecraft forms an excellent environment for teaching logic, logic programming, and logical thinking, warranting future research on the development of pedagogical material.

ABSTRACT. This paper describes the logic training workspace "IMPLY - interactive models for pre-literate youth". IMPLY aims at pre-literate children by using images as literals for concepts or events in a small game universe. The literals are used in general propositional Horn clauses in implication form to describe the valid models, or worlds, of this specific game universe. Multiple game modes allow the exploration of the implication as a fundamental building block of logic programming.

ABSTRACT. This paper presents a methodological approach for introducing core concepts of Logic Programming through a gamified and communicative activity based on encrypting and decrypting messages. Students take on the role of secret agents who exchange ciphered messages with unknown partners, using Prolog knowledge bases to construct, apply, and interpret simple ciphers. The proposed didactic sequence connects the learning of facts, rules, lists, and recursion with a concrete practical task: designing a cypher, encrypting a message, and decrypting a received one. The methodology is organised as a series of progressively more complex learning moments, in which each new concept emerges naturally as a response to a communicative or technical need. The paper also discusses LP education initiatives in Bulgaria and Argentina and reports on a preliminary international experience involving students from both countries. Initial observations suggest that combining Prolog, cryptography, role-based interaction, and cross-cultural communication can foster logical thinking, motivation and collaboration in school education.

ABSTRACT. This paper presents a web-based platform for symbolic artificial intelligence, designed for both knowledge creators and end users. It has been developed in response to the growing interest in explainable and transparent AI, offering an alternative to the limitations of contemporary statistical approaches and large language models. Unlike these approaches, symbolic AI relies on the explicit representation of knowledge through facts and rules, combined with deterministic mechanisms for logical inference. The platform employs SWI-Prolog as its core inference engine, integrated with a server that manages user sessions and facilitates interaction with the logic layer. The architecture is multi-layered and modular, comprising a web interface, a business logic layer, and a data layer with domain-specific knowledge bases. The system supports multiple user roles—administrator, teacher, student, and knowledge user—allowing both the creation and the use of expert systems. From an educational perspective, the platform serves as an environment for learning artificial intelligence and logic programming through project-based and constructivist approaches. Students develop Prolog programs, while teachers validate and integrate the resulting knowledge bases. The platform encourages teamwork, international collaboration, and the development of analytical and logical thinking skills. The paper provides an overview of the IDEAS platform and its application in building artificial intelligence systems, where users are not just consumers, but active creators of intelligent solutions.

ABSTRACT. Ninety-Nine Prolog Problems (P-99) is a famous set of Prolog exercises. We solved the first thirty three just by prompting an LLM (Large Language Model). We used Claude from Anthropic. By ``solved'' we mean: generate the Prolog code and a test file, run the tests and check whether they pass, then formally prove types, groundness, termination, uniqueness, existence and also sometimes functional correctness with LPTP (Logic Program Theorem prover). Hence our approach is an experiment in vibe-coding/vericoding of P-99. It is a vibe-coding experiment because we started from informal specifications written in English and let Claude generate the Prolog code. It also fits within vericoding because the LLM proved reliability guarantees on the generated Prolog code. Claude wrote 58 logic procedures, 508 tests, 257 lemmas for a total of 11800 proof lines. We manually checked each file generated by the LLM. We checked the Prolog code, ran the tests, examined the logical statements generated by Claude and proof-checked Claude's proofs with LPTP. This paper describes this experiment and provides the main details so that it can be reproduced by the interested reader. It also includes some pedagogical considerations for teaching certified logic programming with the help of an LLM.

ABSTRACT. Learning to program is challenging regardless of paradigm. Logic programming offers a particularly natural approach: the programmer describes \emph{what} should hold rather than \emph{how} to compute it. But like any skill, it requires practice with concrete feedback. This paper presents the \emph{Logical Robots} platform~\cite{Skvortsov2026LogicalRobots} as a tool for teaching Logica, an open-source logic programming language that compiles to SQL. The platform comes with a combined view of the logic program and the visual simulation environment where robots act. Users can start, pause, and restart simulations as well as drag robots within the environment to create debugging scenarios interactively. In Logical Robots, autonomous agents in this simulated world are controlled by logical predicates that determine their immediate reaction to sensory data. By starting the simulation, the student can immediately see how logical rules drive the robot's behavior. Learning to program through this feedback loop of writing a rule and seeing the robot act, is yet another way an instructor can make learning logic programming more engaging.

ABSTRACT. Logical English (LE) is a controlled natural language for expressing legal and business rules in a form that is at once human-readable and executable. We compare two implementations of the language: the original system (LE1), developed at Imperial College and partner organisations and oriented toward research breadth across multiple reasoning back-ends and several natural-language front-ends; and a full reimplementation (LE2), undertaken by the same working group with the goal of consolidating the stable core of LE into a self-contained development environment. We summarise the language, sketch the design and scope of LE1, and describe the language-level and tooling-level changes introduced by LE2 --- in particular first-class scenario testing, structured diagnostics, an integrated browser IDE with template-aware editing and interactive justification trees, an LLM assistant for authoring, a graph view of the knowledge base, and early support for the Model Context Protocol and the Debug Adapter Protocol.

ABSTRACT. We argue that Turing tests as means for defining what falls within AI and what does not are insufficient and in addition, pernicious, as they tend to identify the whole discipline with an impersonating, humanity-fooling function: that of passing as human or as credible at all costs.

ABSTRACT. Understanding parallel complexity is increasingly important in the era of multicore processors and distributed computation. Yet parallel complexity is rarely taught at the undergraduate level, in part because it has lacked an accessible programming model. We show that Logica, an open-source logic programming language, makes parallel complexity practical and teachable. In Logica, the parallel time of a computation corresponds to the recursion depth of the program, and the total work corresponds to the size of the computation across all steps. We illustrate this through a series of short, readable programs, which consists of transitive closure, shortest paths, connected components, edit distance, matrix exponentiation, and PageRank, each expressing a parallel algorithm in a few lines. We also discuss the connection to tabling in XSB Prolog. Several of our examples can be expressed via lattice answer subsumption, and we present an experimental \texttt{logica\_table} library for SWI-Prolog that enables broader use of aggregation in tabling.

ABSTRACT. In this paper we describe our experience in using Prolog in compiler construction courses, in the context of an undergraduate degree in Computer Science. We show some of the most emblematic usages of Prolog or other Logic Programming languages, to solve problems in compiler design and implementation. Established language extensions such as constraints or more experimental ones, such as contextual logic programming are also drawn upon to tackle specific aspects of the design of a compiler. We also briefly discuss how this approach fares with students.

ABSTRACT. Mathematical and computational logic are basic subjects in most university degrees related to Computer Science. Not only are they fundamental in the study of programming languages, artificial intelligence, and the design and verification of hardware and software systems, but they also constitute the basis for logical reasoning. The process of acquiring these logical skills also contributes to the development of fundamental metacognitive abilities, which have significant implications for one’s approach to problem setting and problem solving, as well as for the structuring and expression of thought. However, students do not immediately recognize the connections between logic and what they will need to learn (and apply) in their future studies. To mitigate this feeling of disconnect between basic and more advanced content, we have developed DeduccionNatural[.pl, .blockly], two open-source tools built with the Ciao Prolog Playground and Google’s Blockly library. Available online and at https://github.com/Xuaco/DeduccionNatural, these tools allow users to verify the correctness of natural deduction proofs in propositional logic. In DeduccionNatural, the proofs are programs, the inference rules are functions of a predefined library, and the derived rules are subroutines (to refactor the proof) defined by the users/students. We have used the Prolog version of DeduccionNatural in the Cybersecurity Engineering and Artificial Intelligence degrees at Universidad Rey Juan Carlos (URJC) since 2021, and since 2025 we have also incorporated the Blockly-based version. Statistical evaluations of teaching experiences using the Prolog version have been published in educational conferences and received the third prize in the XI Edición del Premio Docentes Innovadores 2024 at URJC, while the evaluation of the Blockly-based version is ongoing.

ABSTRACT. We present ProverSLD, an interactive web-based tool for teaching and visualizing the SLD resolution process. The system supports first-order logic and a fragment of Prolog syntax. It provides a complete transformation pipeline from formulas to clausal form, visualizes the full SLD tree, and enables synchronized, step-by-step derivation via a resolution table linked to the tree. Users can choose between depth-first and breadth-first search strategies. We conducted a user study where the tool was deployed and evaluated in classroom settings, yielding positive feedback and results.