Two Languages, One System:Tightly Connecting XSB Prolog and Python

Abstract: Python, ranked first on the May 2022 Tiobe index, is a hugely popular language, heavily used in machine learning and other applications. Prolog, ranked twenty-first the May 2022 Tiobe index, while less popular has important reasoning and knowledge representation capabilities, particularly since modern Prologs support constraint-based reasoning, tabling-based deduction, and probabilistic inference. Despite their differences, Prolog and Python share important commonalities. First, both Prolog and CPython (the standard Python implementation) are written in C with well-developed interfaces to other C programs. In addition, both languages are dynamically typed with data structures that are recursively generated in just a few ways. Infact, nearly all core data structures of the two languages can be efficiently bi-translated, leading to a tight connection of the two systems. This talk presents the design, experience, and implications of such a connection using XSB Prolog version 5.0. The connection for XSB to call Python has led to XSB orchestrating commercial projects using interfaces to Elastic search, dense vector storage, nlp systems, Google maps, and to a 14.6 billion triple Wikidata graph. The connection for Python to call XSB allows XSB to be imported as any other Python module so that XSB can easily be invoked from Jupyter notebooks or other graphical interfaces. On a more speculative level, the talk mentions how this work might be leveraged for research in neuro-symbolic learning, natural language processing and cross-language type inference.

pickup at 18:00 from the Technion (Tour at Haifa, no food will be provided)

50th anniversary of the birth of Prolog: Some reflections on Prolog's Evolution, Status, and Future [LINK TO SLIDES]

Abstract: This year we celebrate Prolog's 50th anniversary, and the continued relevance of Prolog and logic programming as a basis for both higher-level programming and symbolic, explainable AI.  This talk will provide an overview of Prolog's evolution, status, and future. It will start with a quick tour of the major milestones in the advancement of the language and its implementations, from the original Marseille and Edinburgh versions, to the many current ones, with more appearing continuously. This will be followed by some reflections on issues such as what still makes Prolog different and relevant as a language and tool, the best approaches for teaching Prolog, some landmark applications, relation to other logic programming languages, or the use of Prolog and Prolog-related tools for other programming languages.  The talk will also attempt to dispel along the way some common misconceptions about the language, while discussing some past weaknesses and others that may still need addressing. It will conclude with some reflections on challenges and opportunities for the future.

Part of the contents of this talk appear in the recent TPLP paper "50 years of Prolog and Beyond", by Philipp Körner, Michael Leuschel, João Barbosa, Vítor Santos Costa, Verónica Dahl, Manuel V. Hermenegildo, Jose F. Morales, Jan Wielemaker, Daniel Diaz, Salvador Abreu, and Giovanni Ciatto, written for Prolog's 50th anniversary and TPLP's 20th anniversary. See prologyear.logicprogramming.org for pointers to this paper and other initiatives related to the Year of Prolog. The Year of Prolog is organized by the Association for Logic Programming and the Prolog Heritage foundation.

When solving combinatorial problems, pruning symmetric solution candidates from the search space is essential. Most of the existing approaches are instance-specific and focus on the automatic computation of Symmetry Breaking Constraints (SBCs) for each given problem instance. However, the application of such approaches to large-scale instances or advanced problem encodings might be problematic since the computed SBCs are propositional and, therefore, can neither be meaningfully interpreted nor transferred to other instances. As a result, a time-consuming recomputation of SBCs must be done before every invocation of a solver.

To overcome these limitations, we introduce a new model-oriented approach for Answer Set Programming that lifts the SBCs of small problem instances into a set of interpretable first-order constraints using a form of machine learning called Inductive Logic Programming. After targeting simple combinatorial problems, we aim to extend our method to be applied also for advanced decision and optimization problems.

From "ASP modulo CSP" to "ASP modulo X", or how clingcon paved the way for clingo[X] systems

Abstract: Clingcon extends the ASP system Clingo with linear constraints over integers. As such, its series of variants served as a consecutive design study onhow to extend ASP systems with foreign inferences. Meanwhile this has culminated in the generic theory reasoning framework of the ASP system clingo.We will sketch the evolution of this technology, look at its current state, and emphasis the role of semantics in view of a successful outcome.

Probabilistic Logic Programming: Semantics, Inference and Learning [LINK TO SLIDES]

Abstract: Probabilistic Logic Programming is now more than 30 years old and has become an active field of research at the intersection of Logic Programming and Uncertainty in AI.Among the various semantics, the Distribution Semantics (DS) has emerged as one of the most intuitive and versatile. The talk will introduce the DS and its various extensions to deal with function symbols, continuous random variables and multiple stable models.Computing the probability of queries is the task of inference, which can be solved either by exact or approximate algorithms. Exact inference is usually performed by means of knowledge compilation while approximate inference by means of Monte Carlo.Inference is at the basis of learning programs from data that has recently received much attention. The talk will provide an overview of algorithms for learning the parameters and for learning the structure of programs, discussing the various Inductive Logic Programming techniques that have been adopted and the various tradeoffs between quality of the model and speed of learning.

Panel: Past, Present and Future of Prolog