Constrained text generation to measure reading performance: A new approach based on constraint programming

Tuesday, February 13, 2024 | 10:00 am (CET) | Room: S.1.42 | Alpen-Adria Universität Klagenfurt

Jean-Charles Régin| Professor at the Université Côte d’Azur and Chair for Decision Intelligence at the Interdisciplinary Institutes of Artificial Intelligence (3IA)

Abstract: This talk introduces a new approach to generate strongly constrained texts. We consider standardized sentence generation for the typical application of vision screening. To solve this problem, we formalize it as a discrete combinatorial optimization problem on words and show how constraint programming and multivalued decision diagrams (MDD), a well-known data structure to deal with constraints can be used to solve it. We show how part of the language is kept thanks to n-grams. Once the sentences are obtained, we apply a language model (LLM: GPT-3) to keep the best ones. We detail this for English and also for French where the agreement and conjugation rules are known to be more complex. We will also discuss about some possible improvements for a better integration of LLM into constraint programming.

(This presentation, does not require any knowledge in LLM or in constraint programming.)

Bio: JC Régin is an internationally recognized expert of Constraint Programming (CP). Innovation and ground-breaking research are a constancy in his career and his contributions are recognized internationally in the academic and non-academic worlds. He is one of the inventors of the global constraints in CP (i.e. algorithms to quickly eliminate incoherent values). His paper describing the All different global constraint is highly cited (> 1000 cit.). This constraint is now implemented in almost all CP solvers and routinely used in many applications by companies such as IBM, Oracle, Microsoft, Google or SAS. The cited article, published in 1994, received the „Classical Paper Award“ in 2013 from the American Association on Artificial Intelligence: „For ground-breaking contributions to constraint programming via the development of one of the first propagators for global constraints.“

In 2013, he received the Association for Constraint Programming Research Excellence Award „in recognition of a program of seminal and outstanding scientific contributions to both the theory and practice of constraint programming“.