Article Catalog

Abstract This paper addresses intelligent access to structured and semi-structured data using large language models (LLMs). We develop an approach to LLM-driven query planning in which a user’s information need is handled as a

[…] Read more

sequence of interdependent operations: query analysis, grounding in data sources and the schema, plan construction, execution, validation of intermediate results, and, when needed, replanning. The study draws on a comparative analysis of the open benchmarks BIRD, Spider 2.0, and HybridQA and of representative published approaches, from Text-to-SQL pipelines such as DIN-SQL and CHASE-SQL to hybrid methods such as BlendSQL and Weaver. The proposed architecture separates planning from execution and allows SQL operations to be combined with semantic processing by an LLM. We formalize the plan as a directed acyclic graph of typed SQL, semantic, and lookup operations with validation predicates, define the conditions for partial replanning, and illustrate the procedure with a worked example. Our analysis of existing experimental results indicates that decomposing complex queries, combining heterogeneous sources, and monitoring intermediate state are each associated with higher accuracy or lower cost in multi-step data access tasks. The evidence supports assigning the LLM the role of a control component that coordinates the sequence of actions, although the proposed architecture still needs to be evaluated quantitatively under uniform experimental conditions. To that end, we propose an evaluation protocol with baselines, ablation variants, and falsifiable predictions.

Abstract Regulation of artificial intelligence in education is currently formulated as principles and institutional obligations. UNESCO’s guidance on generative AI calls for a human-centered approach; the EU AI Act classifies

[…] Read more

several educational uses of AI as high-risk, and Russian policy frames the digitalization of schooling through national strategies and state information systems. These instruments share a blind spot: they say little about the level at which AI and digital platforms actually enter education, namely the encounters, face-to-face and screen-mediated, between students, teachers, and machines. This review argues that this level, which Goffman called the interaction order, has its own regularities and that these regularities systematically refract policy. Drawing on Goffman’s dramaturgy and frame analysis, on ethnomethodology and conversation analysis, and, in a supporting role, on symbolic interactionism and phenomenology, the paper identifies three mechanisms through which the interaction order reshapes regulatory intentions: the situated completion of rules, the redistribution of exposure and face-work, and the reconfiguration of participation frameworks when AI systems become participants in classroom talk. Each mechanism is illustrated with empirical research on video-mediated teaching, camera policies, AI-detection tools, and generative AI tutors, and the three mechanisms are traced in recent developments in Russian education. From these mechanisms, it derives three propositions for research and policy evaluation, including the claim that visibility rules convert unequal backstage resources into unequal classroom standing. It concludes that principle-based regulation needs to be complemented by interaction-level assessment if it is to achieve its own stated aims.