Рақамли технологияларнинг назарий ва амалий масалалари Volume 9 Issue 3 (2026) · pp. 16-24
Structural-semantic indexing and selective federated Top-K retrieval algorithms in a multi-agent fact repository
Хужамбердиев, Д.Э.
Abstract
In a multi-agent information environment, facts are stored in local repositories maintained by individual agents. As repositories expand and the number of agents increases, evaluating every query against all facts and all agents leads to higher computational and communication costs. This paper develops a structural-semantic indexing model in which referent information, event semantics, temporal context, and evidence-confidence attributes are represented in separate but coordinated index layers. On the basis of this model, three algorithms are proposed: fact indexing, local Top-K retrieval using composite similarity, and selective query routing to semantically relevant agents. The local results returned by the selected agents are then merged federatively, while semantically related facts are grouped into meta-facts that preserve links to the original evidence. Formal analysis shows that the candidate fact set does not exceed the complete repository and that the selected agent set does not exceed the total number of agents. A functional comparison with exact-match retrieval and full broadcast search demonstrates that the proposed algorithms jointly account for semantic similarity, temporal context, evidence-based explainability, and inter-agent communication load within a unified retrieval process.
многоагентная системарепозиторий фактовструктурно-семантическое индексированиекомпозиционное сходствоTop-K поискселективная маршрутизацияфедеративный запросмета-фактmulti-agent systemfact repository
Metadata source: the journal's OAI-PMH archive · Sindex does not store the full text; it links to the source.