Asosiy kontentga oʻtish
Maqola

An Interactive Affective Computing Framework Utilizing NoSQL Architecture and Explainable Large Language Models for Real-Time Cognitive State Tracking

Ruziev XamrokulUniversity of Economics and Pedagogy, UzbekistanDavlatova NavbahorHardware and Software Support of Computer Systems, Karshi State Technical UniversityDaminov BobosherPostal Communication Systems, Karshi State Technical UniversityAbdurasulov AnvarInformation Technology Service, Karshi State Technical UniversityMurodjonova YulduzKarshi State Technical UniversityGaniyeva InobatComputer Engineering, Karshi State Technical University
2026
ABI

Annotatsiya

General Background Modern software engineering systems are currently shifting from traditional command-driven interfaces toward emotionally intelligent human-computer interaction frameworks. Specific Background Within this domain, tracking facial expressions via multi-dimensional symbolic blendshape weight vectors represents the primary modality for identifying continuous human affect without the massive storage overhead associated with raw multimedia files. Knowledge Gap However, conventional configurations frequently experience severe transactional write-amplification bottlenecks in legacy relational storage tiers alongside semantic parsing constraints and black-box opacity within generative artificial intelligence components. Aims To resolve these structural performance limitations, this paper introduces a novel, decoupled interactive affective computing framework that isolates real-time telemetry from heavy back-end computational processing. Results Empirically, the deployment of a specialized Document-Oriented Affective Database Schema within a non-relational environment maintains an ultra-low database write latency between 3.2 ms and 8.9 ms under concurrent high-throughput workloads, while the implementation of a Contextual Glyph Generator slashes language model context ingestion overhead by 74.2%. Furthermore, the framework achieves an expert-verified F1-score of 0.941 across complex emotional sub-clusters with an end-to-end multi-turn runtime latency of 112.5 ms, sitting comfortably beneath the critical 150 ms human-cognitive interactivity threshold. Novelty This architecture introduces a unique symbolic translation pathway that programmatically converts raw numerical coordinate streams into discrete structural textual tokens, ensuring that downstream transformer processes run with zero hallucination anomalies. Implications These findings demonstrate that explanation-driven transformer algorithms can be reliably deployed as transparent, accountable, and stable control layers within high-stakes computer engineering and behavioral monitoring environments. Keywords: Affective Computing, Explainable Artificial Intelligence, NoSQL Database Architecture, Real-Time Cognitive Tracking, Human-Computer Interaction Key Findings Highlights Document-oriented database configurations achieve an ultra-low write latency profile under concurrent high-throughput stream workloads. Contextual glyph generation reduces downstream large language model context ingestion overhead by over seventy percent. Explanation-driven validation protocols completely eliminate text generation hallucination anomalies during multi-turn interactive dialogue states.

Hali tarjima qilinmagan

Identifikatorlar

Iqtiboslar va manbalar