Tehnički vjesnik, Vol. 33 No. 6, 2026.
Izvorni znanstveni članak
https://doi.org/10.17559/TV-20251230003251
A Hybrid Computational Framework for Flow Economy Optimization: Integrating NLPDriven Linguistic Metrics with Econometric Modeling
Li Zou
; 1) College of International Studies, Yangzhou University, Yangzhou, Jiangsu 225009, China 2) School of Foreign Languages, Chongqing Technology and Business University, Hanzhi Building, No. 19 Xuefu Avenue, Wugongli, Nan'an District, Chongqing 400067, China
*
Hongliang Yu
; Yangzhou University, College of International Studies, Jiangsu, 225009, China
Chenlu Li
; Symbio (Shenzhen) Information Technology Co., Ltd., Shenzhen, 518057, China
* Dopisni autor.
Sažetak
This paper presents a hybrid computational framework for the quantitative assessment of information flow efficiency in e-commerce systems by integrating automated linguistic feature extraction with econometric modeling. Existing flow-based analytical models typically conceptualize information flow as homogeneous, thereby neglecting the measurable heterogeneity of linguistic content that influences user decision-making and conversion outcomes. To address this limitation, we propose the Linguistic-Enhanced Five-Factor Flow Economy Model (LE-FFEEM), a multi-layer computational architecture that combines natural language processing techniques with econometric production functions. The proposed framework decomposes unstructured textual information into five computable linguistic dimensions: discourse coherence, semantic network density, cognitive load, interpersonal tone, and emotional resonance. Linguistic features are extracted through a unified pipeline employing contextualized embeddings, syntactic dependency parsing, and graph-theoretic density measures. The model is empirically evaluated using multi-source textual data collected from major Chinese e-commerce platforms. Results obtained from log-linearized regression analysis indicate that aggregated linguistic quality functions as a statistically significant explanatory variable for engagement-to-conversion efficiency (λ = 0.324). Validation across 12 platform-year observations demonstrates a 21% improvement in predictive accuracy (R² = 0.948) compared to baseline flow economy models. Computational analysis shows that semantic network construction operates with a time complexity of O(N log N), enabling scalable processing of large-scale textual corpora. The proposed framework provides an automated and extensible decision-support approach for optimizing platform-level information flow and contributes to the computational modeling of communication efficiency in digital commerce environments.
Ključne riječi
computational economics; decision support systems; e-commerce optimization; flow economy; NLP; semantic network analysis
Hrčak ID:
353090
URI
Datum izdavanja:
7.11.2026.
Posjeta: 0 *