Article: Simulation of consumer sentiment and optimisation of retail decisions under multi-agent reinforcement learning Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.85 pp.95 - 117 Abstract: In complex retail systems, existing process models rely on static historical data, failing to capture the nonlinear dynamics of fluctuating consumer emotions throughout the market lifecycle. This paper proposes the emotion-decision synergistic simulation network, a novel simulation framework for lifecycle strategy optimisation. First, a deep extraction module using bidirectional long short-term memory and multi-head attention models the temporal evolution of emotional states from unstructured textual feedback. These continuous state vectors are then integrated into a multi-agent environment, explicitly simulating the dynamic state transitions and behavioural processes of virtual consumers under marketing stimuli. Extensive experiments across three large-scale datasets demonstrate that EDSSN significantly enhances process evaluation, reducing the mean absolute percentage error by up to 4.15% compared to traditional baselines. Ultimately, ours model provides a robust simulation testbed, successfully bridging qualitative human factors and quantitative process modelling for reliable decision support. Inderscience Publishers - linking academia, business and industry through research

Title: Simulation of consumer sentiment and optimisation of retail decisions under multi-agent reinforcement learning

Authors: Zhankui Li

Addresses: School of Business Administration, Zhengzhou University of Science and Technology, Zhengzhou, 450064, China

Abstract: In complex retail systems, existing process models rely on static historical data, failing to capture the nonlinear dynamics of fluctuating consumer emotions throughout the market lifecycle. This paper proposes the emotion-decision synergistic simulation network, a novel simulation framework for lifecycle strategy optimisation. First, a deep extraction module using bidirectional long short-term memory and multi-head attention models the temporal evolution of emotional states from unstructured textual feedback. These continuous state vectors are then integrated into a multi-agent environment, explicitly simulating the dynamic state transitions and behavioural processes of virtual consumers under marketing stimuli. Extensive experiments across three large-scale datasets demonstrate that EDSSN significantly enhances process evaluation, reducing the mean absolute percentage error by up to 4.15% compared to traditional baselines. Ultimately, ours model provides a robust simulation testbed, successfully bridging qualitative human factors and quantitative process modelling for reliable decision support.

Keywords: multi-agent reinforcement learning; bidirectional long short-term memory; Bi-LSTM; multi-head attention mechanism; demand forecasting; consumer sentiment.

DOI: 10.1504/IJICT.2026.155482

International Journal of Information and Communication Technology, 2026 Vol.27 No.85, pp.95 - 117

Received: 20 Mar 2026
Accepted: 20 May 2026
Published online: 03 Aug 2026 *

Title: Simulation of consumer sentiment and optimisation of retail decisions under multi-agent reinforcement learning

Keep up-to-date