The dynamic understanding of customer behaviour, in real time, is a very crucial issue in the current platforms of e-commerce, whereby the intentions of the users vary dynamically as they engage in browsing. The existing techniques mainly rely on post-session analysis or static analysis and, cannot react to any immediate reorientation of intention, such as purchase hesitation, comparison behaviour, or exit intent. A novel model, the Real-Time Customer Intention Shift Detection, a newly constructed Real-time Intention Shift Classification Network (RISC-Net), is proposed in this research. The primary objective of this work is to discover and classify the intention transitions under the real-time conditions of the modelling of the low-level behavioural indicators and time dynamics of the active user session. The proposed methodology would integrate the application of micro-behaviour characteristics encoding, sequential time learning, and a novel intention drift computation framework that quantifies non-conformance to historical behavioural tendencies on a slip time window. A decision module, which ought to be reached with a threshold, enables the deterministic and interpretable discovery of intention changes in real time. Continuous experimental evaluation conducted on e-commerce behavioural information has indicated that RISC-Net is far more effective in intention shift detection and has lower levels of false shift alert and response latency than existing sequential and static models. The results indicate that it is more responsive and predictive in the case of dynamic browsing. In short, the presented RISC-Net is a powerful and scalable explainable system of real-time intention-aware decision support enabling the development of proactive personalization, timely interventions, and more effective customer interactions in intelligent e-commerce systems.
Jayalathalakshmi et al. (Thu,) studied this question.