Learning user real-time intent for optimal dynamic web page transformation

摘要

Retailer websites use advanced information technologies to attract, retain and convert users into buyers. However, users change intentions over time or in response to stimuli and information encountered while browsing during a website visit. The present research work, based on the stimulus-organism-response (S-O-R) framework used in environmental psychology, proposes a new individual-level, dynamic learning model by observing individual users' shopping cart choices during navigation and inferring each user's unobserved real-time intent. Then the model automatically performs optimal intent-based web page transformations for the next page before the user exits the site in order to increase purchase conversion while reducing cart abandonment rates. The S-O-R paradigm is used to improve website performance. The user's shopping cart choices (R) are examined and reverse reasoning is used to infer the user's shopping intent states (O) with a hidden Markov model (HMM). Then the proposed model implements optimal page transformation by adjusting marketing and web stimuli (S) on the next page to influence 0 to generate positive outcomes R. The individual-level model system with a theoretical foundation is further demonstrated with mathematical descriptions to show that employing real-time, dynamic, optimal page transformation that reflects a user's intent leads to more positive outcomes, including higher purchase conversion rates and lower shopping cart abandonment. (7 refs.)