Unified User Intention Inference via Latent Variable Mining
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for determining customer intention in a digital environment are limited by incomplete customer profile information, reliance on transactional data that fails to capture intended products or services, and customer unwillingness to provide personal details, leading to ineffective business strategies.
Innovation Solution
A processor-implemented method and system that integrates user data from multiple digital environments to derive unified user intention through weight assignment, latent variable analysis, and structural equation modeling, recommending strategies independent of transactional history and profile information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If customer profile information is used to determine customer intention, then business strategies can be formed, but the customer profile information is incomplete and customers are hesitant to provide personal details
Solution Approach 1:
The patent uses digital environment interaction data as an intermediary to infer customer intention without requiring direct collection of personal profile information. The system mediates between the need for customer intention data and customer privacy concerns by using behavioral traces from digital interactions as a proxy, eliminating the need for customers to willingly provide personal details while still obtaining actionable intention data.
2Measurement precision
If transactional data is used to capture customer intention, then purchase intentions can be measured, but the data fails to capture products or services intended by customers that were not purchased
Solution Approach 1:
The patent analyzes digital environment interactions (clicks, filters, searches, time spent) that occur before the purchase transaction to infer customer intention. By examining these preliminary actions in the digital journey, the system can identify products or services customers intended to purchase but ultimately did not buy, capturing intention data prior to the transaction completion point where traditional transactional data falls short.
3Measurement precision
If surveys are conducted to determine customer intention, then customer intention can be measured, but customers are unwilling to complete surveys and results reliability is compromised
Solution Approach 1:
The patent enables the system to automatically derive customer intention data from digital interaction traces without requiring active customer participation in surveys. The customer's own digital behavior (clicks, filters, search patterns, time spent) serves as self-generated data that the system processes automatically, eliminating the need for customers to willingly complete surveys while maintaining reliable intention measurement through objective behavioral evidence.
4Adaptability or versatility
If location-based strategies are used to form business decisions, then practical strategies can be developed, but fixing catchment area at category or business level is challenging
Solution Approach 1:
The patent segments the catchment area analysis by digital environment and interaction type rather than using traditional geographic boundaries. By dividing the analysis into specific digital interaction categories (searches, clicks, filters applied, time spent) within digital platforms, the system creates adaptable location-based strategies without the complexity of defining geographic catchment areas at category or business levels, using digital behavior patterns instead of geographic demography.
Data Source
AI summary
Traditionally, strategies are carried out based on customer intention indicated from surveys, transactional data, location of the store and customer profile. They have limitations such as partial capture of customer intention, absence of transactional data, challenges in defining catchment area associated with store location and lack of customer profiles. This disclosure relates to determining unified user intention from digital environment for plurality of strategies. An information associated with user is received. The information associated with the user is processed to derive weights associated with the user. The weights are integrated and mined through a variable reduction technique to determine at least one significant latent variable from a plurality of significant latent variables and associated significant attribute values. A structural equation model (SEM) is developed by applying the significant latent variables and the at least one event and a plurality of strategies are recommended for the plurality of applications.