Search Relevancy Scoring via Demand-Supply Desirability
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search algorithms in e-commerce contexts often fail to accurately reflect the user's intent due to their reliance on Boolean rules, leading to irrelevant results, as they do not differentiate between items like 'laptop battery' and 'laptop with battery', and do not consider the desirability of search results based on demand, supply, and user behavior.
Innovation Solution
A system and method that utilize numerical values for demand, supply, and desirability to compute a relevancy score for search results, adjusting scores based on user interactions and historical data, and sorting results accordingly, allowing for real-time textual summarization of search queries.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If Boolean rules are used for search algorithms, then the search process is simple and fast, but the accuracy of reflecting user intent deteriorates
Solution Approach 1:
The patent transforms the binary Boolean matching approach into a multi-parameter scoring system that evaluates relevance based on multiple factors including keyword presence, item attributes, user behavior history, and market dynamics. This parameter change enables nuanced differentiation between items while maintaining computational efficiency through structured scoring algorithms.
Solution Approach 2:
The patent replaces the mechanical Boolean logic system with a probabilistic and behavioral model that incorporates user interaction data, click-through rates, and purchase history. This substitution allows the system to infer user intent dynamically rather than relying on static keyword matching rules.
2Device complexity
If traditional search algorithms are used, then the system complexity is low, but the ability to differentiate between similar items deteriorates
Solution Approach 1:
The patent segments the search evaluation process into distinct components: keyword matching score, item attribute compatibility score, user behavior-based relevance score, and market demand score. Each segment is calculated independently and then aggregated, allowing complex differentiation without overwhelming system complexity.
Solution Approach 2:
The patent creates a universal relevancy scoring framework that can evaluate diverse item types (products, services, digital content) using the same multi-factor approach. This universal system handles differentiation across various domains while maintaining consistent evaluation principles.
3Device complexity
If static keyword matching is used, then the search algorithm is simple, but the responsiveness to user behavior and market dynamics deteriorates
Solution Approach 1:
The patent implements feedback loops where user interactions (clicks, purchases, time spent viewing) are continuously collected and used to update relevancy scores. This feedback mechanism allows the system to adapt to changing user preferences and market conditions dynamically, transforming static keyword matching into a responsive behavioral model.
Solution Approach 2:
The patent transitions from static keyword weights to dynamic scoring parameters that adjust based on real-time user behavior data and market trends. The relevancy scores are continuously updated to reflect current user preferences, making the search algorithm adaptive rather than fixed.
Data Source
AI summary
A system and method to sort search results based upon a desirability value is illustrated. This desirability value may be based upon the difference between a demand value and a supply value. Demand may be based upon user activity such as click-throughs, purchases, price, or location. Supply may be based upon a supply of keywords that may be the number of times a word is used in search or item title. The system and method may include receiving a search query, associating a first numerical value with a keyword that is a part of the search query, tracking user activity associated with the keyword, associating a second numerical value with the keyword based upon the user activity, finding a difference value between the first and second numerical values, associating this difference value with the keyword, sorting keywords based upon the difference values, and returning the search results of the sorting.