Search Result Diversification via Dynamic Reordering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Search engines often prioritize results from a single brand, leading to an incomplete representation of available products and making it difficult for users to compare diverse items, as listings are typically ordered by score without considering diversity in search results.
Innovation Solution
Implementing a system that uses soft constraints to diversify search results by introducing agents that evaluate and adjust the placement of items based on diversity scores, balancing relevance and diversity to ensure a more comprehensive display of search results.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If search results are ordered by relevance score, then search accuracy is improved, but diversity of brands and attributes deteriorates
Solution Approach 1:
The patent applies dynamics by making the search result ranking flexible and adaptive rather than static. The system dynamically adjusts the ordering of search results based on multiple factors including diversity metrics. Different ranking strategies can be applied depending on the situation, and the diversity of displayed items can be adjusted in real-time to balance relevance with variety, allowing the system to adapt between accuracy-focused and diversity-focused modes.
Solution Approach 2:
The patent implements parameter changes by modifying the ranking parameters beyond simple relevance scores. The system introduces diversity scores as an additional parameter, and can adjust weighting parameters to balance relevance and diversity. By changing the parameters used for sorting (from solely relevance-based to diversity-inclusive), the system resolves the contradiction between maintaining search accuracy and ensuring brand/attribute diversity in the displayed results.
2Productivity
If search results prioritize top-scoring items, then productivity is improved, but loss of information deteriorates
Solution Approach 1:
The patent applies partial or excessive action by displaying not only the top-scoring items but also additional items that provide diversity. Instead of showing only the absolute highest-ranked results, the system supplements the top results with diverse alternatives, ensuring that users see both the most relevant items and a broader representation of available products. This approach maintains productivity while reducing information loss.
Solution Approach 2:
The patent uses an intermediary approach by introducing diversity as a mediating factor between relevance scoring and final result presentation. The system doesn't simply display top-scoring items nor does it randomly select diverse items; instead, it uses diversity metrics as an intermediary mechanism to balance both concerns, selecting a mix of high-relevance and diverse items to comprehensively represent available products while maintaining search efficiency.
Data Source
AI summary
Disclosed is a system comprising a computer-readable storage medium storing at least one program, and a computer-implemented method for generating search results. A data access module accesses search results data indicative of a plurality items and that is indicative of ranking values of the plurality of items. A controller module generates page data based on the search results data. The controller module selects a first item from the plurality of items. A diversification module accesses a first rule of the page data. The diversification module determines a deviance value and a reordering-cost value of a second item from the plurality of items. The controller module can select, based at least on a first combination of the deviance and the reordering-cost values of the second item, the second item for placement ahead of the first item on the page data.


