Search Result Reordering via User Metadata
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing search engine techniques often rank newer, more relevant search results lower due to fewer links, and return numerous results that are not tailored to individual user needs or subgroups, making it difficult for users to find relevant information.
Innovation Solution
A multiple-phase search system that uses user-specific metadata, such as bookmarks and group associations, to reorder search results and prioritize dynamically collected information, ensuring that more relevant results are displayed at the top.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If search results are ranked based on the number of links associated with each result, then the ordering reflects popularity and established relevance, but newer and potentially more relevant search results are ranked lower than older results
Solution Approach 1:
The patent implements dynamic re-ranking of search results based on user-specific metadata such as bookmarks, browsing history, and group associations. The ranking algorithm transitions from static link-based ordering to dynamic user-context-aware ordering, allowing newer relevant results to be promoted based on individual user preferences and behaviors rather than being constrained by traditional popularity metrics
Solution Approach 2:
The system changes the ranking parameters from solely link-based metrics to a composite scoring system that incorporates user-specific metadata weights. By adjusting the importance of different metadata types (bookmarks, history, group associations), the system can dynamically modify ranking parameters to surface more relevant results regardless of their link count or age
2Quantity of substance
If a user enters a search query that produces many pages of search results, then comprehensive coverage is achieved, but the user must sort through many results to find what is relevant
Solution Approach 1:
The patent applies local quality by customizing the search results ranking according to each user's specific metadata profile. Instead of applying a uniform ranking algorithm to all users, the system tailors the ranking to individual user preferences, bookmarked sites, browsing history, and group associations, making the most relevant results appear at the top for each specific user context
Solution Approach 2:
The system performs preliminary action by pre-processing and storing user metadata (bookmarks, browsing history, group associations) before the search occurs. This pre-collected information is then rapidly applied during search execution to immediately re-rank results according to user preferences, eliminating the need for users to manually sort through irrelevant results
3Reliability
If search results are ordered based on popularity of all users as a whole, then aggregate relevance is maximized, but results are not tailored to the needs of individual users or subgroups
Solution Approach 1:
The system changes the ranking parameters from aggregate popularity metrics to user-specific weighted combinations of metadata. By allowing different users to have different parameter weights based on their personal bookmarks, browsing patterns, and group memberships, the system achieves both overall accuracy and individual customization simultaneously
Solution Approach 2:
The patent segments the user base into individual users with unique metadata profiles and optionally into groups with shared associations. This segmentation allows the ranking system to treat each user (or group) as a distinct entity with customized ranking parameters, rather than applying a single aggregate ranking to all users
Data Source
AI summary
Embodiments of the invention improve the quality of search results returned for a given set of search terms based on metadata associated with the user performing the search. A search query may specify metadata elements to consider in ranking the search results. The metadata used may include bookmarks set by the user (either locally or at a social bookmaking site), group bookmarks, etc. In such a case, search results may be reordered to improve the ranking of websites that are both in the search results and in the bookmarks.