Generalized Linear Mixed Models for Search Personalization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Generalized linear models used in information retrieval systems lack personalization, leading to irrelevant content being prioritized over relevant content, resulting in excessive electronic resource consumption due to the generation and display of irrelevant search results.
Innovation Solution
Implementing generalized linear mixed models that incorporate query-based and user-based models to personalize search results by generating scores for candidates based on their profiles and user interactions, thereby selecting and displaying the most relevant content.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If generalized linear models are used for search results, then the system is simple and fast, but personalization is lacking and irrelevant content is prioritized
Solution Approach 1:
The patent segments the search model into two distinct components: a query-based model for general relevance and a user-based model for personalization. This segmentation allows the system to incorporate personalized user behavior data without completely redesigning the search algorithm, thus improving adaptability while managing complexity through modular architecture.
Solution Approach 2:
The patent merges the query-based model and user-based model into a unified generalized linear mixed model framework. This combination integrates both general search relevance and personalized user preferences, allowing the system to deliver personalized results without sacrificing the foundational search capabilities, thereby resolving the contradiction between personalization and complexity.
2Reliability
If personalized search results are implemented, then relevant content is prioritized, but electronic resource consumption increases due to processing user-specific data
Solution Approach 1:
The patent implements preliminary action by pre-computing user-based scores and caching them for future queries. This allows the system to quickly retrieve and apply pre-analyzed user behavior patterns during search operations, reducing real-time computational overhead and electronic resource consumption while maintaining high search result relevance through personalized ranking.
3Adaptability or versatility
If generalized linear mixed models are implemented, then personalization and relevance improve, but computational expense and processing time increase
Solution Approach 1:
The patent applies partial action by selectively applying the computationally intensive user-based model only to candidates that pass initial filtering by the query-based model. This approach computes personalized scores for a subset of candidates rather than all candidates, reducing overall computational expense and improving search processing speed while still delivering personalized results for the most relevant candidates.
Data Source
AI summary
Techniques for improving search using generalized linear mixed models are disclosed herein. In some embodiments, a computer-implemented method comprises: receiving a search query comprising at least one search term and being associated with a user; extracting features from corresponding profiles of a plurality of candidates; for each one of the candidates, generating a corresponding score based on a generalized linear mixed model comprising a generalized linear query-based model and a random effects user-based model; selecting a subset of candidates from the plurality of candidates based on the corresponding scores; and causing the selected subset of candidates to be displayed to the user in a search results page for the search query.