Recommender System Policy Module for External Influence
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Recommender systems in electronic commerce lack a mechanism to influence recommendations based on external parameters unrelated to user correlations or item characteristics, such as the origin of content, similar to search engine 'pay for placement' features.
Innovation Solution
Implementing a method within recommender systems to apply programmed placement policies that influence recommendations by setting up policies based on time, keywords, demographic variables, item characteristics, release dates, or the brand/source of items, allowing for preferential treatment of specific content providers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If recommender systems use collaborative filtering or content-based filtering to generate recommendations, then recommendations are based on user correlations or item characteristics, but there is no mechanism to influence recommendations based on external parameters such as content origin or brand
Solution Approach 1:
The system segments the recommendation process into distinct modules: a base recommender system that generates initial recommendations based on user-item correlations, and a separate policy module that applies external influence parameters. This segmentation allows the system to maintain its core recommendation functionality while adding influence capabilities without fundamentally restructuring the entire system.
Solution Approach 2:
The patent introduces a policy module as an intermediary between the base recommender system and the final recommendation output. This policy module receives initial recommendations and applies programmed policies (such as pay-for-placement, brand preferences, or content origin biases) to modify the recommendation list before presenting it to users, thereby enabling external parameter influence without directly altering the core recommendation engine.
2Adaptability or versatility
If pay-for-placement policies are implemented in search engines, then companies can buy high ranking for keywords, but recommender systems lack similar functionality to favor specific content providers
Solution Approach 1:
The system changes the parameters of the recommendation process by introducing policy parameters that allow external influence. These parameters include pay-for-placement amounts, brand preference weights, content origin biases, and temporal factors. By adjusting these parameters, the system can dynamically favor specific content providers while maintaining the ability to revert to unbiased recommendations when policies are not applied.
3Productivity
If multiple policies are applied to influence recommendations, then content providers can benefit from biased recommendations, but the system complexity increases
Solution Approach 1:
The system performs preliminary actions by pre-configuring policies with specific parameters (such as budget allocations, preference weights, and target content providers) before the recommendation process begins. These policies are stored and automatically applied based on matching criteria, eliminating the need for real-time complex calculations and reducing system complexity during actual recommendation generation.
Solution Approach 2:
The system applies policies partially rather than comprehensively to all recommendations. Policies are applied only when specific conditions are met (such as when a user queries for particular content types or when budget thresholds are reached), allowing the system to maintain simplicity for the majority of recommendations while providing enhanced visibility to content providers when appropriate.
Data Source
AI summary
A system and method for influencing a recommender system and advertising uses a controlled, programmed policy. The recommender system uses the policy to adjust recommendations made to users for particular items, so that a policy holder receives some measure of preference in recommendations. The preference may be implemented in the form of a filter, or a presentation of items. The policy can be applied selectively, and be based on pay for placement type consideration, demographics, time, and other related parameters. Advertising for the users can be similarly adjusted in coordination with the policy.