Content Recommendation Scoring for Conversion Attribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing content recommendation systems struggle to determine the highest quality recommended content from a large number of combinations of content elements due to independent recall, synthesis, and delivery policies, making it difficult to attribute conversion effects to specific policies and improve recommendation efficiency.
Innovation Solution
Analyze historical conversion data to determine contribution scores of content elements, calculate importance scores for candidate recommended contents, and select high-quality content based on these scores to enhance the quality of recommended content delivery.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If independent recall, synthesis, and delivery policies are used for content recommendation, then system complexity is reduced and operation is simplified, but the ability to attribute conversion effects to specific policies is lost and recommendation quality deteriorates
Solution Approach 1:
The patent segments the recommendation system into independent modular policies (recall policy, synthesis policy, delivery policy), each responsible for specific functions. This segmentation allows each policy to be evaluated independently for its contribution to conversion effects, resolving the contradiction by enabling both independent operation and precise measurement of each segment's performance.
Solution Approach 2:
The patent implements a feedback mechanism that tracks conversion parameters throughout the recommendation pipeline. By monitoring how each policy stage contributes to final conversions and feeding this information back to the system, it enables precise attribution of conversion effects to specific policies while maintaining their independent operation.
2Adaptability or versatility
If a large number of content element combinations are generated, then content diversity and user choice are improved, but the difficulty of determining highest quality content increases and processing time is lost
Solution Approach 1:
The patent performs preliminary evaluation of content elements and combinations using conversion parameter analysis before final recommendation. By pre-assessing the quality and conversion potential of different content combinations based on historical data and contribution scores, it reduces the complexity of determining highest quality content from a large number of combinations.
Solution Approach 2:
The patent introduces contribution scores as a new evaluation parameter that quantifies the impact of each content element on conversion. By changing from subjective quality assessment to objective parameter-based evaluation, it simplifies the measurement of content quality across diverse combinations, enabling efficient identification of highest quality recommendations.
3Measurement precision
If conversion parameters are attributed to individual content elements, then recommendation quality and targeting precision are improved, but the computational complexity and processing requirements increase
Solution Approach 1:
The patent extracts and isolates the contribution of individual content elements to conversion outcomes by analyzing conversion parameters at each policy stage. This extraction process separates the complex attribution problem into manageable components, allowing precise measurement of each element's contribution without requiring complete system reconfiguration.
Solution Approach 2:
The patent introduces contribution scores as an intermediary metric that bridges the gap between raw conversion parameters and actionable recommendation decisions. This intermediary layer simplifies the computational complexity by providing a standardized measure of content element value, making it easier to process and act on attribution data without overwhelming system complexity.
Data Source
AI summary
According to embodiments of the present disclosure, a solution for content recommendation is provided. A method for content recommendation includes: obtaining historical conversion parameter values corresponding to a set of historical recommended contents related to a plurality of content elements; determining respective contribution scores of the plurality of content elements in the conversion based on the historical conversion parameter values corresponding to the set of historical recommended contents; determining, based at least on the respective contribution scores of the plurality of content elements, respective importance scores corresponding to a plurality of candidate recommended contents, each candidate recommended content including at least one content element of the plurality of content elements; and selecting, based on the importance scores corresponding to the plurality of candidate recommended contents, at least one recommended content from the plurality of candidate recommended contents for providing to a first user group.