Cross-Cohort Ranking Model for Intent-Based Content Presentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional content management systems statically configure the presentation of content item cohorts, leading to suboptimal user engagement as they do not dynamically adjust to individual user intentions, resulting in irrelevant content being prioritized over relevant content.
Innovation Solution
A cross-cohort optimization model that utilizes machine learning to score and rank cohorts based on user intentions, both known and unknown, ensuring that relevant content is presented to users by optimizing the order of content item categories on their landing page.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If cohorts are presented in a predefined static order, then the system structure is simple and easy to implement, but user engagement decreases when user intentions do not align with the predetermined order
Solution Approach 1:
The patent implements dynamic cohort ranking by training a machine learning model (e.g., gradient boosted decision trees) that processes user profile attributes, session features, and cohort features to generate dynamic ranking scores. This allows the cohort presentation order to adapt in real-time based on user intentions, resolving the contradiction between static simplicity and dynamic adaptability.
Solution Approach 2:
The system changes the parameter of cohort ranking from fixed predefined values to dynamic machine learning-generated scores. By computing relevance scores based on user profiles, session context, and cohort characteristics, the system transforms the ranking parameter to be both adaptive and personally customized, overcoming the limitation of static ordering.
2Adaptability or versatility
If cohorts are dynamically ranked based on user intentions, then user engagement increases, but system complexity increases due to machine learning model requirements
Solution Approach 1:
The patent applies preliminary action by pre-training the machine learning model offline using historical user interaction data before deployment. This allows the complex model to be prepared in advance, and during runtime, only inference is performed, reducing the operational complexity while maintaining high adaptability to user intentions.
Solution Approach 2:
The system introduces an intermediary layer (the trained machine learning model) that sits between the user data and the cohort ranking output. This intermediary processes complex patterns and relationships, enabling the system to achieve high adaptability without requiring complex real-time computation logic throughout the entire system.
3Speed
If a machine learning model is trained offline and then used for ranking, then real-time performance is improved, but training time and computational resources are consumed
Solution Approach 1:
The patent implements preliminary action by performing the computationally intensive model training process offline before deployment. The trained model is then deployed for real-time inference, where only lightweight prediction operations are performed. This separates the time-consuming training phase from the speed-critical inference phase, achieving both goals.
Solution Approach 2:
The patent segments the system into two distinct phases: an offline training phase where the machine learning model is developed and optimized using historical data, and an online inference phase where the pre-trained model rapidly generates cohort rankings. This segmentation allows each phase to be optimized independently for its specific requirements.
Data Source
AI summary
Technologies for scoring and ranking cohorts containing content items using a machine-learned model are provided. The disclosed techniques include a cross-cohort optimization system that stores, within memory, cohort definition criteria for each cohort of a plurality of cohorts. The optimization system, for a particular user, for each cohort, identifies a plurality of content items that belong to the specific cohort based upon the cohort definition criteria. Using a machine-learned model, the optimization system generates a score for the specific cohort with respect to the particular user's intentions. The optimization system generates a ranking for the plurality of cohorts based on the respective scores of each cohort. The optimization system causes the plurality of content items of each cohort to be displayed concurrently on a computing device of the particular user. Display order for the plurality of cohorts is based on the ranking determined for the plurality of cohorts.