Promotion System Candidate Selection Using Connection Graphs
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automatic content classification and promotion systems for internet services lack efficiency in targeting and engagement prediction, leading to suboptimal content promotion strategies.
Innovation Solution
A real-time messaging platform with a promotion system that performs initial candidate selection using connection graph information, applies filtering for fatigue management, and utilizes a prediction model with an auction model to rank promotions for effective user engagement.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If expensive processing is applied to all candidate promotions, then prediction accuracy improves, but system cost increases
Solution Approach 1:
The patent segments the candidate promotion set into different processing tiers: initial candidate selection using connection graph information (low-cost processing) and subsequent detailed evaluation using prediction models (high-cost processing) only for selected candidates. This segmentation allows the system to apply expensive processing only where necessary, resolving the contradiction between accuracy and cost.
Solution Approach 2:
The system performs preliminary candidate selection using connection graph information before applying expensive prediction model processing. This preliminary action filters out unlikely candidates early, ensuring that costly processing is applied only to promising candidates, thereby maintaining prediction accuracy while controlling system costs.
2Manufacturing precision
If more candidate promotions are evaluated, then promotion quality improves, but processing time increases
Solution Approach 1:
The evaluation process is segmented into multiple stages: initial filtering using connection graphs, intermediate filtering using fatigue management, and final ranking using prediction models. This segmentation allows comprehensive evaluation of promotion quality while managing processing time through progressive filtering.
Solution Approach 2:
The system performs preliminary filtering using connection graph information and fatigue management before applying time-intensive prediction models. This preliminary action reduces the candidate set size, allowing thorough evaluation of quality metrics within acceptable time constraints.
3Measurement precision
If connection graph information is used for candidate selection, then targeting precision improves, but system complexity increases
Solution Approach 1:
The connection graph serves as an intermediary data structure that pre-computes and stores relationship information between users and promotions. This intermediary allows the system to achieve high targeting precision by querying pre-computed connections rather than calculating relationships in real-time, while managing complexity through dedicated graph management infrastructure.
4Ease of operation
If fatigue management filtering is applied, then user experience improves, but candidate selection complexity increases
Solution Approach 1:
The fatigue management system uses feedback from user interactions with previous promotions to adjust candidate selection. By monitoring user responses and incorporating this feedback into filtering decisions, the system improves user experience by avoiding over-exposure to promotions, while managing complexity through systematic feedback collection and processing mechanisms.
Data Source
AI summary
A real-time messaging platform and method are disclosed which can be used to promote content in the messaging platform. In one embodiment, the promotion system is disclosed which performs initial candidate selection so as to narrow down the set of candidate promotions before applying more expensive processing. The candidate selection takes advantage of the connection graph information associated with accounts in the messaging platform to identify targeted accounts. In another embodiment, the promotion system uses a prediction model to predict a user's engagement with the promotion and utilizes the prediction to assist in ranking the candidate promotions. Promotions can be assigned metrics based, for example, on a weighted combination of user engagement rates, decayed with time to reflect an intuition that recent interactions by one or more users with the promotion will have a greater impact than older interactions with the promotion.