Exploration Traffic Ranking Model Using Return on Exploration Metrics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing content serving platforms face inefficiencies in serving exploration traffic due to the non-deterministic manner in which exploration content items are served, leading to wasted resources and potential exploitation by content providers, as these items are randomly selected and lack accurate user engagement predictions.
Innovation Solution
The implementation of a model that ranks exploration content items using return on exploration impression metrics to allocate percentages of exploration traffic to content providers, ensuring fair and efficient serving by considering the spend and exploration impressions of each provider, thereby mitigating exploitation and optimizing resource utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If exploration content items are served in a non-deterministic manner (randomly selected), then the user engagement model can be trained to predict likelihoods of user engagement, but resources are wasted and content providers can exploit the system by submitting large numbers of exploration content items
Solution Approach 1:
The patent changes the selection parameter from completely random to based on return on exploration impression metrics. This parameter change allows the system to maintain the exploration function (training the model) while reducing waste by prioritizing content items that have demonstrated better engagement potential, thus resolving the contradiction between training reliability and resource efficiency
Solution Approach 2:
The patent implements feedback by tracking return on exploration impression metrics for content providers and using this feedback to adjust the selection process. Content providers who generate better engagement results receive higher allocation percentages, creating a feedback loop that reduces resource waste while maintaining effective model training
2Reliability
If exploration content items are served in a non-deterministic manner, then the user engagement model can be trained, but content providers can exploit the system by obtaining more than their fair share of impressions
Solution Approach 1:
The patent uses feedback from return on exploration impression metrics to adjust allocation percentages for content providers. This feedback mechanism prevents exploitation by ensuring that content providers receive exploration traffic proportional to their actual performance, rather than allowing those with more submitted items to dominate
Solution Approach 2:
The patent changes the selection parameter from random selection to metric-based selection, where the return on exploration impression metric serves as a controlling parameter. This parameter change eliminates the exploitation vulnerability while preserving the training function
3Adaptability or versatility
If a large number of exploration content items are submitted by content providers, then more content is available for exploration and training, but the system becomes less efficient and fair in allocating exploration traffic
Solution Approach 1:
The patent introduces return on exploration impression metrics as a new parameter to control allocation, transforming the system from quantity-based (number of items submitted) to quality-based (engagement performance) allocation. This resolves the contradiction by maintaining high adaptability while improving productivity through efficient resource distribution
Data Source
AI summary
One or more computing devices, systems, and/or methods for implementing a model for serving exploration traffic are provided. An amount of spend by a content provider to provide content items of the content provider through a content serving platform to client devices of users is determined. A number of exploration impressions of users viewing exploration content items of the content provider over a timespan is determined. A return on exploration impression metric is determined for the content provider based upon a ratio of the amount of spend to the number of exploration impressions. The return on exploration metric is used to rank available exploration content items of content providers for serving exploration traffic.