Offline Machine Learning Scores for Request-Agnostic Communication Selection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing platforms face challenges in efficiently selecting and presenting electronic communications to users due to the large volume of available content, leading to network and hardware resource overburden and suboptimal user experiences, as current methods often require real-time scoring that does not consider all relevant communications.
Innovation Solution
Implementing a machine learning model to generate request agnostic predicted interaction scores for electronic communications, allowing for pre-indexing and efficient retrieval, which considers multiple features to predict interaction probabilities, enabling faster and more comprehensive selection and presentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If real-time scoring is used to select electronic communications, then response time is reduced, but computational burden increases and not all relevant communications can be considered
Solution Approach 1:
The patent pre-computes and stores interaction scores for electronic communications before requests are made. The scoring system processes and ranks communications in advance, so when a request arrives, the pre-scored results can be quickly retrieved and used without performing heavy real-time computation. This preliminary action resolves the contradiction by shifting computational work from request-time to off-line periods.
2Reliability
If all electronic communications are transmitted to users, then user experience is improved by providing comprehensive content, but network and hardware resources are overburdened
Solution Approach 1:
The patent extracts only the most relevant electronic communications from the full corpus by using pre-computed interaction scores to identify and select a manageable subset of high-value content. Instead of transmitting all available communications, the system extracts and transmits only those that meet relevance thresholds, thereby improving user experience with quality content while conserving network and hardware resources by avoiding transmission of less relevant material.
3Loss of energy
If a small fraction of electronic communications is transmitted, then network and hardware resources are conserved, but user experience deteriorates due to limited content selection
Solution Approach 1:
The patent changes the parameter used for selection from simple heuristics or real-time ad-hoc scoring to pre-computed interaction scores that capture comprehensive relevance factors. By changing the selection parameter to these pre-analyzed scores, the system can confidently transmit a smaller fraction of communications while maintaining high user experience quality, because the pre-computed scores ensure that even the reduced subset contains the most relevant content.
Data Source
AI summary
Training and/or utilizing a machine learning model to generate request agnostic predicted interaction scores for electronic communications, and to utilization of request agnostic predicted interaction scores in determining whether, and/or how, to provide corresponding electronic communications to a client device in response to a request. A request agnostic predicted interaction score for an electronic communication provides an indication of quality of the communication, and is generated independent of corresponding request(s) for which it is utilized. In many implementations, a request agnostic predicted interaction score for an electronic communication is generated “offline” relative to corresponding request(s) for which it is utilized, and is pre-indexed with (or otherwise assigned to) the electronic communication. This enables fast and efficient retrieval, and utilization, of the request agnostic interaction score by computing device(s), when the electronic communication is responsive to a request.