Location-Based Interaction Cache for Faster ML Recommendations
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems face challenges in efficiently processing and actively influencing interactions between entities based on historical interaction data and location data, leading to computational burden and delayed recommendations.
Innovation Solution
A system that dynamically caches historical interaction data on a processor device based on computing capacity when an entity is proximate to a location, clears the cache when the entity moves away, and uses a machine-learning model to generate timely and relevant interaction recommendations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If historical interaction data is continuously stored and processed for all entities, then interaction recommendations can be generated, but computational resource burden increases and processing efficiency decreases
Solution Approach 1:
The patent applies local quality by caching historical interaction data selectively based on entity location rather than uniformly for all entities. When an entity device is detected at a specific location, only that entity's historical interaction data related to the location is cached and processed, while other entities' data remains uncached. This location-based differentiation reduces overall computational resource consumption while maintaining recommendation quality for relevant entities.
Solution Approach 2:
The system performs preliminary action by pre-caching historical interaction data when entity devices are detected at specific locations before interaction recommendations are needed. The location detection trigger initiates the caching process in advance, so when interaction recommendations are requested, the data is already available in cache, eliminating the need for real-time data retrieval and processing from storage systems.
2Loss of time
If historical interaction data is cached for all entities at all locations, then timely recommendations can be provided, but memory usage and system complexity increase
Solution Approach 1:
The patent implements dynamics by making the cache content adaptive to entity location. The cache dynamically adjusts which historical interaction data is stored based on real-time location detection of entity devices. When an entity arrives at a location, the cache populates with relevant historical data; when the entity leaves, that data can be evicted. This dynamic approach maintains minimal cache size while ensuring timely recommendation generation.
Solution Approach 2:
The system extracts only the necessary historical interaction data for caching based on location context. Instead of caching all historical data for all entities, the system extracts and caches only the subset of data relevant to the detected location and entity combination. This selective extraction reduces memory usage and simplifies cache management while maintaining recommendation quality.
3Speed
If interaction data is processed in real-time without caching, then data freshness is maintained, but processing speed and responsiveness decrease
Solution Approach 1:
The system performs preliminary action by caching historical interaction data in advance when entities are detected at specific locations. This pre-caching ensures that when interaction recommendations are needed, the data is already available in memory rather than requiring real-time retrieval from storage. The data freshness is maintained because the cache is updated whenever entities visit locations, ensuring recommendations are based on the most recent available interaction history.
Data Source
AI summary
A system can receive, from a computing device associated with a first entity, a threshold value that indicates an entity preference relating to an interaction type. The system can intercept a pending interaction before the pending interaction is executed. The pending interaction can originate from an interaction terminal configured to facilitate transfer of an electronic resource to a second entity. The pending interaction can include an interaction value of the pending interaction. The system can generate an interaction recommendation for the first entity using a trained machine-learning model based on the pending exchange value of the pending interaction, location data associated with the pending interaction, and historical exchange values of historical interactions associated with the first entity. The system can also cancel the pending interaction or deactivate access to a network account associated with the computing device.