Estimation System Using Merged User Data for Location Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems struggle to accurately predict future locations visited by users with limited historical data, leading to lower estimation accuracy.
Innovation Solution
An estimation system that uses a learning model to analyze the relationship between past and future locations visited by users, incorporating factors like usage patterns and demographics to improve prediction accuracy for novice users.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the system estimates future locations using only the user's own limited historical data, then the system can operate with minimal data requirements, but the estimation accuracy becomes lower when the user has not visited many locations in the past
Solution Approach 1:
The patent combines the user's own historical location data with aggregated historical data from multiple other users to improve estimation accuracy. The server stores historical location information from multiple users and uses this combined dataset to generate more accurate future location predictions, especially when the target user has limited historical data.
Solution Approach 2:
The patent introduces a server as an intermediary that aggregates and processes historical location data from multiple users. The server acts as a mediator between individual users with limited data and the estimation algorithm, providing enhanced prediction capabilities by supplying additional contextual information from the aggregated user base.
2Measurement precision
If the system waits to accumulate more historical data before making estimates, then the estimation accuracy would improve, but the system cannot provide timely predictions for users with limited history
Solution Approach 1:
The patent performs preliminary actions by pre-aggregating and storing historical location data from multiple users in advance. This pre-processing allows the system to immediately provide accurate predictions when needed, without waiting to accumulate sufficient historical data from the target user at the moment of estimation.
Solution Approach 2:
The patent creates a universal historical location database that serves multiple users simultaneously. The aggregated data from one user base benefits all users in the system, allowing any user to receive accurate predictions regardless of their individual data history, thus eliminating prediction delays.
3Device complexity
If the system uses only individual user historical data, then the system architecture remains simple, but the estimation accuracy for users with limited visit history cannot be improved
Solution Approach 1:
The patent merges individual user data with aggregated multi-user historical data in a unified estimation framework. The server combines these datasets and applies estimation algorithms that leverage patterns from the broader user base, improving accuracy without requiring complex distributed architectures at the user device level.
Solution Approach 2:
The patent introduces a server as an intermediary that handles the complexity of data aggregation and pattern recognition. This centralizes the computational complexity in a manageable location, allowing individual user devices to remain simple while still benefiting from sophisticated multi-user pattern analysis for improved prediction accuracy.
Data Source
AI summary
A storage of an estimation system stores a learning model which has learned a relationship between first position information which is based on a position of a first location visited by a first user in past and second position information which is based on a position of a second location visited by the first user after the first location. At least one processor of the estimation system acquires third position information which is based on a position of a third location being a location visited by a second user in the past. The at least one processor acquires output of the learning model corresponding to the third position information as an estimation result of fourth position information which is based on a position of a fourth location being a location that is likely to be visited by the second user in future.