Mobile Geolocation Accuracy via Habit Probability Distributions
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current geolocation determination methods in mobile devices are prone to errors and lack accuracy, especially in distinguishing between adjacent locations, and are not optimized for real-time use on devices with limited computing power, affecting the reliability of location-based recommendations.
Innovation Solution
A method that combines three probability distributions based on global user habits, personal user habits, and local user habits to determine the likelihood of the user's current location, using a Bayesian approach to improve accuracy and adaptability over time, while preventing overfitting and handling unusual activities.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If current geolocation determination methods are used, then the device can provide location information, but the accuracy is insufficient to distinguish between adjacent places
Solution Approach 1:
The patent combines multiple probability distributions (global user habits, personal user habits, and local user habits) to determine the user's current place. This merging of multiple data sources and probability models resolves the contradiction by achieving both measurement precision and reliability through composite analysis rather than relying on geolocation alone.
Solution Approach 2:
The patent changes the parameter from raw geolocation coordinates to probability distributions based on user habits and behavioral patterns. By transforming the determination parameters from spatial coordinates to probabilistic models of user behavior, the system achieves higher accuracy in place determination while maintaining reliability.
2Measurement precision
If complex probability distribution calculations are performed, then place determination accuracy improves, but computational power requirements exceed mobile device capabilities
Solution Approach 1:
The patent segments the probability distribution calculation into three distinct components: global user habits, personal user habits, and local user habits. This segmentation allows the complex calculation to be broken down into manageable parts that can be processed sequentially on mobile devices, reducing computational complexity while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary actions by pre-establishing the three probability distribution models (global, personal, and local user habits) before real-time place determination. These pre-computed models are stored and reused during actual location analysis, significantly reducing the computational burden during real-time operation on mobile devices.
3Adaptability or versatility
If the system adapts to personal user habits, then recommendation relevance improves, but the system becomes less robust with limited initial data
Solution Approach 1:
The patent applies local quality by creating three distinct probability distribution models with different scopes: global user habits (broad population patterns), personal user habits (individual-specific patterns), and local user habits (region-specific patterns). Each model serves a specific quality requirement, and their combination provides both personalization and robustness by balancing individual adaptation with general reliability.
Data Source
AI summary
A mobile device and a method for determining a place according to geolocation information is disclosed. In one aspect, the method includes triggering an action in obtaining a first set of information related to the user, and, according to the current geographical location, a second set of places. The method may also include, for each place of the second set, determining a combined probability distribution that the user is currently located in said place, according to the first set of information, a first probability distributions based on a set of parameters related to global user habits, a second probability distributions based on a set of parameters related to user habits, and a third probability distributions based on a set of parameters related to local user habits. The method may further include triggering at least one action according to the combined probability distributions of the places of the second set.