AI Travel Service Pattern Learning via DNN
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing travel service methods are limited in providing accurate and comprehensive services during a user's travel period, as they primarily rely on current location-based information and fail to account for daily patterns, leading to inadequate travel-related services.
Innovation Solution
A method using a deep neural networks (DNN) model to learn user daily patterns, including wake-up time, commuting time, and travel route, to determine if a user is traveling by comparing these patterns with predetermined conditions, and providing travel information specific to the travel period and post-travel period.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing travel service methods rely only on current location-based information, then the service implementation is simple, but the accuracy and comprehensiveness of travel service provision is insufficient
Solution Approach 1:
The system performs preliminary learning of user daily patterns (wake-up time, commuting time, travel routes) using a DNN model before actual travel detection. This pre-training enables the system to accurately determine whether a user is traveling by comparing current behavior against established patterns, thereby improving measurement precision without requiring complex real-time analysis infrastructure
Solution Approach 2:
The patent introduces an intermediary determination module that acts as a mediator between raw location data and travel service provision. This module analyzes multiple patterns (wake-up time, commuting time, location) and generates a determined travel state, which then triggers appropriate services. This intermediary layer improves accuracy by synthesizing multiple data sources while maintaining system manageability
2Adaptability or versatility
If the system provides only current location-based travel services, then the service scope is limited, but the service delivery mechanism remains simple
Solution Approach 1:
The system dynamically adjusts service provision based on determined travel states. Instead of static location-based service triggers, the system continuously monitors pattern deviations and adaptively determines travel states, enabling versatile service coverage (during-travel and after-travel services) while maintaining a manageable service delivery framework through state-based control
3Measurement precision
If the system uses DNN model to learn daily patterns, then the accuracy of travel determination improves, but the computational resources and processing time increase
Solution Approach 1:
The DNN model performs pattern learning in advance during periods when computational resources are available and travel determination is not immediately required. By pre-learning wake-up time, commuting time, and travel route patterns, the system avoids performing complex DNN computations in real-time during travel detection, thereby improving accuracy while minimizing processing time delays
Solution Approach 2:
The system applies partial DNN processing by focusing on learning and comparing specific key patterns (wake-up time, commuting time, location) rather than analyzing all possible user behaviors. This selective application of DNN capabilities achieves sufficient travel determination accuracy while reducing overall computational burden and processing time
Data Source
AI summary
A method for controlling an artificial intelligent (AI) device providing a travel service, and which includes learning, via a processer of the AI device, a plurality of daily patterns related with a user's wake-up time, a commuting time, and a travel route using a deep neural networks (DNN) model; determining, via the processor, whether the daily patterns satisfy predetermined conditions each time a corresponding daily pattern among the plurality of daily patterns is generated; determining, via the processor, that the user is currently traveling when at least one daily pattern among the plurality of daily patterns that does not satisfy the predetermined conditions is detected; setting, via the processor, a time when the at least one daily pattern that does not satisfy the predetermined conditions is detected for a first time as a starting point of travel; setting, via the processor, a time from the starting point of the travel to an end point of the travel as a travel period of the user when an end of the user's travel is detected; and outputting travel information related with the travel period.