Real-Time Vehicle Demand Prediction via Map Search Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing information pushing methods for mobile Internet-based vehicle users are not timely or relevant, as they rely on historical travel data to predict vehicle demand, failing to account for current situations.
Innovation Solution
A method and apparatus that analyze search terms from a map application, match them with a pre-generated sequence, acquire log data, and use a pre-trained vehicle demand probability model to determine user demand, pushing information only when the demand exceeds a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If information pushing is based on historical travel data to predict vehicle demand, then the system can provide personalized information, but the timeliness and pertinence of the information push is insufficient
Solution Approach 1:
The system pre-generates search term sequences based on historical map application search operations and stores them for future matching. This preliminary preparation enables rapid comparison with real-time search terms without requiring complex historical data analysis at the moment of information pushing, thus improving both prediction accuracy and timeliness
Solution Approach 2:
The patent replaces the traditional mechanical approach of analyzing historical travel data with a machine learning-based vehicle demand probability model. The model takes real-time search terms and user behavior data as input to predict vehicle demand probability, substituting complex data processing with an intelligent prediction system that operates rapidly on current information
2Loss of time
If the system analyzes real-time search operations and matches them with pre-generated sequences, then the timeliness of information pushing is improved, but the device complexity increases
Solution Approach 1:
The system segments the information pushing process into distinct modules: real-time search term analysis, sequence matching, log data acquisition, vehicle demand probability prediction, and information pushing execution. This modular segmentation manages system complexity by organizing functions into independent, manageable components that can be developed and maintained separately
Solution Approach 2:
The patent introduces a vehicle demand probability model as an intermediary between real-time search term analysis and information pushing decisions. This intermediary layer processes real-time data through a trained prediction model, transforming complex real-time analysis into a simple probability threshold comparison that determines whether to push information, thus managing system complexity
3Measurement precision
If the system uses a pre-trained vehicle demand probability model to predict user demand, then the pertinence of information pushing is improved, but the loss of time in data processing increases
Solution Approach 1:
The vehicle demand probability model is pre-trained offline using historical data before deployment. This preliminary training action transfers the computationally intensive learning process to an offline phase, allowing the model to be ready for rapid real-time predictions without incurring training time delays during actual information pushing operations
Solution Approach 2:
The system changes the parameter of model operation from training mode to inference mode. During real-time operation, the model operates in inference mode where it rapidly processes input data through established parameters and weights, significantly reducing processing time compared to training mode while maintaining high prediction accuracy
Data Source
AI summary
A method and apparatus for pushing information. A specific embodiment of the searching method comprises: analysing, based on a second search operation sequentially performed to a map application installed on a client by a user, a search term corresponding to the second search operation; matching the analysed search term with a first search term in a first search term sequence to determine whether the matching is successful; acquiring log data of the map application installed on the client, in response to the matching being successful; determining a vehicle demand probability of the user based on the log data, acquired feature information of the user and a pre-trained vehicle demand probability model; and pushing information to the client in response to the determined vehicle demand probability being greater than a preset vehicle demand probability threshold.