Web Call Log Pattern Mining for Real-Time Topic Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing data mining techniques for episode mining in services like human resource and financial services are inefficient in quickly and accurately identifying patterns for real-time task execution, as they lack effective methods to associate and predict call topics based on user behavior from web and call logs.
Innovation Solution
A method and system that associate web log records with call log records by identifying patterns of web accesses and timestamps, and predict call topics for new calls based on these patterns, using a processing device to analyze user behavior and correlate web clicks with calls within a specific time frame.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional episode mining systems search for patterns in data and utilize them in real-time, then pattern identification capability is provided, but the speed and accuracy of real-time task execution remain insufficient
Solution Approach 1:
The system performs preliminary pattern mining and association between web log records and call log records before real-time execution. By pre-identifying patterns and storing them in a database, the system can quickly retrieve and apply relevant patterns during real-time task execution without performing complex pattern mining operations at the moment of execution, thereby improving both speed and accuracy.
Solution Approach 2:
The patent introduces an intermediary database that stores pre-mined patterns and associated web-call log relationships. This intermediary structure acts as a mediator between the raw log data and real-time execution requirements, enabling fast pattern retrieval and application while maintaining high accuracy through pre-computed associations.
2Measurement precision
If the system associates web log records with call log records to identify patterns, then prediction accuracy improves, but processing time and computational complexity increase
Solution Approach 1:
The system performs the time-consuming association between web log records and call log records, as well as pattern identification, as a preliminary batch processing step. By completing these computationally intensive tasks in advance and storing the results in a database, the system minimizes processing time during real-time execution, achieving both high prediction accuracy and fast response time.
Solution Approach 2:
The patent segments the processing into two distinct phases: a batch processing phase for associating logs and identifying patterns, and a real-time execution phase for retrieving and applying pre-identified patterns. This segmentation allows the system to optimize for accuracy during batch processing while maintaining speed during real-time operations.
3Manufacturing precision
If the system analyzes user behavior patterns from web and call logs, then task completion accuracy improves, but device complexity and data processing requirements increase
Solution Approach 1:
The system segments the data processing complexity into a centralized batch processing component that handles the complex association and pattern identification tasks. By isolating these complex operations in a dedicated processing layer that runs periodically, the system simplifies the real-time execution environment while maintaining high task completion accuracy through pre-computed pattern associations.
Data Source
AI summary
Methods and systems of performing data mining may include receiving a plurality of web log records and a plurality of call log records; associating one or more web log records with a call log record, wherein the associated user for each of the associated one or more web log records and the call log record are the same; identifying one or more patterns among the web log records for the plurality of call log records, wherein each pattern comprises one or more web accesses, a time stamp at which each of the one or more web accesses is performed and the call topic for the call log record; identifying one or more web log records associated with a new call, and predicting a call topic for the new call based on at least one pattern and the one or more web log records.