Model-Based Call Home Data Updating
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for call home data collection in computing systems often result in a gap in data collection, as necessary data is identified only after the initial decision to act on a defect has been made, and changes in reference code owners can lead to a loss of knowledge on what data is required to resolve defects.
Innovation Solution
A model-based updating approach is implemented, where problem analysis data is used to determine reference codes associated with defects, and a machine-learning model is trained using data usage tracking to output data confidence scores, dynamically updating the association of reference codes with defects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional call home data collection methods are used, then data collection is performed after defect identification, but data collection completeness is poor and knowledge is lost when reference code owners change
Solution Approach 1:
The system performs preliminary actions by collecting and tracking data usage information before the initial defect resolution decision is made. The machine learning model is trained in advance on historical data usage patterns, enabling it to predict and identify relevant data collection requirements proactively, rather than reactively after the defect is identified and acted upon.
Solution Approach 2:
The system implements feedback mechanisms by continuously tracking actual data usage during defect resolution and using this information to retrain and refine the machine learning model. The model learns from real-world usage patterns and feedback, progressively improving its ability to accurately identify and collect relevant data for defect resolution.
2Speed
If data collection is performed after initial defect action decision, then immediate response is maintained, but necessary data may be missed
Solution Approach 1:
The system performs preliminary data identification and collection actions by using the trained machine learning model to predict relevant data requirements before the final defect resolution decision is made. This allows the system to prepare and collect necessary data in advance, maintaining rapid response capability while improving data completeness and relevance.
Solution Approach 2:
The system dynamically adapts its data collection strategy by continuously updating the machine learning model with new usage patterns and feedback. This dynamic approach allows the system to adjust its data identification and collection processes in real-time, optimizing both response speed and data relevance accuracy as defect resolution progresses.
3Adaptability or versatility
If reference code owners change, then system flexibility is improved, but knowledge about required data is lost
Solution Approach 1:
The system implements self-service by using the machine learning model to automatically identify, track, and manage data collection requirements independent of specific reference code owners. The model autonomously learns from usage patterns and maintains knowledge about required data, eliminating dependency on individual owners and enabling seamless transitions when ownership changes.
Solution Approach 2:
The system uses feedback mechanisms to continuously update the machine learning model with actual data usage information from defect resolution processes. This feedback loop ensures that the model maintains accurate and up-to-date knowledge about required data, preserving this knowledge even when reference code owners change or leave the organization.
Data Source
AI summary
Model-based updating of call home data includes receiving problem analysis data associated with a computing system, and determining one or more reference codes associated with one or more defects within the problem analysis data. A portion of problem analysis data to utilize as training data for a model is determined based on data usage associated with the one or more defects. A representation of each of the one or more reference codes within the portion of problem analysis data is input to the model. The model is configured to output one or more data confidence scores based on the one or more reference codes. The association of the one or more reference codes with the one or more defects is updated based on the one or more data confidence scores.