Vector Embedding for Failure Detection in Metric Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing failure detection methods struggle to quickly identify relevant metrics from a large number of metrics, often dropping user-intended failure metrics during the pruning process, which prevents accurate reflection of user intention.
Innovation Solution
A failure information detecting apparatus that acquires multiple metrics and a failure query, encodes them into vector representations, and calculates similarity to detect failures based on user intention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If the number of metrics is reduced to improve failure detection speed, then detection speed is improved, but user-intended failure metrics are dropped
Solution Approach 1:
The patent introduces an embedding model as an intermediary that transforms both the metrics data and user queries into a common vector space. This mediator enables semantic matching between user intentions and metrics without requiring direct filtering or pruning, thus preserving user-intended metrics while achieving fast detection through vector similarity computation.
Solution Approach 2:
The patent changes the parameter representation from raw metric values to embedded vector representations. By transforming metrics and queries into vector space, the system enables efficient similarity computation while preserving the semantic meaning of user-intended metrics, resolving the contradiction between speed and information loss.
2Ease of operation
If traditional pruning methods are used to extract relevant metrics, then analysis is facilitated, but user intention cannot be reflected
Solution Approach 1:
The patent replaces traditional mechanical pruning methods (filtering, clustering, dimensionality reduction) with a semantic embedding approach. Instead of mechanically removing metrics based on statistical criteria, the system uses vector similarity to implicitly identify and focus on user-intended metrics, preserving user intention while facilitating analysis.
3Reliability
If a large number of metrics are analyzed, then comprehensive coverage is achieved, but detection time increases
Solution Approach 1:
The patent performs preliminary embedding of metrics into vector representations before actual failure detection. This pre-processing step organizes the large number of metrics in a structured vector space, enabling rapid similarity-based filtering during detection without sacrificing comprehensive coverage of user-intended metrics.
Data Source
AI summary
A failure information detecting apparatus according to an embodiment includes a data acquisition unit that acquires a plurality of metrics including time series data and metadata and a failure query indicating a failure that a user desires to specify from a monitored system, a metric encoder that calculates a first vector representation of the plurality of metrics, a failure query encoder that calculates a second vector representation of the failure query, and a failure information detecting unit that calculates a first similarity between the first vector representation and the second vector representation and detects a failure in the metrics based on the first similarity.