Automatically Generated Time-Series Labels for Event Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing time series analytics and machine learning systems struggle to provide meaningful interpretation across different applications due to complex algorithms that are difficult to understand and apply outside their specific contexts, limiting their adaptability and effectiveness in detecting anomalies and forecasting.
Innovation Solution
A framework that automatically generates descriptive labels for time series data points, allowing users to define custom queries and rules based on these labels, and integrates artificial intelligence to learn and refine patterns for enhanced analytics and automated responses.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If complex algorithms are used for anomaly detection and forecasting, then detection accuracy is improved, but interpretability and ease of operation deteriorate
Solution Approach 1:
The patent introduces natural language processing as an intermediary layer between the user and complex analytical algorithms. Users can define queries using simple natural language descriptions, and the system automatically translates these into complex analytical operations, thereby maintaining high detection accuracy while significantly improving ease of operation.
Solution Approach 2:
The patent replaces manual configuration of complex algorithmic parameters with automated natural language interpretation. Instead of requiring users to manually configure complex detection algorithms, the system uses NLP to automatically interpret user intent and configure the appropriate analytical operations.
2Measurement precision
If application-specific algorithms are used, then detection accuracy for that application is improved, but adaptability to other applications deteriorates
Solution Approach 1:
The patent creates a universal natural language interface that can handle multiple applications and domains. The system is designed to interpret user intent across different application contexts (e.g., server monitoring, financial data, healthcare) without requiring application-specific configuration, thereby achieving both accuracy and adaptability.
Solution Approach 2:
The patent implements dynamic adaptation where the system learns from user interactions and adjusts its interpretation capabilities across different applications. The natural language processing engine dynamically adapts to different domains while maintaining a consistent interface, allowing the system to be effective across multiple applications.
3Loss of information
If manual analysis of time series data is performed, then interpretability is improved, but productivity and time consumption deteriorate
Solution Approach 1:
The patent enables the system to automatically perform complex analytical operations in response to natural language queries. The system self-services by automatically interpreting user intent, configuring appropriate algorithms, executing analysis, and presenting results without requiring manual intervention at each step, thereby maintaining interpretability while dramatically improving productivity.
Solution Approach 2:
The patent performs preliminary processing and interpretation of user queries before executing complex analyses. The natural language processing component prepares and structures user intent in advance, allowing the system to efficiently execute analytical operations and deliver results quickly while maintaining full interpretability of the analysis process.
Data Source
AI summary
Techniques for performing analytics using automatically generated labels for time series data and numerical lists are disclosed. In some embodiments, a system loads a set of one or more time series datasets. A respective time series dataset may include a set of data points based on varying values of a metric of one or more computing resources over a window of time. The system assigns labels to a subset of the data points in the time series datasets. The label assigned to a given data point may be descriptive of a pattern reflected by the data point relative to other data points in the time series. The system further identifies a pattern of automatically assigned labels that is indicative of an event affecting the one or more computing resources. Responsive to identifying the pattern of labels, the system may trigger a responsive action.