Noisy Ticket Classification via Syntactic Grammar and Event Correlation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing ticket analysis approaches are inadequate for handling noisy and incomplete ticket text, which often results from human operators being measured on resolution time, leading to challenges in interpreting unstructured texts with existing grammar-based systems.
Innovation Solution
A computer-implemented method that correlates ticket data with event data to generate enriched data, applies syntactic grammar to identify domain-dependent and independent concepts, and uses noisy natural language processing (NLP) to parse and classify ticket text, categorizing tickets into contextually disjointed and non-disjointed categories.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing grammar-based systems are used to analyze ticket text, then the system complexity is low, but the accuracy of interpreting noisy and unstructured text deteriorates
Solution Approach 1:
The patent segments the ticket analysis process into multiple specialized components: correlation component for matching events with tickets, syntactic analysis component for grammar-based parsing, noisy NLP component for handling unstructured text, and type identification component for classification. Each component handles a specific aspect of the analysis, improving overall accuracy while managing complexity through modular architecture.
Solution Approach 2:
The patent introduces intermediate processing stages between raw ticket text and final classification. The correlated text data serves as an intermediary that enriches original ticket information with event context, while derived structures act as intermediate representations that bridge unstructured text and classification outcomes, thereby improving interpretation accuracy.
2Measurement precision
If extensive training data is used to improve classification accuracy, then the measurement precision improves, but the loss of time increases due to longer training periods
Solution Approach 1:
The patent performs preliminary enrichment of ticket text by correlating it with event data before classification. This preliminary action adds contextual information and structured data to otherwise noisy tickets, enabling more accurate classification with less training data required, thereby reducing the time needed for model training while maintaining or improving accuracy.
3Productivity
If ticket text is kept concise to maintain fast resolution, then the productivity is high, but the loss of information increases due to incomplete descriptions
Solution Approach 1:
The patent merges ticket text with correlated event data to create enriched correlated text data. This combination preserves the conciseness of original ticket descriptions while supplementing them with detailed event information, logs, and contextual data, thereby maintaining fast resolution processes while compensating for information loss in brief descriptions.
Data Source
AI summary
Methods, systems, and computer program products for structured representation and classification of noisy and unstructured tickets are provided herein. A method includes correlating one or more items of problem incident text data from a given problem incident identifier with items of event text data to generate items of correlated text data within the given problem incident identifier; applying a syntactic grammar to the items of correlated text data to identify in the given problem incident identifier (i) one or more domain-dependent concepts and (ii) one or more domain-independent concepts; parsing noisy and/or unstructured text from the given problem incident identifier based on analysis of (i) the domain-dependent concepts and (ii) the domain-independent concepts; and classifying the given problem incident identifier by classifying features derived from text of the given problem incident identifier subsequent to said parsing of said noisy and/or unstructured text from the given problem incident identifier.