Seq2Seq Hierarchical Classification for Ticket Logging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional helpdesk systems face inefficiencies in ticket logging due to inconsistent and incorrect labeling, which prolongs the time taken to resolve issues and increases the number of tickets raised, as users must manually assign hierarchical labels, often leading to human error and incomplete descriptions.
Innovation Solution
A processor-implemented method using a Sequence to Sequence (Seq2Seq) Hierarchical Classification Model and Seq2Seq Slot Filling Model to automatically process problem descriptions, assign hierarchical class labels, and dynamically update confidence scores based on user responses, enabling the system to ask relevant questions and log tickets accurately.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If users manually assign hierarchical labels to tickets, then tickets can be routed to the right domain expert, but labeling accuracy decreases and time consumption increases due to human error and complexity
Solution Approach 1:
The system performs automatic ticket labeling without requiring user intervention. The Seq2Seq Hierarchical Classification Model autonomously processes problem descriptions and assigns hierarchical labels, eliminating manual labeling operations and their associated errors and time consumption.
Solution Approach 2:
The patent replaces the mechanical manual labeling process with an automated machine learning system. The Seq2Seq model with attention mechanisms substitutes human cognitive processing, automatically analyzing problem descriptions and generating accurate hierarchical labels through computational algorithms.
2Productivity
If users provide incomplete problem descriptions, then ticket logging is faster, but labeling accuracy decreases making it hard for supervised algorithms to learn
Solution Approach 1:
The system performs preliminary analysis of the problem description to identify missing information before final label assignment. The attention mechanism detects which parts of the description are insufficient, and the system proactively seeks additional information through conversational queries, ensuring complete data before labeling.
Solution Approach 2:
The system implements a feedback loop where the Seq2Seq model analyzes the problem description, identifies missing information, and engages in conversational exchanges with users to obtain additional details. This feedback mechanism ensures that labeling accuracy is maintained without sacrificing logging speed, as the system only proceeds when sufficient information is available.
3Measurement precision
If domain experts modify ticket labels after discussion with users, then labeling accuracy improves, but data consistency decreases creating incorrect labeling data
Solution Approach 1:
The system autonomously handles label assignment and updates without requiring domain expert intervention. The Seq2Seq Hierarchical Classification Model continuously processes problem descriptions and updates labels automatically, eliminating the need for manual modifications by experts and preventing data inconsistency between original and modified labels.
Solution Approach 2:
The patent creates a universal labeling system that handles all ticket classification tasks through a single automated model. This multi-functional approach replaces multiple human experts with one consistent AI system that applies the same labeling logic across all tickets, ensuring data consistency while maintaining accuracy through the attention mechanism and hierarchical classification.
4Adaptability or versatility
If a large number of hierarchical labels are used to cover all IT domains, then ticket routing coverage improves, but system complexity increases making it non-trivial to describe issues completely
Solution Approach 1:
The patent segments the complex hierarchical labeling task into manageable components using the Seq2Seq architecture with attention mechanisms. The model processes problem descriptions sequentially, focusing attention on relevant parts of the text and generating labels hierarchically from top-level categories to specific sub-categories, making the complex routing system tractable.
Solution Approach 2:
The system transforms the complex multi-dimensional label assignment problem into a sequential processing task. By using the Seq2Seq model with attention mechanisms, the system processes the problem description in one dimension (textual sequence) and outputs labels in another dimension (hierarchical structure), simplifying the mapping between issue descriptions and the extensive label taxonomy.
Data Source
AI summary
Users have to assign labels to a ticket to route to right domain expert for resolving issue(s). In practice, labels are large and organized in form of a tree. Lack in clarity in problem description has resulted in inconsistent and incorrect labeling of data, making it hard for one to learn/interpret. Embodiments of the present disclosure provide systems and methods that identify relevant queries to obtain user response, for identification of right category and ticket logging there. This is achieved by implementing attention based sequence to sequence (seq2seq) hierarchical classification model to assign the hierarchical categories to tickets, followed by a slot filling model to enable identifying/deciding right set of queries, if the top-k model predictions are not consistent. Further, training data for slot filling model is automatically generated based on attention weight in the hierarchical classification model.