Support Summary Generation via Sentiment and Semantic Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Technical support technicians face challenges in efficiently reviewing and extracting relevant information from multiple interactions with users, leading to time-consuming processes and potential human errors.
Innovation Solution
A method and system for generating a support summary by tokenizing natural language data from support chats, performing sentiment and semantic analysis, and using a language model to create a concise summary of support issues and relevant information.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If support technicians manually review prior user interactions and technician notes to assess support issues, then they can extract relevant information, but the process becomes time-consuming and vulnerable to human error
Solution Approach 1:
The system performs preliminary analysis of user interactions and technician notes before the support technician needs to review them. Language models pre-extract relevant information, identify support issues, and generate summaries in advance, so when the technician reviews the case, the critical information is already organized and highlighted, reducing both time and error risks
Solution Approach 2:
An automated language processing system acts as an intermediary between the raw support data and the human technician. This intermediary uses NLP techniques to parse, understand, and structure the unstructured text data from chat logs and notes, presenting processed insights to the technician rather than raw data, thereby improving accuracy while reducing manual review time
2Productivity
If support technicians review large volumes of historical data quickly, then they can respond faster to user needs, but they risk missing relevant information and making errors
Solution Approach 1:
The system extracts only the most relevant information from large volumes of historical data using language models and NLP techniques. Instead of requiring technicians to review all chat logs and notes, the system identifies and extracts key support issues, relevant facts, and important context, presenting a curated subset that maintains reliability while enabling faster response
Solution Approach 2:
The patent replaces the mechanical process of manual human review with an automated language processing system. The language model automatically parses, understands, and extracts information from historical data, substituting human cognitive processing with computational analysis that can handle large volumes of data quickly and reliably without fatigue or distraction
3Loss of information
If support technicians repeat questions or request data users have already provided, then they can ensure complete information collection, but this increases user impatience and frustration
Solution Approach 1:
The system implements feedback by automatically analyzing prior user responses and technician notes to identify what information has already been collected. This feedback loop ensures that when a support technician interacts with a user, the system provides real-time alerts or suggestions about previously provided information, preventing redundant questions and maintaining user satisfaction while ensuring completeness
Data Source
AI summary
A method of generating a support summary includes extracting chat information including natural language data from a support chat with a user. Tokenized language is generated by performing feature extraction on this natural language data. This tokenized language is subjected to sentiment analysis to produce sentiment data reflecting sentiment of the support user during the support chat, and to semantic analysis to extract support-relevant features. A support summary made up of natural language text identifying a support issue and information germane to the support issue is then generated from the extracted chat information using a language model, at least in part from the extracted support-relevant features and the sentiment data.