Sentiment-Driven ML Self-Service System for IT Support
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Information technology departments face user frustration due to disorganized self-service systems, leading to increased reopened tickets, as users struggle to find relevant solutions and contact representatives, often resulting in negative experiences.
Innovation Solution
A system utilizing a trained machine learning engine to identify issue categories, retrieve custom solutions from a knowledge base, request feedback, and retrain based on sentiment scores, providing solutions in formats like articles, videos, or scripts to enhance user experience and reduce representative contact.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If a self-service system is provided for users to resolve issues, then representative contact is reduced, but the system becomes disorganized and users experience frustration due to inability to find relevant solutions
Solution Approach 1:
The system captures user feedback through sentiment analysis of communications and support interactions. This feedback is fed back into the machine learning model to continuously improve solution recommendations, creating a closed-loop system that adapts to user needs and preferences over time.
Solution Approach 2:
The system enables users to autonomously resolve issues by providing personalized solution recommendations based on their specific problem descriptions and historical data. Users can access relevant knowledge articles, troubleshooting steps, and resolution guidance without needing to contact representatives.
2Adaptability or versatility
If multiple knowledge articles and applications are provided for general support, then solution coverage is increased, but users experience confusion due to lack of organization and relevant filtering
Solution Approach 1:
The system segments the large knowledge base into relevant portions by analyzing user communications and matching them with appropriate knowledge articles. The machine learning model identifies and presents only the most relevant solutions from the broader knowledge base, breaking down the overwhelming information into manageable, targeted segments.
Solution Approach 2:
The machine learning model serves multiple functions: it categorizes issues, recommends solutions, analyzes sentiment, and personalizes recommendations based on user behavior. This multi-functional approach consolidates what would otherwise require multiple separate systems into a single unified platform.
3Loss of energy
If users are required to search for solutions manually, then system resources are conserved, but user frustration increases and ticket reopening rates increase due to time consumption
Solution Approach 1:
The system performs preliminary actions by proactively analyzing user communications and automatically generating personalized solution recommendations before users need to search manually. The machine learning model prepares and presents relevant solutions in advance, based on the user's problem description and historical data.
Solution Approach 2:
The system replaces manual searching and browsing mechanisms with automated machine learning-based recommendation. Instead of users manually navigating through knowledge articles, the ML model automatically identifies and presents the most relevant solutions, substituting mechanical user effort with intelligent automation.
4Reliability
If feedback collection is implemented for unsuccessful solutions, then solution quality improves through learning, but additional user interaction steps are required
Solution Approach 1:
The system implements automated feedback collection by analyzing user communications and sentiment to determine whether solutions were successful. This feedback is automatically incorporated into the machine learning model to improve future recommendations, creating a self-improving system that learns from each interaction.
Data Source
AI summary
Systems and methods for capturing sentiments and delivering elevated proactive user experience are disclosed. In one embodiment, a method may include a solution recommendation computer program: receiving, from a user electronic device, a message comprising an identifier for a computer issue; identifying, using a trained machine learning engine, a solution category for the computer issue, wherein the trained machine learning engine is trained using historical service data and historical sentiment scores; retrieving a custom solution for the solution category from a knowledge base; determining whether the custom solution was successful; in response to the custom solution being unsuccessful, requesting feedback on the custom solution from the user electronic device; receiving the feedback from the user electronic device; assigning a sentiment score to the custom solution based on the feedback, wherein the sentiment scores is positive, neutral, or negative; and retraining the trained machine learning engine using the sentiment score.