ML Event Tree Workflow Visualization for Customer Service Diagnostics
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing workflow engines for customer service problem resolution are hindered by a linear and sequential diagnostic process, leading to lengthy and often inaccurate solutions, high repeat call rates, and increased operating costs.
Innovation Solution
A system that utilizes machine learning-enabled event trees to generate workflow construction specifications, which are then translated into workflow visualization interpretation files to assist customer service agents in understanding the logic behind machine learning recommendations, and dynamically builds personalized workflows for problem resolution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a linear and sequential diagnostic process is used, then the workflow is simple to implement, but the problem resolution time increases and accuracy decreases
Solution Approach 1:
The patent implements a dynamic diagnostic workflow that transitions from linear/sequential to event-driven/parallel execution. The workflow engine continuously monitors system events and dynamically adjusts the diagnostic path based on real-time conditions, allowing multiple diagnostic steps to execute in parallel rather than sequentially, thereby reducing resolution time while maintaining implementation feasibility through automated event listening and response mechanisms.
Solution Approach 2:
The patent establishes continuous monitoring of system events throughout the diagnostic process. Instead of discrete sequential steps, the system maintains continuous observation of system state changes, enabling immediate detection and response to events. This continuous action allows the diagnostic process to progress without idle sequential transitions, reducing overall resolution time while the workflow engine manages the complexity of coordinating continuous monitoring across multiple systems.
2Device complexity
If a linear and sequential diagnostic process is used, then the workflow is easy to manage, but the number of repeat calls increases due to inaccurate solutions
Solution Approach 1:
The patent implements feedback mechanisms where the workflow engine monitors the outcomes of diagnostic actions and system events, using this information to adjust and refine subsequent diagnostic steps. The system learns from repeated events and outcome data to improve the accuracy of root cause identification, thereby reducing repeat calls. The feedback loop allows the workflow to adapt to real-world conditions and improve solution accuracy over time while the engine manages the complexity of feedback coordination.
Solution Approach 2:
The patent replaces manual sequential diagnostic steps with automated machine learning-based event analysis. The workflow engine uses ML algorithms to analyze system events, identify patterns, and determine root causes automatically, substituting the mechanical sequential process with an intelligent automated system. This substitution significantly improves solution accuracy by leveraging pattern recognition and data analysis, while the workflow engine manages the complexity of coordinating these intelligent systems.
3Speed
If machine learning provides root cause recommendations, then diagnostic speed improves, but agent trust in the recommendations decreases due to lack of transparency
Solution Approach 1:
The patent introduces an explanation module as an intermediary between the machine learning system and the customer service agent. This intermediary component translates the black-box ML recommendations into human-understandable explanations, showing the logical reasoning and evidence behind each root cause identification. The explanation module acts as a bridge that maintains the speed advantage of ML while building agent trust through transparency, as agents can understand and verify the reasoning process.
Solution Approach 2:
The patent implements visualization components that provide visual representations of the diagnostic reasoning process, using visual indicators, flow charts, and highlighted paths to make the ML decision-making process transparent. These visual elements change the 'information color' from opaque algorithmic output to clear visual narratives that agents can easily interpret. The visualization system maintains rapid diagnostic speed while significantly improving agent trust by making the invisible ML reasoning process visible and understandable.
4Device complexity
If pre-configured workflows are used, then the implementation is straightforward, but the ability to personalize for different customers is limited
Solution Approach 1:
The patent transforms static pre-configured workflows into dynamic, customer-specific workflows. The workflow engine automatically adapts the diagnostic path based on individual customer profiles, historical data, and real-time system events. Each customer receives a customized diagnostic sequence tailored to their specific conditions and preferences, maintaining implementation simplicity through automated adaptation rather than manual configuration for each scenario.
Solution Approach 2:
The patent changes workflow parameters dynamically based on customer-specific data. The system adjusts diagnostic priorities, selected procedures, and communication styles according to individual customer profiles and historical patterns. By modifying workflow parameters rather than creating entirely new workflows for each customer type, the system achieves high personalization while maintaining implementation simplicity through parameter-based adaptation.
Data Source
AI summary
Concepts and technologies disclosed herein are directed to interpretation workflows for machine learning-enabled event tree-based diagnostic and customer problem resolution. According to one aspect, a system can receive a workflow construction specification derived from a machine learning-enabled event tree (“MLET”). The MLET can be generated for use by a customer service agent to resolve a customer problem. The workflow construction specification can include a plurality of objects, each of which represents a navigation path through the MLET. The system can traverse the workflow construction specification and can create a set of workflow creation commands based upon at least one policy. The system can generate a workflow visualization interpretation file based upon the set of workflow creation commands. The workflow visualization interpretation file can identify how the MLET derived a root cause of the customer problem. The system can then present the workflow visualization interpretation file to the customer service agent.