Asynchronous Messaging Topic Classification for Contact Centers
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Asynchronous communication channels lack an indication of session end, making it difficult to determine if an interaction has concluded, which hinders efficient management and resumption in contact centers.
Innovation Solution
Implementing a system that analyzes customer interactions to differentiate between resuming a previous conversation topic and starting a new one, using topic modeling and confidence scores to determine the intent behind new messages, allowing for seamless interaction resumption and agent allocation based on skill-based routing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If asynchronous communication channels are used to enable flexible customer interactions, then adaptability and customer convenience are improved, but the ability to determine session end and manage interactions efficiently deteriorates
Solution Approach 1:
The system performs preliminary analysis of customer messages by evaluating topic confidence scores before routing decisions are made. This allows the system to proactively determine whether a message indicates a new topic or continuation of existing conversation, enabling efficient session management without requiring explicit end-of-session signals from customers.
Solution Approach 2:
The system implements feedback mechanisms by continuously monitoring conversation topics and confidence scores to dynamically adjust session management decisions. The topic confidence scores provide feedback about conversation state, enabling the system to adaptively determine when interactions should be transferred or concluded without explicit customer indication.
2Measurement precision
If topic modeling and confidence score analysis are implemented to determine conversation intent, then interaction management precision is improved, but system complexity increases
Solution Approach 1:
The system segments the complex task of conversation analysis into distinct components: topic modeling, confidence score calculation, and routing decision-making. This segmentation allows each component to be optimized independently while maintaining overall system precision in determining conversation intent.
Solution Approach 2:
The system introduces topic confidence scores as an intermediary metric that bridges raw message content and routing decisions. This intermediary layer simplifies the overall system architecture by providing a standardized measure that can be used to make consistent routing decisions without requiring complex direct analysis of conversation content.
3Measurement precision
If the system analyzes each new message to determine topic continuity, then interaction accuracy is improved, but processing time increases
Solution Approach 1:
The system changes the parameter used for topic determination from analyzing entire conversation histories to evaluating topic confidence scores based on current message context. This parameter change maintains high accuracy in topic determination while significantly reducing processing time by focusing on relevant features rather than complete historical analysis.
Data Source
AI summary
Embodiments of the disclosure provide a method of processing messages received in an asynchronous communication system. In some embodiments, the method includes receiving a message from a customer communication device, determining that a conversation is already established in association with the customer communication device, including the message among a plurality of messages that are already assigned to the conversation, analyzing the message to determine a topic classification and a topic confidence score for the message, and based on the analysis of the message, determining whether the topic classification determined for the message corresponds to a continuation of a topic classification for the plurality of messages or whether the topic classification determined for the message corresponds to a different topic classification than the topic classification for the plurality of messages.