Multi-Participant Sentiment Modeling for Context-Aware Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for sentiment analysis in contact centers primarily focus on individual participant sentiments without considering the synergistic effect of analyzing utterances from different participants in a discussion, leading to inaccurate insights.
Innovation Solution
A method and system that concurrently analyze utterances from multiple participants in a discussion using a composite machine-learning model, incorporating an attention-based encoder and sentiment classification models tailored to each participant type, to enhance sentiment classification accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If existing sentiment analysis methods focus on individual participant sentiments, then the analysis process is simpler, but the sentiment classification accuracy is insufficient
Solution Approach 1:
The patent segments the interaction into multiple textual segments, each associated with a specific participant type (e.g., customer, agent). This segmentation allows the system to analyze sentiments of different participants separately while maintaining their individual characteristics, thereby improving sentiment classification accuracy without overwhelming complexity
Solution Approach 2:
The patent combines the analysis of multiple participants' sentiments into a unified framework. By concurrently analyzing utterances from different participants and considering their interactions, the system achieves more accurate sentiment classification than individual analysis while managing complexity through integrated processing
2Measurement precision
If the system analyzes utterances in isolation, then the processing speed is faster, but the sentiment interpretation is less accurate
Solution Approach 1:
The patent performs preliminary actions by obtaining and labeling multiple textual segments from different participants before conducting the actual sentiment analysis. This preparation includes identifying participant types and organizing utterances in advance, which enables more accurate context-aware sentiment interpretation while optimizing processing efficiency
3Adaptability or versatility
If the system uses a single sentiment classification model, then the model complexity is lower, but the adaptability to different participant types is reduced
Solution Approach 1:
The patent applies local quality by using different sentiment classification models or different classification criteria for different participant types. For example, customers may be evaluated on positive/negative sentiment while agents are evaluated on empathetic/unhelpful behavior. This targeted approach enhances adaptability to each participant type's specific characteristics while managing overall system complexity
Data Source
AI summary
A system and method of determining a sentiment of a participant in an interaction may include: obtaining a plurality of textual segments, each representing a portion of the interaction, and labeled according to a specific participant; inferring a language model on one or more textual segments of the plurality of textual segments, to generate respective semantic embedding vectors, each representing a semantic meaning of the respective textual segment in a semantic vector space; compiling a semantic vector set that includes (i) a target semantic embedding vector, corresponding to a target textual segment of a target participant, and (ii) one or more peripheral semantic embedding vectors, respectively corresponding to one or more peripheral textual segments of the plurality of textual segments; and inferring a composite machine-learning (ML)-based model on the semantic vector set, to classify a sentiment of the target participant, as expressed in the target textual segment.