Call Center Topic Classification via LDA and Stop Word Removal
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Call centres face challenges in processing large volumes of voice call data, making it difficult to effectively label calls and forecast future calls using traditional methods like point estimate models and time series approaches.
Innovation Solution
A computer-implemented method using a topic classifier model to classify communication topics by preprocessing voice-to-text data, removing stop words, generating a bag of words model, and applying latent Dirichlet allocation (LDA) for topic identification, linked with other data sources for prediction and clustering to anticipate call purposes and volumes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional methods like point estimate models and time series approaches are used, then call labeling and forecasting can be performed, but the processing of large volumes of call data is difficult and inefficient
Solution Approach 1:
The patent segments call data into distinct components: call metadata, call transcripts, and customer interaction details. This segmentation allows each component to be processed independently using appropriate techniques, improving overall processing efficiency while managing complexity through modular approaches.
Solution Approach 2:
The patent introduces machine learning models as intermediary components between raw call data and analytical outputs. These models act as mediators that automatically process and transform unstructured call data into structured insights, reducing the manual processing burden and improving efficiency.
2Measurement precision
If traditional call labeling methods are used, then calls can be identified and categorized, but accurate prediction of future call patterns is difficult
Solution Approach 1:
The patent implements preliminary action by training machine learning models on historical call data in advance. This pre-training enables the system to rapidly analyze new calls without performing extensive analysis in real-time, thereby improving prediction accuracy while reducing analysis time for future calls.
Solution Approach 2:
The patent incorporates feedback mechanisms where prediction results are continuously refined based on actual call outcomes. This feedback loop allows the system to learn from past predictions and improve future accuracy, while the automated nature of the feedback process minimizes additional time requirements.
3Quantity of substance
If manual call data processing is used, then detailed analysis can be performed, but the volume of data that can be processed is limited
Solution Approach 1:
The patent replaces manual mechanical processing of call data with automated machine learning systems. This substitution enables the processing of large volumes of call data without proportionally increasing time requirements, as the automated systems can parallelize processing and operate continuously without fatigue.
Data Source
AI summary
A system receives data associated with a communication between one or more individuals. The data is split between each of the one or more individuals into text associated with that individual. Each of the text is modified to remove stop words and to duplicate key words. The text is merged to form a text corpus, from which a bag of words model is generated. Topics of the bag of words are classified using a topic classifier model. A purpose is identified based on the returned topic and keywords from the topic classifier model. Returned topics and keywords from the topic classifier model are linked to the communication.