Predicting Customer Call Intent for the Auto Dealership Industry from Analyzing Phone Call Transcripts With CNN for Multiclass Classification

, und . International Journal on Soft Computing, Artificial Intelligence and Applications (IJSCAI) 8 (3): 13 (August 2019)


Auto dealerships can receive thousands of inbound customer calls daily for a variety of reasons, or intents, including sales, service, vendor inquires, and job seeking. Given the high volume of calls, it is very important for auto dealers to precisely understand the intent of these calls to provide positive customer experiences. Positive interactions can ensure customer satisfaction and deeper customer engagement leading to a boost in sales and revenue, and even the optimum allocation of agents or customer service representatives across the business. In this paper, we define the problem of customer phone call intent as a multi-class classification problem stemming from the massive data set of recorded phone call transcripts. To tackle this problem, we develop a convolutional neural network (CNN)-based supervised learning model for semantic text analysis to classify customer calls into four intent categories: sales, service, vendor and jobseeker. Experimental results show that with the ability of our scalable data labeling method to provide sufficient training data, our CNNbased predictive model performs very well on long-transcript text classification, according to the model’s quantitative metrics of F1-Score, precision, recall and accuracy on the testing data

Links und Ressourcen

Suchen auf:

Kommentare und Rezensionen  

Es gibt bisher keine Rezension oder Kommentar. Sie können eine schreiben!


Zitieren Sie diese Publikation