Hi, I have a general question about how Mahout might be used to create a classification system…
The Problem: We sell a product for which Customer Support plays a huge role. We offer a "Contact Us" page that serves as the gateway through which customers express to us they have an issue and request the assistance of a customer support agent. Currently, we ask customers to choose a problem category, based on that problem category, we offer a selection of relevant FAQ articles that might quickly address the issue. Assuming the customer's issue is still unresolved, we then ask the customer to fill out a form and describe there issue in a few sentences. When the form is submitted, the customer is placed in a queue where they will eventually be able to communicate with a support agent (either via Phone or via Instant Messaging). The problem is that we are asking the customer to do more work than is necessary. Everything we need to know about the customer's issue is likely captured in the customer authored problem statement. Thus, I am exploring options for intelligent analysis of this text that would lead to a more streamlined "Contact Us" experience. My Specific Question: Can Mahout's classification system be used to classify a customer authored problem statement in real time? I am imagining a system that would periodically train a classifier in an offline fashion and then leverage that classification index to provide real-time classifications of customer problem statements as they are received. Is this possible? Thanks, Tristan
