This document explains how to optimize quality inspection rules in Smart Conversation Analysis.
Feature overview
During the review of a call center or dataset task, you can assess the correctness of each rule. If a rule is incorrect, you can provide details about the issue and suggest improvements. The system triggers an optimization recommendation after the number of reviews for a rule reaches a specific threshold.
A rule optimization recommendation is triggered after approximately 200 reviews. This threshold is a built-in system value and cannot be changed in the console.
Accessing the feature
After the review threshold is met, a list of optimization recommendations appears on the Call Center Tasks, Dataset Tasks, and Rule Configuration pages. The list shows the rule's accuracy rate before and after the proposed optimization. Click View to review the details of the optimized rule and its new accuracy rate. Click the Decline Optimization button to cancel the recommendation. Click Accept and View Verification Task to add the task to the Rule Optimization page. You can then review and annotate the inspection results from the optimized rule and further refine the rule based on the standard data in the task.