QoS Filtering for Call Interaction Evaluation Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems fail to effectively evaluate call interactions due to low network Quality of Service (QoS), leading to poor customer sentiment and inaccurate agent ratings, as they do not consider multiple parameters influencing QoS before forwarding recordings for evaluation.
Innovation Solution
A computerized system in a cloud-based environment that includes a Session Border Controller to regulate network QoS, store metadata, and use an Interaction Manager to extract and index Mean Opinion Score (MOS) data, preventing evaluations of recordings with poor QoS by distributing only interactions with scores above a predefined threshold for evaluation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If all call interactions are distributed for evaluation without QoS filtering, then complete evaluation coverage is achieved, but computing resources and distribution time increase significantly
Solution Approach 1:
The system performs QoS assessment and filtering of call interactions before distributing them for evaluation. The QoS metadata is collected and stored in advance, and the filtering logic is applied during the distribution phase, preventing low-quality interactions from entering the evaluation pipeline in the first place
Solution Approach 2:
The system extracts and separates low-quality call interactions from the overall dataset based on QoS metrics. By identifying and removing interactions with poor network quality, packet loss, or other degradation indicators, the system creates a refined subset of high-quality interactions for evaluation, eliminating the need to process irrelevant low-quality data
2Productivity
If QoS filtering is applied before distribution, then computing resources and distribution time are reduced, but evaluation coverage may be limited
Solution Approach 1:
The system dynamically adjusts QoS threshold parameters to control the filtering strictness. By modifying thresholds for metrics such as packet loss percentage, network latency, or MOS scores, the system can flexibly balance between filtering out low-quality interactions and maintaining sufficient evaluation coverage based on operational requirements
3Measurement precision
If packet loss ratio is used as the sole QoS metric, then network layer issues are detected, but overall call quality assessment is insufficient
Solution Approach 1:
The system implements a multi-dimensional QoS assessment framework that collects and evaluates multiple parameters simultaneously, including packet loss ratio, network latency, jitter, and Mean Opinion Score (MOS). This comprehensive approach allows the system to detect both network layer issues and higher-layer call quality problems, providing a complete picture of interaction quality
Solution Approach 2:
The system creates a composite QoS metric by combining multiple individual metrics (packet loss, latency, jitter, MOS) into an overall quality assessment. This composite approach synthesizes information from different layers of the communication stack, enabling more accurate identification of interactions that truly reflect poor customer experience
Data Source
AI summary
A computerized system in a cloud-based computing environment for preventing an evaluation of agent recording segments having bad customer sentiment due to low network Quality of Service (QoS), is provided herein. The system may include a Session Border Controller to send metadata related to a regulated network QoS to a database; a media server to store the call interaction; an Interaction Manager (IM) module to extract metadata related to the regulated network QoS; an Indexer Micro-Service to listen to real-time data streaming service and to index and store the metadata related to the regulated network QoS, when new metadata arrives; a Multi-Channel Recording unit to retrieve: an interaction call; and indexed metadata from the document-oriented database; a Quality Planner Micro-Service to: read MOS score for the call interaction; distribute a call interaction for evaluation when the read MOS score is above a predefined threshold, according to a predefined quality plan.