Cross-lingual/Cross-channel Intent Detection in Contact-Center Conversations
Abstract
Contact centers play a critical role in providing quality customer service for various businesses. Identification of business-related intents in these conversations is valuable for compliance, quality assurance, and analytics. This is often done using keyword spotting. But, identifying the right key phrases for accurate intent detection is challenging, especially for noisy call transcripts generated by ASR systems. Moreover, the process has to be repeated across different languages & to adapt existing intent definitions from other communication channels to call transcripts. We present a novel Semantic Phrase Search Engine that enables cross-lingual/cross-channel discovery of intent key phrases. The platform facilitates automated expansion of intent key phrases to improve coverage, and in-context translation of phrases across languages/channels, while maintaining semantic integrity. Thereby, our system helps in improving quality as well as reducing cost of contact center operations.