We've had this conversation before: A Novel Approach to Measuring Dialog Similarity

Dialog is a core building block of human natural language interactions. It contains multiparty utterances used to convey information from one party to another in a dynamic and evolving manner. The ability to compare dialogs is beneficial in many real world use cases, such as conversation analytics for contact center calls and virtual agent design. We propose a novel adaptation of the edit distance metric to the scenario of dialog similarity. Our approach takes into account various conversation aspects such as utterance semantics, conversation flow, and the participants. We evaluate this new approach and compare it to existing document similarity measures on two publicly available datasets. The results demonstrate that our method outperforms the other approaches in capturing dialog flow, and is better aligned with the human perception of conversation similarity. # Conversation 1 Conversation 2 1 Customer: I'd like to look for a film to watch. I like adventure films. Customer: I'd like to search for a fun film to watch. 2 Agent: Where are you located? Agent: What is your location? 3 Customer: Could you look for films shown in Napa? Customer: Could you find films shown in San Ramon for me? 4 Agent: I discovered 1 movie -would you like Dumbo? Agent: What is your take on Breakthrough or Captain Marvel? 5 Customer: Please look for other films. I would like to watch at The Lot City Center. 6 Agent: What is your take on Hellboy, Little, or Missing Link? 7 Customer: I'd love Captain Marvel. When can I watch it? I'd like to watch a regular show. Customer: Little is the one for me. When can I watch it? I'd like to watch it today. 8 Agent: What time would you like to watch it? 9 Customer: I'd like to watch it on the day after tomorrow. 10 Agent: I discovered 1 showtime for the film in Century Napa Valley and XD Theater at 10:30 pm.