Conversation analysis should be a part of every conversation design team’s workflow. You may be doing it already without realising exactly what it is.
I reached out to Dr Maria Aretoulaki to find out exactly what conversation analysis is, why it matters and how she would use it for a new project.
Maria is the Director at DialogCONNECTION Ltd and she works with GlobalLogic (a HITACHI Group company) where she’s SME for ML/AI, design-led and data-driven Voice & Conversational AI (VOCAI) and VUI and CUI Design, and a Coordinator of the Web3 and Blockchain CoE.
Maria is a true veteran of conversational AI, so it was fantastic to chat with her!
In this episode
00:20 Maria’s Introduction
02:38 Maria’s VUI experience
05:33 30 seconds, go!
07:01 On ‘copywriting’ as a term for conversation design
08:45 What would you call the craft of utterance writing?
12:18 what surprised you most when you started designing conversations?
15:04 Customers creatively swearing at bots
17:09 What is conversation analysis?
22:03 What is ontology?
26:33 Ontology of human experience
30:10 Where would you expect to use conversation analysis in a new project?
33:10 What would you look for in the data?
36:09 Are recordings better than transcriptions?
38:48 Do transcriptions give enough detail?
43:20 How much detail is enough?
44:48 TTS and voice talent
48:08 Should conversation analysis adapt to different modalities?
51:58 Are ontologies intents?
53:51 Where can people learn more about conversation analysis?
57:51 Where can people follow you?