Company Information

Graham Wilcock has a PhD from UMIST (University of Manchester Institute of Science and Technology), England and is Adjunct Professor of Language Technology at University of Helsinki, Finland. He previously worked in industry with ICL (International Computers Limited) in Europe and with Sharp Corporation in Japan. He was co-organizer of several workshops on NLP and XML including the first Linguistic Annotation Workshop (LAW 2007). He received an IBM Innovation Award in 2008 for work on UIMA (Unstructured Information Management Architecture) and published a textbook Introduction to Linguistic Annotation and Text Analytics in 2009.

Since 2016 he has worked on talking robots and conversational AI. With Kristiina Jokinen he developed WikiTalk, a Wikipedia-based open-domain dialogue system for Nao robots, and co-edited a book Dialogues with Social Robots published by Springer in 2017. He developed CityTalk, a prototype generic dialogue system in which social robots interact using conversational AI and answer queries by searching knowledge graphs in graph databases. In 2018-19 he was Visiting Professor at Kyoto University, Japan where he worked with the ERICA robot in the ERATO project. He has presented papers and demos on talking robots at SIGDIAL 2015, COLING 2016, IJCAI 2018, IJCAI 2019, and most recently at HRI 2022, RO-MAN 2022, ACII 2022, RO-MAN 2023, IWSDS 2024 and HRI 2024.

CDM: Constructive Dialogue Modelling

CDM is a framework for interaction management described by Kristiina Jokinen in her book Constructive Dialogue Modelling: Speech interaction and rational agents (2009).

CDM focusses on the enablements for communication (Contact, Perception, Understanding, Reaction) and the view that interaction is dynamically constructed by the participants via evaluating and reacting to the new information (NewInfo) being exchanged. The participants are assumed to be rational, autonomous, and intelligent agents, either humans or situated robots, capable of coherent, multimodal conversation with the most affordable means: natural communication. Evaluation of the partner's message is based on the current context, while reaction also takes into account the agent's own goals which they intend to achieve via the interaction. The goals can be specific task related achievements or quite general social relation and mutual bond related desires, and they may be born by the agent's own aspirations or, as is usually the case with artificial agents, by the design and implementation of the application.