Peter Charles Wallis, PhD

P. Wallis, Tremarden, Dale End, Bradwell, S33 9HP, UK
Phone: +44 (0)143 3620889, or Mobile: 0791 0059 137
Research Interest: Artificial Intelligence - Machines, human languages & understanding.
My quest is to give conversational agents social intelligence. The focus in the NLP literature is on the information "carried" by language, but utterances are said by someone for an audience. Both speaker and listener will be performing roles and be embedded in a cultural network of norms that must be followed. "Can I borrow 50p?" is okay asked of a friend, but "can I borrow £500?" is not without explanation. There are many thousands of such examples but what makes some "the same" as others? Is there some common underlying structure and if there is, what information is needed to make Machine Learning figure it out?


1995, Doctor of Philosophy
(Semantic Signatures for Information Retrieval)
- RMIT University (Melbourne, Australia)
1987, Bachelor of Arts (2A Hons)
(Philosophy and Computer Science (honours in CS))
- Flinders University of South Australia
1976, Matriculation
(Maths I, Maths II, Physics, Chemistry, Geography)
- South Australia
2000, Graduate Certificate in Management (Science Leadership)
- University of South Australia
1976-1981 Apprentice Fitter&Turner
- Weapons Research Establishment, Australia

Employment History

2013 -  Visiting Research Fellow
Centre for Policy Modelling, Manchester Metropolitan University
2006 - 2012 Senior Research Fellow
Department of Computer Science, University of Sheffield
2003 - 2005 Managing Director Ramjet Software Pty Ltd
2001 - 2003
Software Consultant
Agent Oriented Software Pty Ltd
1995 - 2001
Research Scientist
Defence Science and Technology Organisation, Australia
1989 - 1995 PhD Student Dept of Computer Science, RMIT, Melbourne, Australia
1989 Research Assistant Department of Computer Science, University of Queensland
1983 - 1988
Flinders University, South Australia
1981 - 1983
Technical Assistant
Department of Chemical Engineering, Adelaide University
1976 - 1981
Apprentice Weapons Research Establishment, Australian Defence Forces



FP7-ICT call3, Cognitive Systems and Robotics Principle Investigator
Social Engagement with Robots and Agents (SERA)
With OFAI (Austria - coordinating partner) University Twente (Netherlands) and University Duisburg-Essen (Germany).
The primary objective of the SERA project was to gather recordings of genuine human-robot interaction and then, hopefully, develop some kind of explanatory theory based on real data. Payr and I set out to use qualitative methods but we broadened the range of methodologies under investigation - the Twente people are classic HCI, and those at Duisburg-Essen have a psychology background. Our technique has been to put a Nabaztag "internet connected rabbit" in older people's homes with the ostensive purpose of encouraging exercise. We have collected 300 or so video recorded interactions with associated logs and post event interviews. Although we have not reached consensus on methodology, we have a follow-on proposal in train for the next round of EU funding.
EPSRC Digital Economy: Feasibility Studies in novel ICT developments Co-investigator
Engineering Natural Language Interfaces: can CA help?
April 2008 to March 2009 with Mark Hepple as Principle Investigator.  Joint work with Newcastle's Department of Education Communication and Languages Science.
In 2000 my team at Defence Science in Australia was given the task of developing a dialog system for a ECA acting as a virtual assistant in a data cave.  We conducted a Wizard of Oz study and found, using a semistructured interview technique, that our subjects needed far less semantics, pragmatics and world knowledge than expected, and focused their effort on being polite and generally managing social relations.  Interviewing people about commonsense behaviour (as opposed to expert behaviour) turns out to be problematic and the ethnomethodological version of Conversation Analysis appeared to be a strong contender for an alternate methodology.  One result from this work was inter annotator agreement of over 95% on adjacency pairs and insertion sequences.  A trivial result according to reviewers but perhaps a solid basis for machine learning of more interesting relationships.
FP6-ICT Facilitator
In 2004 we moved to the UK for my wife's post doc, and I was given a desk at Sheffield University, Department of Computer Science, where I started writing papers and proposals with Yorick Wilks. I had admired Yorick's work from a distance since my PhD on lexical semantics and it was an honour to work along side the group and indeed the Department.  After the unsuccessful Copain proposal, we received 13m euro from the European Commission for the Companions IP.   As "Facilitator," my role was to identify problems and potential stoppages and report them.  Unfortunately the project was heading for disaster (funding was ultimately suspended while management issues were sorted out) and I left to pursue independent funding as described above.

PhD Supervision Experience

Teaching Experience

An Introduction to AI (COM 1070) Sheffield, 1st year CS students (80 students)
An Introduction to AI
Object Oriented Design & Programming using Java (COM 162) Sheffield, 2nd & 3rd year Service subject (50)
Human Language Technology Methods, Tools and Applications (TA for Roger Moore)
ART Course on Spoken Dialogue Systems (TA for Marilyn Walker)
1990's ...
Software Project Management, Final year CS students (40) Melbourne University
An Introduction to Natural Language Processing, 4th year CS students (10) Melbourne University
Computers and Society RMIT University - a university wide context subject
Operating Systems Principles I & II second year CS students (120) RMIT University
Introduction to Computers (service subject) RMIT

Memberships, Committees, etc.


Journal articles and book chapters


Refereed conference and workshop papers

Internal reports etc