Promise made real

Chris Hutchison reports from Montpellier on a bold attempt to bring together disparate disciplines at the cutting edge of high technology

Forget the Festival de Cannes - it was computer scientists who were spoiled in the south of France recently when under the collective banner of Informatique '95 four conferences ran concurrently in Le Corum conference and exhibition centre in Montpellier.

The four - Interface '95 (Human-Computer Interfaces), Genie Linguistique '95 (linguistic engineering), RPO '95 (object-oriented technologies) and IA '95 (artificial intelligence) made for an intellectual banquet on an awesome scale.

Epic visions were strengthened by the presence of sponsors IBM, Barco, France Telecom, Silicon Graphics, ASV, and INRIA with the combined listings of exhibiting companies, steering committee, and international programme committee ample enough to bring to mind the opening pages of a telephone directory. And the whole caboodle was captured by the ubiquitous camera of French television channel Canal+.

Indeed, had anyone walking into Le Corum been in a sceptical mood, the joint events will have left them in no doubt that the media-mythologised "information society" is, if not yet quite with us, at least just a very short step around the corner. Around the corner, perhaps, in the "interactive demonstration area" of the Espace Antigone in which, past the "Internet self- service area" where four PowerMacs provided free Internet access to conference participants, there were demonstrations of "virtual squash" in a 12ft by 18ft court ending in a Barco screen wall, the screen and space also used as a giant shuffleboard, a music mixing desk, and - for budding Jackson Pollacks and John Cages - a "virtual canvas" for electronic painting and a lac des sons for musical composition.

Overall, the accent of the exhibition, as of much of the conference, was on virtual reality and its applications in industry, education, and "infotainment". Represented by companies such as Virtual Presence, Simulis, Immersion, Virtools, and Superscape Ltd, VR reaffirmed its continued relevance and value in the wake of some disillusionment after the unfulfilled hype of a couple of years ago. I was particularly impressed by the "cultural" exhibits. Douglas Rushkoff, in his 1994 book Cyberia, had invited us to imagine a "virtual classroom" of geographically dispersed students as, each plugged in to a VR system and following their teacher, they fly over the skyline of ancient Rome, visit the Forum, and explore the Colosseum. ENEL, the Italian electric power company, exhibited a selection of their VR products that were effectively already turning Rushkoff's vision into a reality: impressively good high-definition graphics allowed the user to go back in time and "walk" freely through St Peter's Basilica and its piazza, to explore freely the City of Giotto, or to "enter" the tomb of Nefertari, Light of Egypt. With comparable high-quality graphics, the University of Cincinnati presented a virtual tour of the caves of Lascaux. Given that IRL ("in real life") the Constantinian Basilica is long demolished, Nefertari's tomb in dilapidation, and the caves of Lascaux closed to the public, it is hard to overestimate the potential value of these interactive displays as educational and cultural resources.

Elsewhere in the exhibition, two Virtuality 2000SU series immersive virtual reality entertainment systems spectacularly showed the power and wonder of fully immersive VR while, yards further on along the aisle, the Robotics and Teleoperation Department of CEA demonstrated TAO 2000 - yes, again that prophetic millennial tag - a computer-assisted teleoperation system, with interactive environment modeller, its applications enabling the robotic telepresence of the operator in nuclear and other hazardous environments, as well as in medical applications to help the disabled. The language engineering component of the exhibition was also well represented: IBM's speech-to-text Personal Dictation system and Lernout & Hauspie Speech Products' speech recognition and multilingual text-to-speech systems were clearly heralding a generation of more robust speech interfaces.

The conferences themselves opened with a joint plenary of inaugural addresses. Dominic Bouwhuis (IPO, Netherlands) gave an amply illustrated talk on the role of visual and aural perception (and illusion) in the human-computer interface.

Mark Maybury (Mitre Corporation, United States) followed that with his on intelligent multimedia interfaces, the accent on intelligent interfaces to the Internet. Taken together with Harold Thimbleby's (Middlesex University) mesmerising first-day tutorial on "Autostereograms: how they work and how to program them", conference participants were regaled with an insight into a world as beguiling and inviting as Alice's the far side of the looking glass.

With characteristic Gallic flair for le spectacle, the opening addresses were followed by a brief piece de theatre of interactive dialogue between budding thespians in the auditorium and a real-time animated cartoon Imhotep displayed on a large screen enunciating and exploring in lay terms the main themes of the parallel conferences. Socialist member of parliament and mayor of Montpellier, Georges Freche, joined the spectacle and, after some exchanges with Imhotep, concluded by drawing attention to the fin de siecle renaissance of Montpellier around its flourishing technopole (science and business parks) and IT industries. Elected mayor 16 years ago, and re-elected in each municipal election since, Freche has more than anyone been responsible for driving Montpellier's regeneration as a city for the 21st century.

"McLuhan would have loved Montpellier", claims a publicity brochure for the city; "Whilst many are still blinking at the dazzling new media explosion, Montpellier had the perspicacity to take immediate advantage of its effects: optical fibres are already coursing through its arteries". And, indeed, in 1983 Montpellier was the first city in the world to be entirely cabled in fibre-optics.

As the capital of the Languedoc-Roussillon region located on the western wing of the "Silicon Riviera", a ribbon of technopoles stretching from Montpellier to Sophia Antipolis in Nice, the self-styled "Southern Capital of Europe" is one of the fastest growing cities in France.

It owes its rapid expansion from a sleepy provincial backwater to "EuroCity" (Montpellier is a founding member of the Eurocities Club) to its massive investment over two decades in the information technology, AI, robotics, microelectronics, and telecommunications industries. From its early beginnings with the decision by IBM in 1965 to set up a manufacturing plant here, Montpellier's technopole is now host to more than 1,000 companies and home to, among others, the Grammont international film and television centre, France Telecom's distance learning National Educational Service, and the Gutenberg mediateque with its vast library of audiovisual and electronic sounds and images.

An apt venue, then, for the HCI conference, created in 1991 by the District of Montpellier and EC2 & Co, and now in its fourth year. Under the ensign of "L'Interface des mondes reels et virtuels", the conference is dedicated to highlighting concrete applications of the widest possible range of HCI tools and techniques in industry and the service sector. Its bias towards already operational applications - and in particular to the interface to virtual worlds - was well in evidence this year, as much in the technical sessions as in the exhibition.

The HCI tutorials, running over the first two days of Informatique '95, broadly covered the main topics in human-machine interaction and interface design, from the design and implementation of interactive systems, through the use of powerful object-oriented techniques in the construction of highly interactive graphical user interfaces; from three-dimensional sound for virtual reality and multimedia to multimedia and multimodal interfaces to telematic services on the Information Superhighway; from an introduction to the issues and commercial opportunities in immersive virtual reality to the use of VR in the design of powerful interfaces to integrated sensor-based human-machine robotic systems permitting telepresence and teleoperation in hazardous environments.

Mohan Trivedi, the founder and director of the Computer Vision and Robotics Research Laboratory at the University of Tennessee, also delivered the invited lecture for the conferences, drawing together, under the title "New Generation of virtual and real intelligent robots", the threads of the parallel events in a paper that reported the research in his own laboratory. A fascinating lecture but not an engaging lecturer: although simultaneous translation was provided for this talk, as for most other sessions, the interpreters soon gave up in the face of a strongly accented English rattled out at breakneck speed. A pity, since the interdisciplinary work being undertaken in the CVRR Laboratory well-mirrored the interdisciplinary flavour of the conference.

From Trivedi and from others we were much relieved to hear that AI is not dead, whatever recent invidious rumours might suggest. The IA '95 tutorials, covering topics from constraint programming to development methodology and software for knowledge-based applications to management, bore witness to the continuing vitality of AI and intelligent knowledge-based systems.

Mark Maybury's tutorial on techniques for building intelligent multimedia and multimodal interfaces evinced the formidable value and relevance of AI to leading edge technologies - for example, the potential to generate on the fly, in response to user profile and preferences, any of an infinite number of unique screens and novel paths through multimodal materials.

As with the other conferences, the industrial and commercial bias of IA '95 was clearly evidenced by the nomenclature of the technical sessions. Grouped under major themes such as "maintenance, quality and dependability", "process control", and "banking, finance and business management", sessions with headers such as "risk assessment" brought together papers whose titles - "recent advances in FINSIM: a knowledge-based system to assist loan assessment", "SPARC - an expert system for loan decision analysis", "Innovative risk assessment with an integrated approach of knowledge-based and machine learning components" - testified to a healthy expert-systems industry. Other sessions brought us into the current generation of intelligent systems and AI technologies, such as neural networks.

With technical sessions introduced by no less than 47 papers the language engineering conference provided the context for drawing together state-of-the- art research with a clear orientation towards real-world applications. The shift within the discipline from a linguistic to an engineering bias was mirrored in the papers: a dearth of theoretical natural language processing in favour of applications for "theatre information and booking services", for abridgement of press releases and news reports, for querying "databases in the legal sector", or for "construction industry specification analysis and understanding".

Bringing together contributions from Alain Brunet (La 5ieme TV, France), Joost Breuker (Amsterdam), Michael Baker (CNRS, Lyon), and Jean Sallantin and Marc Nanard (LIRMM, France), and myself, a highlight for me (for more than the obvious selfish reason) was the "teleformations, IHM et Systemes Intelligents" session on the Thursday afternoon, a workshop exploring the huge potential for new directions in telematic distance learning.

Marc Nanard's reflections on his use of videoconferencing in a learning environment, followed by Michael Baker's on support for collaborative learning, and my own - in this and the following panel session - on the Circle distance learning project, clearly showed that, technologies apart, education in the 21st century will entail enormous social upheavals and a massive restructuring of the education system.

The conferences were drawn to a close on the Friday afternoon with a rather disappointing joint panel session on "Information Superhighways: the role of human-machine interfaces and artificial intelligence techniques". It was an unexciting, slightly timid, and perceptably downbeat conclusion to the week, with little said by any of the speakers that we did not know already.

Saturday morning, congress finished and exhibition dismantled, I sat in the Place de la Comedie with my croissant and coffee, taking stock of the week. I wondered how successful the bringing together of the four conferences had been. Subject specialists go off to their separate conferences, losing sight of what each other are doing and of the grander enterprises in which they may engage. The truth of the matter was probably that the computational linguists went to the language engineering sessions, the artificial intelligentsia to the AI sessions, and so on.

Perhaps not the milestone conference it could have been, but at least a plausible herald for greater things to come.

Chris Hutchison is senior lecturer in the school of information systems, Kingston University.