Niels Ole Bernsen

ELSNET, as we all know, is a Network of Excellence, but not the only one: ELSNET has a number of sisters, 21 to be precise. And one of those is i3net, the European Network for Intelligent Information Interfaces.

i3net is a younger sibling - it was created in 1997 - but a robust and fast-growing one: it currently carries out 75 Meuros (85 M $ US) worth of multi-disciplinary research, involving more than 300 researchers from approx. 100 organisations. Its remit is "to explore visionary, human-centred interactive systems for people in their everyday lives."

i3net's structure differs from that of ELSNET in that it consists of a number of research programmes, which each comprise a number of collaborative research projects (currently twenty-five in total). The applications developed by the projects, which range from a virtual planetarium and a conversing parrot to a box that stores memories, may sound futuristic, and a lot of them are: trying to anticipate what life will be like at the beginning of the 21st century, and helping shape it as well, is an important part of the i3 mission.

Another feature of much of the work developed in the context of i3 is that it seems to get even the most rabid technophobe excited.

But does i3net have any relevance for the ELSNET community? Are there significant connections between the two networks in terms of research, and do members of both communities have anything to offer to each other? Niels Ole Bernsen, coordinator of i3net, thinks so. In the following interview he talks about what is distinctive about i3net in terms of its vision, organisation and achievements, and about the role speech and language play in the i3 projects and in i3net as a whole.

Bernsen: At the moment i3 consists of two research programmes: CI (Connected Community and Inhabited Information Spaces) and ESE (Experimental School Environments). Speech and language play a role in both programmes, obviously to differing degrees depending on the focus of the individual projects.

The CI programme, which started in 1997, has two different parts: one explores computing for virtual communities, the other computing for local or physical communities. Perhaps surprisingly, the virtual community part of CI is the more "old fashioned" of the two. The interaction paradigm underlying CI could be described as one of human-human-system interaction (HHSI), where two or more people communicate with each other supported by a system. I believe that this triplet paradigm is the paradigm of the future, and that it will gradually replace, or subsume, the one user - one system paradigm that has been dominant so far.

The basic scenario of the HHSI paradigm generates lots of research directions, and many of these will involve language and speech. When you have humans communicating with one another and with, or through, a system, it becomes important for the system to be as adept as possible at human communication. This covers a wide spectrum of activities. The systems involved in the CI projects need to be capable of such things as multilingual generation, meeting indexation and filtering, summarization, translation, speech recognition and understanding, and so on... And for these things you obviously need to draw on speech and language technologies.

The second research programme, ESE, which started more recently, focuses on experimental school environments for four-to-eight-year olds. Some of the dimensions already implicit in the CI research programme are more prominent in the context of ESE: being creative, for instance, and learning in the process of being creative. Creating, manipulating, understanding and reflecting on different kinds of contents (be it theatre, animation or everyday events) all contribute to learning, and are all central to the ESE programme.

Children between four and eight are, of course, in the transition to being literate, and this, too, is an important focus of ESE, both as a research objective and an object of the research. How can we develop computer systems and new interfaces that enable children to learn creatively, even if they can't read or write yet? And how can we gently support the transition to literacy? These are real challenges, and they have a lot to do with speech and language.

While the focus of the ESE programme is somewhat different from that of the CI programme, there are many important similarities. The emphasis on communities, for instance, is there in both programmes - in CI these might be neighbourhoods or geographically dispersed communities, in ESE schools. Moreover,

Visit a Legible city on a Virtual Bike: one of the applications developed within i3, by the eSCAPE project.

Creative learning in the four-to-eight age range is central to the i3 research programme Experimental School Enviroments

A glimpse of the computer-intergrated classroom designed by the ESE project NIMIS. Communities of "ordinary" users, such as schools, play an important and active role in i3net

Thi CI project Magic Lounge is developing intelligent communication sevices for virtual meeting places

Building such a network is an important objective in itself, but it will also serve as data for the research process. And that is something else that is typical, and distinctive, of the i3 research projects: they are all data- and methodology- intensive. All the projects are contributing to the development of new systems, or systems in new areas, or for new user groups; and reflection on that process is an integral part of the i3 approach. So collecting video data on the process of development, for instance, becomes part of the research. But there is a real need for more and better tools in this respect, and for appropriate markup languages.

Another thing that is distinctive about i3 is that users play a very strong role in the i3net vision. Working with users, and establishing what their needs are is at the heart of i3, as is trying to glean how ordinary people will live their daily lives ten years from now. And this emphasis on ordinary people and their lives is iterated in the project methodologies, where working with users from early on is a requirement specification.

In addition to technology and users, the integration of design is another important characteristic of i3. A restricted perspective on design would primarily view it as concerned with polishing the surface: changing a heavy portable into a smart and attractive product, for example. That is one aspect. But designers -and other creative people, such as theatre directors and artists -are also good at anticipating

A final distinctive characteristic of i3 is the community dimension of the network as a whole. Obviously all Networks of Excellence exist to bring people together. But in i3net this aspect is particularly prominent, because the network supports a number of specific research projects which have a lot in common; and that makes it possible to have more focused targets for community building. The i3 community also meets twice a year and that, too, contributes to a strong sense of community within the network.

But to come back to the relevance of i3net for ELSNET: both current i3 research programmes (CI and ESE) exemplify the need for contact and close collaboration with people in the speech and language area in the near future. The same would definitely apply to any possible third research programme the i3 community might embark on. And for i3 as a whole, speech and language become extremely important as components for supporting technologies.

The i3 projects are typical of next-generation projects, where we're not working so much on basic technologies any more (as has been the case over the past 20-30 years), but are concentrating instead on new ways of putting technologies together, on the creative development of new integrated systems. The needs of such systems will drive speech and language work in the near future: we are increasingly seeing speech and language products being integrated as components in industrial and research systems, or contributing to enhancing components of larger systems. Such a development will give a very different role to speech and language researchers in the future; and because of it, I can see lots of possibilities for collaboration with ELSNET, which is itself opening up, in its next phase, towards speech and language in the context of natural interactivity.