Information systems deal with information-processing software systems that help users carry out tasks by providing relevant information and computational support. They generate, manipulate, analyze and store both structured and unstructured data, and their functionality automates, controls, monitors, or supports the tasks at hand. Central to the information systems area is the inter-dependence of software, humans and various artefacts. This complex mixture of technological and human resources is today crucial in many domains, and information systems integrate and apply more specialized technologies to meet individual or organizational needs.

Information systems include prescriptive and empirical methods and approaches for developing and analyzing both software systems in isolation and the wider concept of information systems. The area includes sophisticated user-computer interfaces, methods for mining, retrieving and managing information, support for collaboration and coordination among computers and people, adaptations to individual and organizational needs, mobile and distributed systems, knowledge discovery and management, and architectural and design considerations. Indirectly, their impact on social and legal structures, as well as related ethical issues, also falls within the study of information systems.

The information systems area is characterized by a wide range of research fields, and there are few of today's research topics that will survive the next decades. There are however trends and tendencies in this area that span several research fields and are likely to dominate both research and industry the next 20 years. The following trends are of particular importance:

Jon Atle Gulla, Ingunn Amdal, Monica Divitini, Helge Langseth, and Hallvard Trætteberg

Position paper at INFOSAM 2020, Trondheim

The information systems area is characterized by a wide range of research fields, and there are few of today’s research topics that will survive the next decades. There are however trends and tendencies in this area that span several research fields and are likely to dominate both research and industry the next 20 years. The following trends are of particular importance:

Interaction – body and environment
Intelligent support
Universal personalized services
Ubiquitous computing
Networked communities and social implications
Emerging development processes

Each of them is discussed below.

Interaction – Body and Environment

With multi-modal interaction we refer to multiple input and output modalities as well as multiple “terminals”. The most common output modalities today are screen (text, graphics, and video) and audio. Rapid development in display technology and speech technology will result in a more diverse set of output modalities in the future. An important aspect for future output is the presentation of information to non-experts (e.g. medical information directly to the user not through physicians). Typical input devices (controlled by user, i.e. not a sensor) today are keyboard and various pointing devices (buttons, mouse, and stylus). Speech is an obvious additional modality that is becoming more important as the technology improves. The same is also the case for eye tracking, gesture tracking, biometrics, etc.

Sensor technology today includes devices for GPS, temperature, humidity, light, etc. We will see a rapid development in especially medical and military sensors. As sensor technology improves, they will be a natural input/output modality for information systems.

The terminals as we know them will probably be on the museum in 2020. All items will have sensors, and computers will be everywhere (“disappearing computers” (ISTAG 2004)). With wearable computing the sensors may track human condition and behavior (when you enter a zone – indoors or outdoors, biosensors) and be an integral part input and output devices.

The next 20 years we expect substantial development in sensor technology, speech technology, displays, and audio/video research. Speech technology will improve (at least languages with enough funding, small languages like Norwegian will need special care) and can be taken into account (see Language Technology position paper).

Whereas we today have mostly one interface per sensor or input device, the future will bring us interfaces that combine information from several sensors and input modalities. Essential concepts here are fusion, fission and dialogue management. As interfaces get more complicated the human centered design gets more important to let the user be in control of the interaction. Ideas from human-human-interaction will find their ways into human-computer interaction, though there are situations in which humans may want computers to behave differently from humans.

Many of these input technologies and sensors have an inherent uncertainty (speech recognition errors, the interpretation of a medical sensor regarding a specific human condition, noise in transmission, etc.). This calls for context-based interaction (intelligent dialogue), where we allow delayed decisions using probabilities and models from the different modalities as well as dialogue histories and user preferences. Fusion and fission will be more complex, and for a user friendly interaction the semantics of the input/output should be coordinated with the semantics of the information providing part of the system. The system must be able to detect and repair misunderstandings.

Intelligent Support

There is a class of intelligent systems that makes use of published data to analyze issues, help users solve problems or come up with recommendations. The quality of these systems grows as more data is added, provided that the system has the technology needed to deal with the data made available to it. Except from document management systems, this class also includes many learning systems, expert systems, and even tools for finding and using web services.

Tools for analyzing structured and numerical data are available today. As more data is added to these systems, the tools come up with more precise and more reliable analyses that help the users understand the issues and make decisions. For unstructured data, like text documents or images, the situation is however somewhat different. Methods for information retrieval, information extraction, text mining, and categorization and clustering have been introduced the last few years, though there are still fundamental problems in this area. The methods make use of statistical approaches like normalized word frequencies and simple morphological information about words to analyze document content. These methods address syntactic aspects of the data and does not go into the semantic logical part. Semantic approaches are still not reliable enough, and the processing limitations prevent the methods from being employed in many information management and learning systems.

The next 20 years will see huge amounts of data being published and information systems that are far better integrated than today. With this plethora of data, the potential of these intelligent systems will rise dramatically (PC Magazine 2003). At the same time, both processing capacity and semantic analysis methods will improve to the extent that the systems can start dealing with the semantics of this data rather than various statistical and morphological aspects. Methods and approaches from linguistics, statistics, and logic will be tightly integrated and used on-the-fly to help users find and interpret the information they need.

As envisioned by the Semantic Web initiative (W3C 2004b), the systems of the future will distinguish between the documents or services themselves and knowledge about these representations. Since all documents, services, rules or records are given formal logical definitions, the systems can answer questions directly rather than just pointing the user to sites that may have the information they need. The systems can combine data from different sources by reasoning about the different semantic representations. They can also cooperate on an ad-hoc basis and make deductions about the result from analyzing the semantic description of each system. Every time a new document, service or rule is added, the system has access to another piece of knowledge that will improve the system’s ability to provide the user with intelligent support.

Structured domain knowledge in the form of ontologies and models will be important to interpret the data in the right context. There will be large domain ontologies available, but they will also build ontologies for individual or more specialized use. The users may maintain their own ontologies that reflect their experiences, interests, and preferences. The same data may mean different things to different users in different contexts, and the use of these ontological representations make sure that the available and sometimes fragmented and ambiguous data is meaningful and relevant to the users. Ontologies will also allow the user to specify only what he wants to achieve and let the system be responsible for figuring out how to do it by retrieving and combining the necessary data and services.

Universal Personalized Services

Through continuous interaction with computerized systems, human users will bit by bit release information about their interests and preferences. They leave traces that can be collected and used to assess their past behavior. By collecting and analyzing these bits of information, automated systems (intelligent artificial agents) will help to find relevant information, carry out the appropriate transactions, and provide a computer-supported environment that is tailored to the individual user. This development brings (at least) two technologies forwards:

· Personalization

· Decision support systems

Both these technologies are already finding their commercial niches today. First-generation personalization systems built on techniques like collaborative filtering, clustering, and classification algorithms can be found on the internet. Limited decision support systems (a.k.a. expert systems) have also emerged, and are today capable of automating relatively simple decision making in limited domains. Examples include screening of medical images, classification of spam-mail, troubleshooting of electronic systems, etc.

Emerging technologies related to data warehousing and data mining make distributed learning and decision support available. In the near future these technologies will be more integrated in everyday life. Information overflow through digital television and an increasingly voluminous internet makes filtering and personalization of the available information important. Smaller and cheaper computer systems make decision support systems available everywhere. Venetian blinds with built-in intelligence and t-shirts with an opinion on which other clothing it prefers to be washed together with are but two examples.

Ubiquitous Computing

When Apple Research did a study on the future of computing, they tried to analyze how people in the future will carry out their tasks, which tools they will use, and how they will use them. They came up with a new concept, ubiquitous computing, that describes the way users, their tools and network resources will be working together:

"The Concept of computers as things that you walk up to, sit in front of and turn on will go away. In fact, our goal is to make the computer disappear. We are moving towards a model we think of as a 'personal information cloud'. That cloud has already begun to coalesce in the form of the Internet. The Internet is the big event of the decade [...]. We'll spend the next 10 years making the Net work as it should, making it ubiquitous."

- Frank Casanova, director of Apple Computer Inc.'s Advanced Prototyping Lab

The underlying goal is to move computers away from the user’s attention, where they are applied subconsciously and embedded in many small and highly specialized devices within our continuously changing environment. This vision of ubiquitous computing has been around for a while, but it is only now starting to take shape. Ubiquitous technology is a new computational paradigm that is important to embrace for assuring the widespread adoption of IS in the society. We envision that in the next decade ubiquitous applications will get out of research laboratories and into everyday life of people. This is made possible by a growing interest of the industry in these applications and in the advances at the level of connectivity. However we expect this adoption to be problematic for some years to come because of limitations in the technology, hardware and running costs, but also and foremost for the worries of the general public around these technologies. This will force the developers of these applications to take seriously into consideration the social and ethical implications that the developed technology brings along.

Ubiquitous computing has implications for research fields like

software architectures,
computer-supported cooperation,
interaction design, and
multimodality

Networked Communities and Social Implications

The European Union is actively working with its research funding programs to bring, within 2010, “IST [Information Society Technologies] applications and services to everyone, every home, every school and to all businesses.” Even if this objective can be reached until 2010, we expect that the consequences at the societal level will take much longer to be fully visible. In the nineties the Web has allowed the emerging of web based communities. At the beginning of the millennium we are witnessing the emergence of new forms of social aggregation, called by someone “social mobs”, supported by widespread usage of mobile technologies. We can expect that pervasive usage of IS better integrated into everyday life of people will lead to different social communities, probably more ad-hoc and volatile. These new forms of aggregation, bypassing traditional forms of governance and control, are promoting new and richer forms of participation and, potentially a more democratic participation of individuals to community life. Everybody is online most of the time (we are only offline when we choose to be), both in private and professional life, and we are part of numerous networked communities within which we communicate and coordinate our activities.

If on one side the widespread of IS can lead to a more democratic society, on the other it also brings the risk to more control on citizens and lack of privacy. We expect that the next fifteen years will be largely characterized by the struggle between these opposites and the consequent definition and continuous revision of different policies:

Communities will support local democracy through the spread of relevant information and gathering of opinions. Networked interest groups will be formed to fight for particular issues (which both may strengthen and weaken democracies). Different groups of the population will be strengthened and weakened by the move to online democracy, e.g. education, economy and accessibility will affect how online you are.
Our constant use of networked communication devices will leave electronic traces all over the place. A complete trace will reveal a lot about our private life, and care must be taken to protect our privacy and avoid intrusion, like spam and little brother surveillance. End-users must be able to understand and control how information about themselves spreads in the network. At the same time, both the citizens and institutions must be protected against network-based criminal activities, like identity theft, electronic trespassing and illegal spreading of protected information, through both technical and legal mechanisms (like coordination of national and international standards and laws).

Also from a legal point of view different challenges are arising. It is interesting for example to remind here the discussion about the distribution of legal responsibility connected to the exchange of copyrighted music shared over the net thanks to peer-to-peer infrastructure, or about the collection of an unprecedented quantity of individual data for different reasons, ranging from medical record to terrorism prevention. The sensitive nature of this information arises a set of issues related to security and ethical usage and sharing. We expect these issues to be only the top of the iceberg, with the full scale of the problem visible in the ten or fifteen years.

The adoption of Information Systems by new users will bring along a new set of requirements:

“Design for all” has been a recurrent slogan for the last decade. In the next decade this has to become common practice.
More flexible systems that are able to evolve with their users, but also be flexible enough to adapt to usages different than expected
Clearer distinction between inscripted behavior and policies

From a process point of view this requires more interdisciplinary work and more awareness in the design team of the social, legal, and ethical implications of their systems. We also need to pay more attention to emerging patterns of usage. For example, SMS and chat started as tools for teenagers and are now getting into the workplace. We believe that the introduction of the technology in different environments will bring new patterns of usage that can possibly be adopted in other contexts, with a positive contamination between different sectors. Adoption must be based on acceptance and consensus, which will dramatically change the marketing and deployment strategy of IS.

Emerging Development Processes

Most large IS development projects take more time and cost more than expected, and often still do not satisfy their users. This highlights two issues that developers are struggling with: 1) Understanding their

users, their environment and what they really need, and 2) controlling the process of gaining this understanding and developing the relevant technology. Apart from reducing development time and costs, the focus the next 20 years will be on making development projects more predictable, flexible and less vulnerable.

The major trend will be to reduce the size of each project, evolve rather than revolve by support continuing change rather than major overhauls. The relationship between the customer and developer will change into closer cooperation, with more shared learning and reduced risk. The customer will not want to rely on only one developer organisation, so the process must be open, have smaller steps and be easier to change.

Systems will be based on open standards, not just for infrastructure and middleware, but also for vertical components established by specific segments. Component-based development will have matured to the point that systems development implies less programming and more configuration of standard and generic components. In some sense, every system will already exist, it will just be a matter of describing the needs and perform the necessary customization. The customers can take care of most of their own needs through visual composition of functionality and scripting.

To be able to continuously evolve and migrate systems, systems will be self-contained and self-describing (no difference between system and documentation) using executable models of function and design. Since systems will be transparent, developers know what to modify to achieve a certain effect, they can verify the consequences of the changes (ripple-effect). New methods for model integration and migration will support gradually changing running systems.

As many important "mainstream" technical problems have been solved, there will be less emphasis on making new technologies feasible and more on understanding, describing and fulfilling end-users' needs. The developers will need to switch roles between viewing technology as a system and seeing it through the eyes of its users. This kind of "stereo sight" will become even more essential in the years to come as no single perspective on technology, society and design alone is sufficient.

Challenges in Education and Research

The anticipated development of this field requires strong multi-disciplinary approaches combined with deep specialist knowledge. The educational paradigm should be related to how students can learn to work with people of different backgrounds while contributing with their particular competences. This is more important than teaching each student fractions of several disciplines.

The need for switching roles between viewing technology as a system and seeing it through the eyes of its users, requires a new generation of computer science professionals and researchers with a training in technology supplemented with courses from the social sciences, design, marketing and psychology. The focus of research at the same time must move from studying technology de-contextualized in the lab to also studying it "in the wild".

Strong analytical and theoretical skills are needed to push the technology forward and take full advantage of the technological basis is established. This also means that education has to be regarded as a continuous activity that has to be encouraged by employers and society at large.

A cooperative and open approach to the development and use of information systems poses new challenges to education and research. On the one hand, we need to adhere more to standards and be more efficient in applying reusable or generic software rather than developing it ourselves. On the other hand, projects will be more loosely organized, and we need to learn how to organize our work at an individual basis and get accustomed to more ad-hoc ways of cooperating with other people.

Finally, we need to encourage creativity. Whereas we in the past have improved the technology by imitating and automating manual approaches, we will soon be in a position to do things that we could never imagine before. We need to stimulate innovation and creativity both in education and research.

References:

Horn, P. (2000). The Future of Information Technology, IBM Research. [http://www.cs.colorado.edu/events/lectures/horn/horn.pdf]

ISTAG (2004) Information Society Technology Advisory Group in EU. [http://www.cordis.lu/ist/istag.htm]

ISWC (2003) IEEE International Symposium on Wearable Computers. [http://www.cc.gatech.edu/ccg/iswc03/]

ELSNET (2004) The Human Language Technologies roadmap. [http://elsnet.dfki.de/].

PC Magazine (2003). 20 Hot Technologies to Watch, July 1 2003. [http://www.pcmag.com/print_article/0,3048,a=43573,00.asp].

Rheingold, H. (2002) Smart Mobs: The Next Social Revolution; Perseus Books.

W3C (2004a) Multimodal Interaction Activity. [http://www.w3.org/2002/mmi/].

W3C (2004b). Semantic Web. [http://www.w3.org/2001/sw/].

Weiser, M. (1991) The Computer for the 21st Century, Scientific American:September 1991, pp 933-940. [http://www.ubiq.com/hypertext/weiser/SciAmDraft3.html]

Members of the working group: