21
According to Craig Knoblock, of the Information Sciences
Institute, the most natural way to model the semantic content of
information sources is in reference to ontologies-knowledge
representations that can be constructed for a given subject area
(e.g., stock market data); machine learning technology is needed to
automate model generation because the body of information is so
large.22 Because the metadata also
will become semantically drifted, Knoblock argued against central
standardization: "There is going to have to be some kind of
distributed solution, where if you have these information providers
that are actually buying this information, they are going to have
to change their model and update things. There has to be enough
information in the underlying structure that it is easy to make
those changes. But there is no way that you can anticipate all
those changes."
Moshe Zloof, also at the workshop, cautioned against reinventing
lessons from decades of experience with database management. New
approaches such as using agents to model the semantics of
unstructured data now flooding the Web may be less effective than
structuring the data to begin with (e.g., by using a relational
database model). As Austin Henderson observed, however, fixed
structures-whether embodied in a database or modeled from diverse
sources on the Web-inevitably become
OCR for page 64
Page 64
out of date: "Suddenly my database shifts, not because anything
in it shifts, but because the world shifted. This is the well-known
problem of semantic drift. ... [We are] never going to get everyone
to agree [on a structure. We are] always going to be in the
position of negotiating."
Pleasant to Use
An important factor in attracting new users and helping them
overcome their fear of technology is creating interfaces that are
naturally attractive and fun to use. A key historic differentiator
between home and work, underscored by social scientists at the
workshop, is the element of "desirability": home uses of technology
have tended to be discretionary, and interfaces or other aspects of
technology that do not appeal to consumers are often not used in
the home. The HomeNet study, for example, focuses much more on what
people want to do than what they need to do, as would be the case
at work. At the workshop, Robert Kraut, of Carnegie Mellon
University, explained that there is no direct connection among
utility, usability, and desirability and that much is not
understood about how those qualities do or can relate to each
other. Making systems less threatening, less technical looking,
more familiar, and more interesting and fun will be important
components in creating interfaces that will actually be
approachable and used by many individuals. As mentioned above,
these systems, however, must gracefully lead to more efficient
interface strategies whenever the interesting/fun interfaces are,
themselves, not efficient for long-term or general use.
The pleasure, fun, or desirability of use is part of a broader
pattern of interaction with behavior that should be considered in
designing interfaces. For example, the ability, in a communications
context, to see people on screen, especially in real time, can
affect how involved an individual is but also tends to result in
payment of more attention to physical appearance, associated
symbols and cues that can be removed with other communication
modes, and increase in cognitive load, which affects attributions
to others and persuasiveness. Regardless of context, as Sara
Kiesler observed at the workshop, every change in an interface
implies changes in social psychology, organizational processes, and
other side effects for organizations and individuals-effects that
can be studied and anticipated.
Telepresence, in the form of casual video conferencing and
collaboration, is the subject of much speculation about how
technology can eliminate barriers to intimacy. Although many
technologists and business analysts tout video teleconferencing as
a possible "killer app" for the NII, Robert Kraut and Sara Kiesler
noted that research over 25 years suggests limited payoff to
it-conversations accompanied by video are not clearer, information
exchange is not better-but some do like it better than simple
OCR for page 65
Page 65
audioconferencing. Similarly, different attitudes have been
recorded for participation in text e-mail versus systems with image
transmission. More optimistic technologists hold out the promise
that within a year or two some people will be able to glance into
the offices-home- or business-based-of perhaps 60 people with whom
they normally interact, in contrast to most video conferencing used
for remote meetings with many people in attendance. With the advent
of wideband digital networks, ease of use, and better quality, the
telecommunications center is moving into personal computers or
workstations. As the technology becomes more widespread, most
meetings could consist of two people collaborating under casual
circumstances; as experimentation on the Internet's Mbone multicast
system suggests, extremely large (e.g., in the thousands or more
participants) or variable-size meetings also will become easier and
may become more common. As discussed in Chapters 4 and 5, large
group interactions appear to be an area where more understanding of
social dynamics is needed.
Pulling It All Together: Eci
Interfaces In The Year 20xx
To describe the future is to risk being wrong, but it is a
useful technique for showing how it may be possible to integrate
the key concepts of an ECI to work seamlessly together to create a
whole new paradigm for interaction between information
infrastructure and people. A simple scenario, focusing primarily on
the input/output aspects, is provided below.
Along with demonstration or prototype projects, scenarios, per
se, were suggested by workshop participants as useful elements of
an interdisciplinary research program because of their amenability
to computer and social science explorations that begin with their
design and continue through assessment of the resulting
roles/relationships/outcomes under different rules and starting
assumptions. As illustrated at the workshop by Michael Traynor's
telemedicine scenario, research could develop and explore scenarios
that involve multiple stakeholders and diverse interests that
converge on cases of NII use. Scenarios might also provide a
training/teaching paradigm related to how new media affect extant
procedures, expectations, and so on; similarly, simulation games
aimed at policy analysis have already shown that scenarios hold
promise for providing a framework or vehicle for collaborative
policy deliberation among diverse stakeholders and for arriving at
negotiated agreements on policy inputs to the NII decision-making
process, but the methodology calls for systematic evaluation.
OCR for page 66
Page 66
In The Year 20xx-One Scenario
It is the year 20XX. Systems with ECIs abound and take a wide
variety of forms. Some appear on workstations; others are on
accessories carried around as a notepad or cellular phone was
carried in the 1990s. These accessories, however, are
multifunctional in nature and can be used to access almost any type
of information or service available. Many of the ECIs are simply
integrated into the environment as part of rooms, vehicles,
appliances, and even clothing that people wear. The generalized
information systems themselves are integrated so that people can
begin a task (e.g., sending a message) in their office on the
system built into a desk or wall there and continue the activity
seamlessly as they walk out the door to get into their vehicle and
leave on a trip.
The systems are modality independent with regard to both input
and output. In the office, they may be primarily visual display
based (especially if one works in a shared office space). However,
as the user gets up and leaves, they are able to seamlessly move
from interacting in a visual fashion to interacting in a verbal
fashion, completing the "e-mail" as they walk down the hall, get
into their vehicle, and head for the airport. While en route, the
voice interface can be used to access any of the information
transaction or communications systems, in a purely verbal fashion.
This might include checking weather "maps," buying a gift for one's
spouse, touching base with other colleagues, etc. Since the systems
can all work either visually or verbally (words), these same
systems work equally well for colleagues who have low vision or
blindness or are hard of hearing or deaf. Because the verbal
information can be rendered as Braille or speech, the systems could
also be used by individuals who are deaf or blind or who are unable
to read at all because of specific learning disabilities that
prevent them from learning to read or read well visually.
Individuals who have difficulty learning the new systems or new
functions on the systems find that there are built-in agents that
will help them through whatever task they are interested in and
that will interact with them in a friendly, natural language format
(or that can interact with the user's own personal agents). They
can either speak to the systems aloud, type on the built-in keypad,
or use any other technique or device to input information. As users
become more expert, they can begin using shorthand phrases, codes,
gestures, and other more efficient but less obvious strategies. The
user and the systems that the user interacts with develop these
strategies naturally over time. These conventions are also passed
from one device to another so that users' familiarity with them is
interchangeable as they move between physical systems. In addition
to using verbal input, many of the systems will have the ability to
monitor both the environment around the individual and the
individuals themselves for contextual information.
In addition, they can use global positioning systems to
determine physically where the individual is and environmental
databases to help understand the context or surroundings the
individual is in at any point in time. This may include a knowledge
of which other people are in the immediate vicinity (inasmuch as
they allow this information to be known). With information about
the environment, the context, and the individual, the device can
much more easily
OCR for page 67
Page 67
interpret the interactions and requests it receives from the
user and may be able to anticipate or facilitate the activities of
the user as well.
When an individual does not understand information that is being
presented or how to achieve some objective, intelligent agents in
the system are available to assist the individual in representing
the information in a simpler format or to assist with instructions
or with carrying out the activity. Some agents might be autonomous
and carry out tasks automatically for the user. Most agents,
however, are collaborative and interactive and act more like an
intelligent colleague or assistant. They are able to interact with
the individual at whatever level and in whatever modality (visual,
aural, tactile) is most appropriate and most effective given the
environment (noisy, etc.), situation (person's eyes or hands are
occupied, or a situation requires silent operation as in a
meeting), task, and user abilities or preferences.
Although many situations and environments may require the use of
only one or another of the available interface modalities (e.g.,
visual only, verbal only), there will also be times when the full
abilities of the user are available, including simultaneous use of
whatever visual, auditory, and tactile manipulative abilities the
user may have. In these cases the individual can take advantage of
this by using a full immersive environment. For example, the user
may use an immersive environment to simulate transport to another
virtual environment. Instead of traveling to meet colleagues, the
user can sit at a desk and move into a mode where he or she
visually, aurally, and manipulatively (and, eventually, tactilely
and olfactorially?) joins with other colleagues from around the
country in a virtual meeting room where they communicate and
exchange virtual documents or exhibits and carry out their meeting.
The colleagues around the table who are deaf can have the system
invoke a speech recognizer and present its output on the screen.
(In the next decade or so, the speech recognition technology will
probably still make errors. But for clear speakers and narrow
domains of discourse, recognition may be sufficient for
understandability.) The text may appear to float in space in front
of the speaker, or the user can drag the text displays for
different people closer together in the space in front so that it
is easier to monitor them simultaneously. People who do not have a
hearing impairment also find this feature useful, particularly if
they can read faster than they can listen and find it easier to
focus attention on a particular verbal stream or to check over what
was said when everybody's speech is presented visually. It also
allows them to check back over what was said. This is particularly
valuable if they are trying to listen to multiple overlapping
discourses. Colleagues who are blind or who have difficulty reading
any of the printed materials can have the materials presented to
them aurally or translated into a form that is easier to
understand. Sighted individuals also take advantage of this feature
in order to allow them to continue monitoring the situation or
demonstration with their eyes while the textual information is
being fed to them aurally. Even an individual who is deaf or blind
can have the information translated and presented on a special
dynamic Braille and tactile display that can be attached to the
system.
Immersive environments can be used for a wide range of functions
beyond allowing an individual to travel to and visit most any real
or simulated spot on earth. They also allow the individual to scale
themselves larger or smaller in
OCR for page 68
Page 68
order to explore or better learn about objects or environments
(e.g., the ability to zoom in and out to learn about geography,
biology, etc.). They are also used to allow individuals to explore
things with their senses that are not available to their senses.
This includes seeing the invisible, hearing the inaudible, and
translating concepts that have no physical form (such as
information) into visual or auditory formats in order to gain new
insights. Again, users with all of their senses intact can choose
to have the information presented in simultaneous multisensory
form. Individuals for whom some senses are weaker or absent have
the information presented in forms that best meet their individual
abilities and learning styles.
Notes
1. In addition, early experiences are often limited indicators
of new technology's benefits: the 1980s and early 1990s discussions
of the "productivity paradox" suggest that disappointing financial
returns from early computing investments reflect relatively
simpleminded automation of existing work processes rather than the
fundamental restructuring of those processes that has proved
necessary to realize the full benefits of computing. See CSTB's
Information Technology in the Service Society (CSTB,
1994b).
2. An interagency assessment of issues and opportunities for
kiosks in government applications proposed a staged pilot and
market test program that would support data gathering and
incorporation of feedback into future design and deployment steps
(see Government Information Technology Service (GITS), 1995).
3. The IDC data are at the household level and thus likely to
produce higher percentages than the individual-level data from the
Census Bureau survey. However even if both data sets were at the
individual level, it would still be impossible to draw meaningful
comparisons, because they used different sample weightings in order
to factor their results to the scale of the whole population. This
is one example of the difficulty of identifying trends and making
comparisons from survey data in this field.
4. Project 2000 researchers statistically corrected the CNIDS
data to weight the sample in proportion to the U.S. population for
gender, age, and education-variables known to affect the likelihood
of Internet use. Income was not included because of a high
nonresponse relate for income in the CNIDS survey; education,
however, is a reasonable proxy for income. The researchers also
adjusted the data to omit logically inconsistent responses that the
CNIDS had included, such as those from people who initially
reported having used the Internet but later in the survey reported
the opposite. See Hoffman et al. (1996).
5. Whether new Web-page publishing tools are readily usable by
nonspecialists remains an open question. For an anecdotal account
that illustrates the difficulties novices have with such
applications (among others), see Rigdon (1996).
6. Of course, a number of innovative applications of multimedia
technology have been introduced for education, but several
education experts believe the promise of such technology is only
beginning to be tapped.
7. A further challenge results from the level of exposure to a
given environment, situation, or task; for example, there is a
difference between a mobile phone one rarely looks at, a phone one
never looks at, and a phone one uses frequently.
8. A related concern is whether there are general skills that
people can learn for use in a variety of settings. Does learning in
a specific context ever limit the usefulness of the resulting
knowledge?
9. For example, Internet bridge clubs type in bids without idle
chit-chat; they sit
OCR for page 69
Page 69
locked in their houses, staring at text, punching at
keyboards.
10. By contrast, the encoding features of the Hypertext Markup
Language are mainly used to control the appearance of text (e.g.,
bold type, fonts, blinking text). HTML's descriptive codes are
mostly for low-level constructs such as emphasis or indented lists.
It has no standard mechanism for indicating common document parts
such as author, abstract, keywords, or references. New HTML codes
(tags) for such elements could have the side benefit of being
interpretable by software agents for automatic indexing and
searching. New tags are continually being proposed and implemented;
however, the resulting lack of standardization hinders content
producers, because a conscientious Web author must test pages to
ensure that they appear properly when viewed with a variety of Web
browsers that support different overlapping sets of tags
(Schulzrinne, 1996).
11. Note that capabilities for filtering and blocking stand out
as features specifically contemplated for children as a
subpopulation.
12. According to the Cross-Industry Working Team (1995),
nomadicity refers to the ability of people to easily access a rich
set of services, other people, and content while they are on the
move, at intermediate stops and at arbitrary destinations;
ubiquitous refers to systems that access communications and
computing services via the NII and that will be at least as common
as today's telephone. Moreover, the NII will facilitate
connectivity through a wide range of electronic devices, including
portable, mobile, and wireless computing and communications.
13. U.S. sales of interactive kiosk hardware were $449 million
in 1994 and estimated at $610 million for 1995. The retail sector
accounts for 84 percent of kiosks installed in 1995, but Venture
Development Corporation (VDC, 1996c) expects faster growth in
financial, government, and corporate use. Information-dispensing
kiosks are about 60 percent of 1995 installations; the remainder
are point-of-sale manufacturing kiosks (e.g., greeting card and
business card printing) and transactional (e.g., product ordering,
driver's license renewal). VDC expects faster growth in
transactional kiosks than in information delivery kiosks, partly
because return on investment is easier to justify for a kiosk that
sells something than for one that gives free information.
14. The most detailed recent survey of disabilities by the
Census Bureau is the 1993 Survey of Income and Program
Participation. Although the data are now several years old, it is
unlikely that the percentages of people with various disabilities
have changed significantly. See
http://www.census.gov/hhes/www/disable.html.
15. The survey involved interviews with over 26,000 adults. It
measured skills likely to be required in work, home, and community
contexts, such as locating and integrating information in a prose
passage; writing new text; interpreting lists, charts, and graphs;
and reading and using numerical information (U.S. Department of
Education, 1992).
16. In addition, as Wallace Feurzeig observes in his paper in
this volume, spoken-language interfaces allowing literacy training
systems to integrate spoken and written communications could
enhance training by enabling learners to build on their spoken
language abilities.
17. Interoperability is being advanced through standards such as
MPEG for video coding and H.323 for multimedia conferencing. There
appears to be a trend for Web-oriented multimedia products
(telephony, conferencing) to conform to these standards, which, in
the immediate future, will make it much easier to communicate
without elaborate prearrangements. Interoperability is also
advanced, as described earlier, through the use of transportable
software (in a standardized language and virtual machine), which
removes the necessity of every party to a session needing to have
all of the application software in advance, and distributed object
systems, which allow existing applications on diverse computing
platforms to interact with one another. Standards are taking on new
meaning as a means for facilitating interaction between
applications or customizing equipment with
OCR for page 70
Page 70
transportable software, rather than being rigid constraints on
the equipment. Networks, as well as applications, can be customized
with transportable software, and programmable networks are a
significant near-future research topic.
18. How and when these developments take place depends in part
on relevant public policy parameters (e.g., the evolution of
cryptography policy).
19. A document-centered approach represents a midpoint, in which
people could use various applications, but would still have to
access whole documents rather than data from within documents.
20. At the workshop, Robert Kraut, of Carnegie Mellon
University, noted that because information is not a passive,
inactive thing, it can have different values for consumers and
producers. For example, a babysitter who wants to advertise to
parents in the neighborhood probably values that information more
than the parent who feels bombarded with advertisements from many
sources.
21. See
http://www-db.stanford.edu/∼gravano/standards.
22. Knoblock refers to information in the Web, but the
observation applies more generally to all forms of information in
the NII.
Representative terms from entire chapter:
interface design
Items in cart [0]
