Cybrarian Technical Specification Ranking by General Characteristics

for Cybrarian DfES

Date: 02/08/2002

0 Introduction

0.1 This document is based upon feedback from Cybrarians’ for Focus Groups, namely:

Elderly/Disabled
Ethnic Minorities
Socially Excluded
Untapped Mainstream

0.2 Obviously, some requirements are more important to one group than another, so this ranking attempts to strike some balances but some minority requirements are vital for the minority in question i.e. voice out for blind people and language translation for those who do not understand written or spoken English.

0.3 The fact that the characteristics are generalised does not relegate the importance of the wealth of technical detail in the supporting documentation.

0.4 The technical requirements have some legal implications but these are not dealt with in the commentary. They are listed here for the Departments reference.

Data Protection Act
Disability Discrimination Act
Copyright, Designs and Patents Act
Broadcasting Acts (NB Forthcoming Legislation on communications)
A variety of Acts governing education and learning

The following paragraphs are commentaries which match the numbers in the table:-

1. Cross Platform

The core Cybrarian target groups are all much more likely to possess a television and/or telephone than a PC/Apple with a bundle of software. It is recognised that the project may have to begin its pilot phase on PCs but it should be designed from the outset so that digital television and 2.5G and 3G mobile phones can be used at a later date. This was one of the recommendations which received a unanimous high priority from all four groups. Perhaps the way to look at this is to design the project in three phases:

Cybrarian PC Companion
Cybrarian Home Channel (for DTV)
Cybrarian Traveller (2.5G and 3G mobiles)

This means that when the data is being re-purposed for the project it should be designed so that it can move across platforms.

This involves the use of tools to make automatic adjustments to pages on the basis of the user interface. A key principle here is the ability to adjust pages scaleably until such point as a tool intervenes to cut out a graphic because the screen on the user interface is too small. There is a vital trade-off between cross platform capability and customisation either by input or through machine auto adjustment. For some time to come television and telephones will only be customisable through user configuration and only through options and menu systems for output characteristics. They may be able to take some account of 'favourites' and 'reading age' but we will need to think about these issues carefully.

2. Input/Output Characteristics Customisable

Simple customisation, such as print size and font, colours of foreground and background is easy in data and achievable with slightly more difficulty in metadata. Brightness and sound volume control are also simple. These few features, though simple, are easy to overlook. For a full set see (Petrie H & Carey K: Harnessing Information Technology: Requirements for usability of Information and Communications Technologies for people with disabilities in Disability Discrimination Act : Access for All, British Computer Society, 2000). Account also needs to be taken of the user's interface so that she can choose to exclude large sound and/or image files. We also need to remember choice between justification and non-justification of the right hand margin; the latter is better for people with reading difficulties.

The basis for any set of requirements must be the World Wide Web Consortium Web Content Accessibility Guidelines (WCAG) Version 1.0 at a compliance level of A and AA (Priority 1 and Priority 2 levels) The project must also be aware of the evolution of these Guidelines into Version 2.0 (See www.w3.org/TR/WCAG10/)

An important component in this area is voice-in/voice-out systems. When voice-in was developed it was rather restrictive but it is becoming ‘more fuzzy’ and the output from search systems is also becoming ‘fuzzier’.

Technology is also rapidly evolving in the area of tutorial and signing avatars and so these technologies need to be taken into account as the Project develops.

3. Output Data Customisable

This is the core of the Cybrarian Project which allows users to define such characteristics as:

Areas of Interest
Reading Age
Formal and Informal Study
Entertainment/Leisure

Vital distinctions here include:

The ability to acquire literacy using adult oriented texts rather than those written for children learning to read
The use of apparently leisure/entertainment material for skills acquisition, e.g. sports statistics
The creation of a 'safe' environment that is not bland

4. Internal Language Engineering

This idea is primarily based on 'reading age', an assessment tool used by tabloid newspapers to check their material, but it can also be based on technical competence levels, i.e. somebody may be very good in one highly technical area but less good in another. This means that a user may want the full version of some documents but a summarised or simplified version of others.

Currently language engineering within rather than between languages is a rather scattered affair: we do have Thesaurus functions and language parsing engines and Departments of English have lexicographic counting and concordance systems. Xerox and Autonomy are working in areas relevant to language engineering. However, I have not seen these collated for a single enterprise to provide written text to a variety of lengths, with a choice of syntactic complexity and of lexicographic ranges. For many reasons this is vital for all areas of Government.

5. Language Engineering Translation

This is in a somewhat primitive state but it needs to be developed so that English and other languages can appear separately or simultaneously according to user preference both in symbols and sound.

6-10

These five are the same as 1-5 except that the information system itself configures itself

according to user behaviour. This is not yet simple except in primitive ranking tools. All have been scored as "Vital" because they make the whole system simple and non-invasive but all are complex in terms of their feasibility. They lead to specification 11 below.

11. Portable Customer Profiles

Whether the customer profile is achieved through configuration of 1-5 or the activation of 6-10 or a combination of the two, the user does not want to go through 1-5 and cannot benefit from 6-10 unless he/she can carry their profile with them or invoke it easily. The options are:

Store on Floppy. Easiest for user but almost certainly unacceptable because of problems with virus
Smart Card. Now technically possible and secure with no virus implications
Stored on Remote Website. Feasible but users might need help downloading their profile.
Cookies. This is a controversial area and although they have their uses Focus Groups were on balance against these.

On balance, the smart card solution is the best.

12. Multimodal

This means that any strand of data or piece of metadata should be processable and output in a variety of media that are at once complementary but discrete. The clearest example of this is the use of voice-in/voice-out simultaneously with keying and displaying text. The achievement of this is simple in some cases, more difficult in others but it is developing rapidly, for example, in the area of signing avatars.

This is an area where Focus Groups contributed a wealth of technical information which needs to be taken into account in a more detailed set of user requirements.

13. Variable Navigation

In many ways this is a derivative of previous characteristics but deserves a separate requirement and note. If a system is to be used effectively, people need to be able to find their way round autonomously, even if help is available. This is an area where multi-modality is particularly important; people should be able to use:

Keyboard only
Voice In only
Numeric Keypad only

ways of working with data and they need different methods for different kinds of information. The following are examples of kinds of navigation and where they might be appropriate:

Key Word Searching (by keyboard or voice-in) for data which is listed alphabetically or chronologically where the user knows what he/she wants
Image Mapping: where a user knows where a phenomenon exists, e.g. a country or part of the body or where the user can use the map assisted by labels
Menu trees: where data naturally bifurcates; this is the most traditional array for PCs but the least appropriate as it wastes the potential of hypertext navigation
Power of Nine: Particularly important where the user only has a numeric keypad from a phone or e remote controller; good for navigating by simple number in cars etc but limits the user to finding fairly large pieces of data, for full text see Power of Nine (PO9) (Carey K, Stringer R: The Power of Nine : A Preliminary Investigation to Navigation Strategies for the New Library with Special Reference to Disabled People: LIC Report 74: Library and Information Commission, 2000). One area of developing importance is the attachment of SMS functionality to all numeric keypads so that PO9 functionality can be augmented with keyword searching.
Point Centred Re-orientation: Where the characteristics or facets which relate to a piece of data can be re-oriented to show its variety of significances, e.g. a fact as an historical, geographical, anthropological signifier.
Flow Chart: Showing information in sequences where it might be chronological, in a tree or a hybrid
Colour Coded: For simple multi-page parallel sequences. NB Problems of various colour contrasts and particularly colour blindness.

How many of these are provided as alternatives depend crucially on the nature of the data and on the brief for the designer.

14. Interface Peripherals/Accessibility

The Project will have some responsibility to see that disabled people in particular have access to user interface peripherals like:

Braille Displays
Screen Readers
Magnification Packages

The degree to which these are required depends upon the ability of the specification to absorb such requirements as magnification (under customisation) and speech out (under its own heading). The general problem here is that most peripherals supplied by the Public Sector come from the Access to Work package of the DWP. The DOH supplies some equipment but not much; and so we have the problem that many potential users will not be eligible for public sector access devices and may have to depend upon the voluntary sector. Nonetheless, the system must be built in such a way that it can be fully accessed by user interface special peripheral access devices.

15. Training & Technical Support Strategies

There is so much ICT training at such a variety of levels that it does not seem effective to construct a new scheme for Cybrarian but it will need a strategy so that it can use other resources. Care must be taken here not to entangle Cybrarian with formal learning systems, at least for that sector of our target group which has been alienated by formal learning.

Technical Support also presents a serious problem but, again, we must look at existing resources. Much of this relates to the discussion of the extent to which the Project depends on legacy equipment.

16. Interactive

Much of what people generally think of as interactive, namely choice of content and rendering, is dealt with elsewhere. In this context interactive literally means the ability of the user to add to the content of the system. If this is totally unmediated it will not work to fulfil the purposes of the Project and/or will replicate other projects. Here are some examples of interactivity which are discussed briefly:

E-mail: This is one of the great successes of the Internet and Cybrarian should provide it free for a limited period for clients so that it is seen as an introduction to commercial services and not in competition with them.
Chat Rooms. Here we must exercise caution because we could easily get into hot water with such a heterogeneous client group. This requires much more detailed discussion. If it is to work at all it will be heavily dependent on human intervention, notably through Cybrarian supporters providing supervision over communities of interest.
Communities of Interest Bulletin Boards. These would require human mediation but I have no doubt that it would be easy to find competent and enthusiastic volunteers in everything from wine tasting to train spotting, local history and fishing; these are easier to deliver than chat rooms because the editing does not have to be in real time. One of the key questions for the Project is the degree of human mediation it can and should afford
Help Desk: As with the above
Hybrid Assistance. Based on the idea of human intervention triggered by a machine which 'notices' a repeated client error; as with the above
User Feedback. Both simple and sensible with an editor
User Contributions/Annotations. Slightly more complex than straightforward comments through an Editor but possible and desirable for showing results to Project backers; also see 17 below.

17. Tools/Prompts

While it is vital that the learning of users is osmotic and not heavy or didactic - many will have been alienated by traditional learning and training regimes - it is important that the user has friendly tools to help her and prompts which can take learning further. These are, of course, highly variable. Some of the most basic prompts are:

What do you want to do next?
Where do you want to go next?
Have you finished with this?
Do you want to look at related information? E.g. You have looked up teams in a competition, do you want to know more about the City/Country they come from.

What users do not want are prompts and messages which they cannot understand, such as “error” without specifying what the error is and what the user can or should do about it.

18. Collaborative

Whether the system has a large amount of interactivity or not, it needs to be collaborative, even if this is largely asynchronous, process driven. As machines continue to eat up jobs that rely upon the division of labour and low levels of literacy and numeracy, it is increasingly important to put weight on those things humans do well but machines do badly; most of these depend on collaboration and many of the problems of our target groups arise from a lack of self esteem, self confidence and social skills. The following are major characteristics which humans should learn in respect of information:

Defining a Search
Allocating weight, value, credibility
Collation
Collage
Lateral Relationships
Organic Development
Radical Discontinuous Development

Where collaborative working is conducted through human mediation/mentoring it can be time consuming and may involve difficulties of timetabling for synchronicity.

Asynchronous working requires:

General Projects Schemers
Work Flow Charts
Table of Dependencies
Dispute Resolution Procedures

This looks like a daunting list but we already undertake most of this informally.

19. Self Testing and Assessment

These are not the words we really need but some people will want to measure their own progress and feel challenged. Here are some examples:

Self Level and Auto Level Testing. A user can guess his own competence level or let the machine do it after a while and the user can then mark progress as the result of learning, perhaps against an external set of criteria such as a Key Stage, an NVQ or some other qualification; nb this must be invoked voluntarily by the user and not 'pushed' by the system and its backers
General Knowledge Quizzes.
Interactive Quizzes with Communities of Interest

20. Automated Assessment Tools

This is a controversial area because users might think an element of Big Brother (In the Orwellian sense) is being brought into the project and this would wreck the ethos. Users, however, should be asked if they want their performance assessed and we should give them incentives to want it, whilst making it perfectly clear that this can only be done if the user activates the assessment tools. The following is an array of such tools, all of which can be automated:

Reading Age Assessment
Lexicographic Counter
Search Analyser for range, depth, 'favourites'
Quizzes
Graded Tests

21. Legacy Compatible

This was one of the most controversial areas for Stakeholders who recognised that there must be a trade-off between Cybrarians’ obvious need to be ‘cutting edge’ and the kind of equipment that people may have. We will have to draw up a minimum specification according to the information and service requirements which have been given high priority in the ranking. We can then assess how much legacy equipment and systems we can accommodate. Nowadays the range of equipment runs to approximately eight generations, so the oldest realistic set-up is a PC 386 struggling to run Windows '95 up to a Pentium Six running Microsoft 2000 or later on XP. There will be pressure from community groups to dedicate much of the Project's resources on adjusting the kit to make it fit for Cybrarian but it would be much better to replace equipment below the minimum specification. As a simple rule for now:

Microsoft Office and related applications '95
Browsers, Version 3.0
HTML 4.0
CSS 2.0
Adobe Acrobat 5.0
PC 486
Modem 56 KBS
2.5G cellular phones (assuming introduction of phone use in 2004)

22. Peripheral Hard Copy Devices

This is not a problem for PC/Apple applications (although it will have to be considered as part of the discussion about copyright) but it does highlight the need for the Government to lobby manufacturers of digital television in particular to design sets with the capacity to output hard copy.

23. E-Commerce

This raises a host of interesting issues about booking holidays, buying music and purchasing on-line services; this essentially breaks the 'walled garden' principle, so would we consider one major commercial partner that can provide all these services in exchange for major sponsorship? I doubt it.

24. Undesirable

These are not included in the table as they are not specifically technical but they should be noted:

Long Forms. People are rightly worried about the length and complexity of the registration and customisation process
Junk Mail. This should be covered by the Data Protection Act
Addresses. One group said that the requirement to state address would be offensive to homeless people a name, initial, postcode recognition system is the simplest form of identifier (see below)
User Name and Password. Most systems require a user name and password to enter but I suggest postcode, family name and initial to distinguish from other members of family
Condescension and Sexism. Worries were expressed by more than one group about the nature of the language in the system. It is going to be very difficult to strike a balance between cultural neutrality and blandness. We will have to visit this again in discussion and content.

25. Summary of Tradeoffs

This analysis has produced three major areas where requirements are incompatible and call for tradeoffs. These are:

Cross platform and Customisation
Cutting Edge and Legacy
Automated and Human Agency

We need to find a way to discuss these. I suggest that we ask the four Focus Group Chairs to join Darren, Rachel, Atul, Heather and me to discuss these early in September.

Technical Specifications Ranking - table

Ranking	Attribute	Priority	Feasibility
1	Cross Platform	Vital	Complex
2	Input/Output Characteristics Customisable	Vital	Variable
3	Output Data Customisable	Vital	Complex
4	Internal Language Engineering	Vital, particularly for people with learning and cognitive difficulties	Complex
5	Language Engineering Translation	Vital for Ethnic Minorities	Intermediate
6	Cross Platform Auto Adjustment	Vital	Complex
7	Input/Output Characteristics Auto Adjustment	Vital, particularly for disabled	Complex
8	Output Data Auto Adjustable	Vital	Complex
9	Internal Language Engineering Auto Adjustable	Vital	Complex
10	Language Engineering Translation Auto Adjustable	Vital	Intermediate
11	Portable Customer Profiles	Vital	Intermediate
12	Multimodal	Vital	Variable
13	Variable Navigation	Vital	Simple
14	Interface Peripherals/Accessibility	Vital for disabled people	Simple
15	Training and Technical Support Strategies	Vital	Technically simple but administratively complex
16	Interactive	Highly Desirable	Variable
17	Tools/Prompts	Highly Desirable	Variable
18	Collaborative	Highly Desirable	Complex
19	Self-testing and Assessment	Highly Desirable	Simple
20	Automated Assessment Tools	Desirable	Complex
21	Legacy Compatible	Desirable	Variable
22	Peripherals Hard Copy Devices	Desirable	Technically simple but politically complex
23	E-Commerce	Desirable	Technically simple but incompatible with other objectives

Return to User Requirements