4.1 UK context
4.2 Wider context
The methods we used for the study provided us with some contextual insights. These form a backdrop to the study's subsequent sections.
People have written about disruptive technologies. I would describe choice as disruptive governance. It really forces you to change your view of the world. It is at this point that power really shifts. It is like that inversion of magnetic north to magnetic south that scientists talk of, where accountability for the first time really starts to flow downwards. Andrew Turnbull, Cabinet Secretary. (3)
We were struck by the way that "Personalisation of public services" has become something of a mantra amongst public-sector policy-makers, almost within the period during which we did this study. For example, in Tony Blair's 23/6/2004 St Thomas' Hospital speech on the future of public services (4); and in "Personalisation through participation: A new script for public services" (5), in which Charles Leadbeater lists 3 "meanings" of personalisation in the context of the provision of public services, all of which are echoed in past and current discourse about the personalisation of information services:
Leadbeater goes on to compare different aspects of public service provision, according to whether the service is characterised as "traditional public sector", "new public management" or "personalisation" (6). It is interesting to note how under this latter category, there is so much resonance with the themes and ideas of this study.
It is too early to say whether this focus on personalisation as a central organising theme for modern public services is just a temporary fad or a more permanent feature. Even if it is only the former we think that a wider range of people in the UK academic community than those whose main focus is on information systems and/or the JISC Information Environment will start to recognise personalisation as a serious policy issue: the view of some in the information systems world that personalisation is a passť topic cannot therefore be sustained. It is partly for this wider audience that our report is written.
There are many similarities between large further and higher education institutions (FHEIs) and other large public and private organisations, but there are also key differences. All large organisations hold personal information about their employees and their customers but a university or college will (in theory) be able directly to relate that data to the information needs of their students. A student's performance will generally be monitored and assessed far more closely and in a more detailed way than that of an employee (though one can think of exceptions - call centres for example). A first year history student will have different information requirements from a postgraduate chemist and many of those requirements can be predicted to a certain extent (reading lists being an obvious non-technical example). The very business of education is exposure to and analysis of appropriate information. FHEIs traditionally have much more scope for using personal information for internal purposes without running into privacy concerns. Compared to, for example, an online retailer, a college should have far more, and more authoritative, information about all of its students than the retailer has about all of its customers.
During this study it became clear that the JISC Information Environment Architecture (JISC IEA) underpinned many of our discussions - our remit uses the word "Presentation" in a very particular way - and we often found the need to refer to the diagrammatic representation of this architecture created by Andy Powell (7). We have reproduced the diagram with accompanying notes in appendix 11.6. We were fortunate to have presentations from Andy Powell and Paul Miller during our workshop consultation process.
The layered diagram of the JISC IEA shows a range of components in a distributed environment. It is not meant to suggest a strictly layered hierarchy (such as the 7 layer OSI Model), it is more a way of categorising components and roles.
The provision layer includes content providers, including academic-based, commercial content via agreements with JISC and other negotiated services; these may be hosted by JISC or by other providers. Institutional content and content available from the Web also appear here.
The presentation layer includes any tools which users use, whether remote (e.g. a portal), or on the desktop, such as an RSS (newsfeeds) aggregator. There is often an assumption that the presentation layer only means portals - this is wrong - it can include anything from mobile phones to Web-based services.
The Fusion layer includes brokers (such as GetRef - http://edina.ac.uk/getref/ - a bibliographic broker based at Edina), aggregators (such as a metadata harvesting service which makes the data available to a portal), catalogues (such as RDN hubs which make metadata records available) and indexes (such as the ISI Web of science, or Google).
Shared infrastructure might include authentication, a service registry to tell services about other services and a trusted way to get systems to talk to each other.
Services may span 2 or more layers of the diagram. Services may span 2 or more layers of the diagram. The JISC IE is underpinned by a set of common standards that enable interoperability between services. These services are formally part of the JISC IE for UK Further and Higher Education, however as the Common Information Environment (CIE) develops, the standards will enable interoperability across sectors. The underlying standards are documented in the JISC Information Environment Architecture Standards Framework:
There is a significant push in the UK, as elsewhere, towards creating institutional portals, a term whose definition varies widely. The essential characteristic is that the portal provides access to a variety of materials or services, originating from inside and outside the institution, through an interface provided by the institution.
An institutional portal should be seen as a framework for the delivery of services. The portal knows who you are, and what you can see is defined by that, so that you as a user see your address, your HR record, your exam timetable, and your HESA census data. Interview participant.
There is currently a division between those working to create so called "thin portals" - where relevant content and services are gathered together in one place but the portal identity may be lost once the user accesses a particular resource - and "fat portals" where everything, including external services which usually have their own interface, is transformed to be displayed within the portal's own interface and identity.
A thin portal provides doorways into other applications, such as to the home page of the learning environment, or quite deep within it, or to "today" in the institution's calendar application. Workshop participant.
The motives for establishing institutional portals are varied. Some claim that portals achieve streamlining of business processes with time saving, for students and staff "Look: there are no queues at the registry in the first week of term - how much is that worth?" Personalisation is often seen as a key element in this improvement. Clearly if the institution can provide a tailored service to students, filling in many of the blanks and saving the enormous duplication that is characteristic of multiple registrations (at registry, faculty, department, course, library, sports centre, medical centre, careers service etc) this is an indisputable efficiency gain and a clear justification for personalisation. Others feel that an institutional portal is part of the institution's branding process. In this scenario personalisation may be seen as the "Wow" factor - demonstrating the up-to-date technical capabilities of the institution, or as a means to maximise the extent to which content and services, of whatever origin, seem to the user to be provided by the institution. Others see an institutional portal as a way of controlling the content (type of content, quality of content, cost of content) put before their students and staff. In this case, personalisation tailors the level and extent of access rights accorded to users depending on their job / role, course, subjects studied etc. Finally, some see a personalised portal as means of finding and correcting "holes" in the institution's data management (8).
If users are in the "wrong place" in the MIS, then they may be denied access to a service they need. Of course the converse of this is that if you know that a user is denied a service they should have, this shines a powerful light on the MIS. Interview participant.
Whatever the motives there is a definite, but not yet decisive, move towards creating institutional portals. This move creates demands - the institutions want to be able to use and reuse resources and services which have been funded for use by the academic community (9), and they may need to integrate core (and often legacy) systems, such as MIS, library catalogues, and learning environments within their portals.
The two market leaders in the FHE learning environment market (Blackboard and WebCT), as well as open source learning environments like Moodle or Claroline, claim the capability to interface with campus portals and with authorisation and authentication services, though it must be said that those we spoke to who are using them for this kind of portal integration found the process difficult and the capabilities somewhat limited.
At least some use of Rules Based Filtering is practically an inherent feature of all learning environments, in that the specific content and services to which a user has access are shaped by preset rules about relationships between items and user profiles. Some learning environments also offer customisation, at varying degrees of complexity. For example the current version of the Ufi/learndirect learning management system allows users to add personal URIs to their homepage, but offers no control over its look and feel. A future release of the Ufi/learndirect learning management system will allow a 3rd party user of content from the system to customise the content "in the colours of the 3rd party".
More complex customisation, in which one class of user (teacher) can customise a service so as to provide personalisation for another class of user (learner) is provided by at least one of the market leading learning environments:
Using information specific to a student, instructors can ... automatically release content to students that meet certain criteria. For example, instructors can release content to students based on participation in a learning group. Or, instructors can use a student grade on a particular assessment, assignment or combination of grades to determine what content should be available to aid that student's learning. More advanced instructors can even ... construct an online course that automatically releases content to students based on a certain learning style preference (auditory, visual, etc). Instructors have tremendous freedom to create online courses that automatically adapt to a student's needs. Questionnaire respondent.
Interestingly, the approach described above sidesteps, by using a customising (human) agent, the technical difficulty of "true" Adaptive Personalisation within a learning environment, a point made by several questionnaire respondents and workshop participants.
.... the ultimate vision of a personalised learning pathway for every learner according to their unique learning needs will require extraordinarily rich metadata if intelligent software is going to be even vaguely accurate at selecting the right content at the right time. The cost implications of this metadata are not yet well understood, and in my view, may make the personalisation argument cost prohibitive for all but the most well understood and universally studied topics (e.g. school maths may have a positive ROI for massive investment in metadata, but minor elements of Australian history may not).
Professor of Learning Technology, James Dalziel. Response to questionnaire.
Like many other systems in use within FHEIs, library management systems offer some services, such as OPACs, which can be used anonymously and others, such as reservations or inter-library loans, which rely on information about the user's identity. Historically, Library Systems maintained their own database of users but, increasingly, this is being integrated with other campus systems by the use of e.g. LDAP directories. At the moment, integration is often limited to a periodic import of user data from third party sources rather than 'real-time' integration. There are a variety of practical reasons for this. For example, libraries often serve a broad constituency (including Alumni, for example) whose details can't easily be accessed elsewhere. Also, institutional directories often only store individuals' institutional email addresses (.ac.uk), whereas for reminders, overdue notices, etc. libraries need to use the email account that users actually make use of. For many students, this is a personal Hotmail or Yahoo mail account.
There are increasing demands being placed on vendors of library automation products to incorporate personalisation into their products. One comes largely from the public library sector which is keen to introduce Amazon-like features, including cover art, reviews and recommendations, into their OPACs. Many public libraries apparently view themselves as being in direct competition with the likes of Amazon.
More directly relevant to the UK academic community is the demand to integrate access to a variety of electronic databases and eJournal subscriptions into a single interface. If a system is aware, for example, of what course a student is on, then it can present a tailored list of search targets for resource discovery and use technologies such as OpenURL resolution to present the appropriate copy of a particular resource.
A brief summary of the main products in use in the UK academic community is provided in section 220.127.116.11.
If you buy something on the Internet, from a filter plug for your ADSL connection through music and books to a car or a house, you will be presented with elements of personalisation. How people react to these elements seems to vary widely. Some object wholesale to insincere greetings messages and inappropriate use of their first name. Others claim that the systems don't match their needs:
Customers don't want relationships with corporations. As I smile at the sound of the store attendant's greeting, I'm actually responding to a person. I'm certainly not responding to a corporate message, and I certainly don't want warm personal "relationships" with the 100 retail businesses I use most.
Nothing annoys me more than if the program thinks about what I would like and produces something I don't like. Peter Gietz, DAASI.
Systems can guess but when they fail it is really a nuisance ... The best personalisation is to build simple and very powerful tools, which can be used separately and share a common logic and UI [User Interface] look and feel ... Then there can be customisation options ... and academic people like suggestions not rules ... people who read/searched/did this also read/searched/did this lists. Lassi Nirhamo.
Most of the people we spoke to during this study agreed that it is more irritating than useful to know that the last person who bought a filter plug also bought 3.5 metres of cable and clips. This problem emanates from what Jared Spool calls "indiscriminate attention". "
It's nice at times for a third party to pay attention to certain interests and make recommendations. For example, if you go to a restaurant regularly, and a waiter knows you like a certain salmon dish that is sometimes available as a special, it's nice for the waiter to point that out. But you probably don't want that same waiter to start commenting on your recent choice in friends. http://www.nmpub.com/blog/archives/000030.html
A significant number (probably not a majority) spoke in favour of system suggestions for future purchases, particularly in the area of books and music, where several had bought items on the basis of system suggestions and were happy with the results. Some were convinced that books and music were a special case and that people will only bother to contribute reviews and ratings in an area of passionate interest rather than an area of work or study.
At first glance it seems that the world of Web commerce has already produced solutions which allow personalisation and implemented them widely.
The public sector is five years behind and the gap is growing. Workshop participant.
Some of the people we spoke to thought that this meant that we were wrong to question whether and where personalisation is necessary, taking the view instead that it is already here and will soon become pervasive. Others, however take a very different view:
I think this topic is still very much in the research phase and premature for practitioners to consider. Professor of Computer Science, Michael M. Danchak.
While we need to continue to research this area to some extent, I think much of the focus on this area is misplaced given my concerns above about the problems of realising the personalised single learner pathway concept ... I think the next year or two may see a major shift away from personalisation in e-learning as the "big idea", and in favour of "collaborative learning activity sequences". Professor of Learning Technology, James Dalziel
What are we to make of this discrepancy?
We believe that personalisation has succeeded in commerce where it fulfils a core function for the business and where the task (although not necessarily the technology) is relatively simple and well defined. In eBay for example, if you are using the service as a seller to produce a regular source of income (and you are the lifeblood of the service, without you it will not work) then you obviously require to see the history of your past sales, state of bidding in current auctions, payment history etc. This is not an improvement or enhancement to the service, it is an essential ingredient. For Internet banking and financial services it is very similar - it is self evident that an accurate personalised data display is vital to the business. For a service like Amazon, however, the argument is not so clear cut. When pushed to come up with a success story for personalisation most people cite Amazon. Is Amazon not creating a new market using personalisation? Is it not using the technology to increase its sales in an innovative way which is unique to the Internet? In the face of the success of the brand and the ever increasing sales it would be contrary to deny it. Is this core to Amazon's business or a clever add on? Many of us are old enough to remember bookshops where the staff were knowledgeable enthusiasts, more than willing to spend time getting to know you, debating the merits of different titles and authors and pointing the customer to new delights. The mass marketing of books and the rise of the large chain bookstore has made this sort of shop very rare, and to some extent Amazon has come to take its place - we value a place where we can browse quietly, undisturbed, occasionally surfacing to get a small but appropriate piece of advice. Perhaps Amazon is a very old fashioned service and personalisation (along with cover pictures, reviews etc) is a vital tool used to accomplish it. For libraries to emulate the success of Amazon comparable, and possibly unaffordable, personalisation features would be needed.
For example, it is not possible to establish a personal relationship between yourself and a bookseller like Amazon.com. There are no people. There are no salesmen. By providing "intelligent" user interfaces to goods and services, some businesses are overcoming these limitations. Online businesses are playing a significant role in changing user expectations when it comes to online experiences. Other things play a role in these changing expectations as well. For example, the sheer numbers of people who use the Internet make it easy to find other people like oneself and discuss increasingly specialized issues. While we are becoming more homogenized we are also becoming more specialized. People are bringing these changing expectations to my library. They are saying, ... "Library, you already know who I am because I have to log in before I can use your services. After I grant you permission, why don't you remember me, keep track of what I do, and offer suggestions? After all, I did such a thing previously when I visited you in person. Ms. Peabody was very helpful when she..." ... if libraries do not provide the products and services people need and desire, then the funding for libraries will dry up. Eric Lease Morgan. Response to questionnaire.
It is stating the obvious to say that in business, personalisation works where there is a clear business case. It should be equally obvious that any application of personalisation in education should have a clear and defined business case. Research and development activity, in both areas, can be less driven by specific objectives, as long as the organisation can clearly distinguish between r&d and application.
We have found little if any systematic user requirements analysis in this field - presumably the major commercial players have conducted their own which is regarded as commercially sensitive. It seems that personalisation is often considered to be an alternative to user testing, with the rationalisation that if the interface, the options and the data are sufficiently flexible the user will choose the most suitable elements. This argument does not convince us. We prefer to agree that:
The majority of UM [user modelling] literature addresses technical frameworks and modelling techniques with scant reporting of user testing. According to Chin (2001) "A quick scan of the first nine years of UMUAI [user modelling and adaptive user interfaces] reveals that only about one third of the articles...includes any type of evaluation. This is much too low of a percentage". Further research on the usability of UM and its effect on technophobic and inexperienced users would greatly add to the understanding of how users perceive and react to such systems and may improve its further development and uptake. de la Flor (2004)
and, in more robust terms:
Rather than spending extensive resources on personalization, Web designers should:
- run usability studies
- structure the site according to the user's view of the world
- write content that is optimized for the online medium
Of course, these steps do not have the magic ring of "let's fix it with some cool technology", but they do have the advantage of working every time (and being cheaper, too). Jakob Nielsen's Alertbox for October 4, 1998
Things have changed since 1998 and we may argue about the relative costs, but the fact remains that developers and those who commission systems tend to be more interested in "cool technology" than in prosaic user testing.
The brief for this study focuses on "presentation services" but we must mention that personalisation can take place outside of the large service context. Email was mentioned more than once as being a personalised medium, Tom Franklin insisting that "push technologies" are a key part of personalisation. The "bottom-up" approach to personalisation can take a number of different forms. Social networks like FOAF <http://www.foaf-project.org/> require personalisation to fulfil their core function and to provide a meaningful end-user interface to the network. Since the mapping of personal relations is one of the main features of social networks, personalisation needs are inherent. Indeed, within the context of adaptive personalisation, a FOAF-like tool like could be used as a 'recommender system' and thus as an alternative to collaborative filtering algorithms.
Peer to peer networking, though popularly portrayed as subversive and pseudo-criminal by the media and others who have a vested interest in such a portrayal, is much more than merely a threat to the recorded music industry. It has the potential to allow a high degree of user control over personalisation of information, communication and entertainment, a fact which has been recognised in academic projects like Lionshare <http://lionshare.its.psu.edu/main/> which aims to facilitate legitimate file-sharing among individuals and educational institutions.
While one of our respondents warns that:
Adaptive personalisation tends to have a connotation of "Big Brother is watching you". Users should be able to turn such features off. Questionnaire respondent
it could be argued that adaptive personalisation in a peer to peer environment might turn this argument on its head, by giving the end user ultimate control over incoming and outgoing traffic
The original Internet was fundamentally designed as a peer-to-peer system. Over time it has become increasingly client/server, with millions of consumer clients communicating with a relatively privileged set of servers. The current crop of peer-to-peer applications is using the Internet much as it was originally designed: as a medium for communication for machines that share resources with each other as equals. Oram
Weblogs or blogs are another example of bottom up personalisation. Whole communities with complex structures, idiosyncratic rules and mores are developing and personalisation is a key element in these communities. Blogs can certainly be an element of larger services, particularly in a collegial environment although
Weblogs go some way but not everyone wants a soap-box. Software enabling "personal digital repositories" is another dimension that would be very useful -- such repositories could be the means for fast-tracking the development of larger institutional repositories or "community of practice" collections. The e-portfolio movement has some potential but has been sidetracked by the "personal transcript" interpretation of such an application. ... Could a service evolve as a result of leading practice? Take a good look at the vibrancy of the various communities using blogs & syndicated content. ... Think outside the box of "delivering a service" -- it's almost always too top-down in conception. Could a service evolve as a result of leading practice? Questionnaire respondent
The boundary between the service approach and the bottom-up approach becomes blurred when a social networking service is highly centralised with restrictive data ownership conditions, for example Google's Orkut <http://www.orkut.com/>