User Issues (Library and Information Studies)

The mass adoption of Web 2.0 type resources such as FlickR, Myspace, YouTube and Del.icio.us means that large numbers of users are now attempting to archive and describe digital objects without recourse to expert classification systems. This is widely acknowledged to have positive and negative implications in terms of the validity and effectiveness of data, but in terms of systems design, there is an interesting ‘desire line’ principle that might be applicable to the construction of indexes and thesauri. ‘Desire lines’ are the worn areas of a landscape, usually quite narrow tracks, where walkers have decided to take a route that doesn’t correspond to, or is completely at odds with, a formal path. Peter Merholz suggests that ‘a smart landscape designer will let wanderers create paths through use, and then pave the emerging walkways, ensuring optimal utility.’

Taking this approach as a model for building effective vocabularies might mean that they take a long time to become established, but used in combination with existing systems, this method could add value. At a recent Methods Network workshop, one of the participants mentioned that the British Library records all search terms entered into the catalogue – whether they return successful results or not – in order to ascertain what search terms people would prefer to use, rather than simply being able to gauge which terms from an existing list are used most regularly. Clearly there is an argument for saying that a lot of community-generated terminology will be faddish and that terms that appear to have significant collocation at one point in time (e.g. Iraq/war-torn – circa 2007) may be subsequently misleading. Counter arguments might point to the value of such conceptual relationships as chronological markers, if temporal data can be applied to the linkages between terms.

The more orthodox approach in terms of LIS principles is to provide structured, secure and authoritative systems that ensure appropriate users have access to licensed and reliable sources of information. One of the enabling components of this strategy is to be able to authenticate and authorise users to access relevant systems and a key tool that is currently being promoted by JISC and the UK Federation for Access Management is called Shibboleth. It defines a set of protocols written in SAML (Security Assertion Markup Language) that enable the passing of secure identity information between institutions and service providers and it allows organisations to provide a single logon for users to access all of the local and remote resources that they are authorised to use. The principle advantage to the user is that they are no longer required to remember a multitude of passwords to access a range of distributed resources. The institution benefits by no longer being reliant on third parties to reset accounts when problems arise, as the devolved architecture of Shibboleth means that the process of establishing who you are (authentication) is separate from the process of establishing where you can go (authorisation), the former being entirely dependent on a single locally-assigned password.

Getting the balance right between creating secure and robust systems, whilst at the same time keeping them attractive, intuitive and usable, is a significant area of LIS research. The use of online surveys and the collection of statistics via transaction log analysis (TLA) are two ways of building up a picture of how users engage with web resources, although both methods have limitations as well as strengths. Online surveys can be valuable for revealing user’s motivations, goals, attitudes and satisfaction levels but it is very difficult to ensure that the survey sample is representative of the user group that is being targeted. Another problem is that users are increasingly reluctant to engage with questionnaires, a factor which does not impact upon TLA methods as data is automatically harvested from computerized log files and provides the analyst with an objective view of user behaviour rather than having to rely on error-prone human accounts of activity. Where TLA methods fall short however, is their inability to accurately identify and track individual users of systems. Log files rely on IP addresses or hostnames to identify client computers and there is no way of ensuring that use of the client system is limited to one defined user. Cached pages and unattended active sessions may also cause problems with the reliability of the results. A combination of both approaches however can prove to be very effective for understanding website usage.

COUNTER (Counting Online Usage of Networked Electronic Resources) is an international initiative that focuses on the effective recording and exchange of online usage statistics. Aimed at libraries, publishers and software vendors, it also liaises with a number of other organisations, one of which, the Association of Research Libraries (ARL), is responsible for a suite of tools and services called StatsQUAL. Comprising of three different component systems, LibQUAL+, DigiQUAL and MINES for Libraries, they are designed to enable effective assessment of the role, character and impact of physical and digital libraries.

One of the critical tasks of any systems development project is to ensure that the human-computer interface (HCI) is thoroughly tested and is entirely fit for purpose. Zabed et al give details of a sophisticated iterative process of designing and evaluating an IR interface and recount the eight-step sequence that resulted in the development of a prototype. Offering quantitative statistics over a number of categories relating to relative satisfaction levels of novice and experienced users, the authors manage to arrive at a number of recommendations for IR interface development.

Syndicate content