Suggestions for a research framework for South Africa : how can we learn from web information seeking / searching studies ?

The growing interest in web informotion seeking/seorching studies, ond their importance for web site design. search engines, intranets, portals, etc. warrants. a research agenda for web information seeking/searching studies in South Africa. A research framework directs research projects and should include: the rationale for and purpose of a study. descriptive information, links to related studies and traditional information retrieval, links to research paradigms and theoretical models, research methodology and methods, rationale for analysing the data, actual statistical analysis, interpretation and findings. and explicit reference to the practical impact of and need for further research. It should build on the importance of research in a particular context, and acknowledge the existing body of research literature. Suggestions are offered for practical and theoretical research projects relevant to the South African context. Although we can benefit from isolated projects, a research agenda, as suggested, may add greater impetus to our research efforts and understanding of the phenomenon under investigation.


Introduction
The exponential growth in web sites, web users and accessto web-based information resources is an indication of their growing importance.The World Wide Web .0NWW) is used by a diverse group comprising scientists, researchers, academics, students, information technology (IT) specialists, members of the general public, schoolchildren and elderly citizens.A growing interest is reported for the last three groups (Bilal, 2000, 200 I;Burwell, 200 I) and in information seeking for everyday needs (Carey et 0/., 200 I) and decision=making (Choo, 200 I, 2002).In South Africa there is also a growing interest in the web as an information source.Apart from its potential to meet academic and businessinformation needs, it can also be an important resource filling the everyday information needs of South Africans.Spink and Cole (200 I :303) note that: [The] Internet is the driving force behind broadening LIS analysis of information seeking beyond work and school.The interactive potential of this hybrid information flow channel should bring the power of information use to many more sectors of society than is now the case, but it also forces researchers interested in these issues to take a wider, more integrative approach to studying information seeking and use that includes nonseeking behavior in its human information behavior perspective.
Web information seeking/searchingstudies can shed light on the design of user interfaces, search engines, navigational features, online help and intelligent agents, information architecture, content description and metadata, the teaching of information skills (e.g. to schoolchildren or senior citizens) and the refinement of our information retrieval (IR) research methodology.The findings may also be useful in web advertising, marketing and e-commerce (Ozmutlu et 0/.,2002), and in drawing interest from designatedtarget groups (e.g.getting teenagers interested in usingthe library).By understanding the variables (also called barriers or influencing factors) that affect web information seeking/searchingwe can improve all facets of web information spaces.Many such variables have been identified and tested in traditional IR (Wilson I 999b).
More recently Hertzum et aJ.(2002), for example, reports on the notion of trust in the information seeking behaviour of software engineers with special reference to virtual agents and e-commerce.Spink and Xu (2000) conclude that 'continued research into Web user behaviour is needed to impact the development of new types of user interfaces and software agents to aid users in better Web searching'.
Information seeking is also important in the context of virtual or online learning environments (e.g.Sheremetov & Arenas, 2002; Yang, 200 I).In the South African context this is particularly important since a number of tertiary institutions have already adopted virtual learning environments to supplement their teaching (e.g. the University of Pretoria), and because of the national emphasis on information literacy skills (addressed in the skills prescribed by the South African Quality Authority).It is therefore clear that not only libraries and information services but also society at large will benefit from a research programme for web information seeking/searchingbehaviour.This article builds on a paper presented at the ProLiSSA 2002 conference (24-25 October 2002, Pretoria, South Africa) (http://www.dissanet.com.).
The conference paper presents a literature review of web information seeking/ searching studies for the period 2000 to July 2002, suggestions for a research framework and research topics of interest to South Africa.To ensure extensive coverage the following databases were searched: ERIC, Wilson Library Literature, Information Science Abstracts, Emerald, ScienceDirect, and the 151 citation databases.Library and Information Science Abstracts could not be accessed.References in bibliographies were also followed up.This article will expand on the initial research framework, and will suggest a more detailed research agenda for South African web information seeking/ searching studies.The suggestions will be linked to benefits to be gained from web information seeking/searching studies and lessons learned from the existing research corpus.In addition to the databases researched for the conference paper, South African Studies was searched to identify literature relevant to a South African research agenda.
This article will cover the following: clarification of key concepts, the need to link WWW information seeking studies to traditional Boolean IR studies, a brief overview of the status quo of web-information seeking/searching studies, benefits to be gained from web information seeking/searching studies, and suggestions for a research framework for South Africa.

Clarification of concepts
Researchers are often accused of not qualifying their use of concepts.Among other things this makes it difficult to verify or compare research results.It is therefore essential to clarify key concepts such as information seeking, information searching, web spaces and web logging for the purposes of this article.
Information seeking and information searching both belong to the broader field of user studies.According to Wilson (1999b), information seeking and information searching models function in a complementary manner and address issues at various levels of information behaviour.For the purposes of this article we will adhere to Wilson's (1999b:361) explanation of information seeking/searching: [I]nformation behaviour may be defined as the more general field of investigation ... with information-seeking behaviour being seen as a sub-set of the field, particularly concerned with the variety of methods people employ to discover, and gain access to information resources, and information searching behaviour being defined as a sub-set of information seeking, particularly concerned with the interactions between information user (with or without an intermediary) and computer-based information systems, of which information retrieval systems for textual data may be seen as one type.
The WWW may be considered a new information space that differs from traditional information spaces such as traditional online databases, OPACs and in-house databases.According to Wang et 01.(2000), the WWW as an information space consists of: • Digital objects (including content, expression, relation, structure, hyperlinks) • Activated objects (e.g.portions of the web that have been activated during interaction, such as home pages that are stored in the cache of a local computer) • Web spaces (i.e.partitioned spaces of the web, such as search engines, directories or specific websites).These would also include home pages, intranets, portals and specific services offered (e.g.document delivery services) • Organisational schemes (e.g.how each space is organised) • Metadata (descriptive and representative data attached to an object and which provide access to the object) The WWW can also accommodate traditional information "spaces" such as in-house web-enabled databases, databases available through commercial database and aggregator services such as Dialog, library catalogues and CD-ROM databases available through an intranet.Libraries and information services should especially monitor users' navigational behaviour in identifying and selecting information channels accessible from their web sites (e.g.databases, library catalogues, full-text journal services, suggested search engines, document delivery services and current awareness services).Apart from research on web-enabled library catalogues (Cooper,200 I;Cooper & Chen,200 I), no reports could be traced on "traditional web-enabled spaces".Any of the above-mentioned web information spaces should actually be covered in an extensive research agenda.
Web log analysis is used in most web information seeking/searching studies to collect and analyse information on users' actions and moves.Other terms that are found in the literature include web logging, web log file analysis, transaction log analysis, web tracking and log file analysis.

Is it necessary to link to Boolean IR research?
There are different opinions on whether web information seeking/searching reflects totally new information behaviour, or whether it is similar to traditional IR.Although there is a great deal of overlap between traditional IR and the web, there are researchers such as Jansen et 01.However, Jansenet 01.(2000) also stress the fact that we need more comparisons between traditionallR and web search findings.This would include mostly Boolean based IR research such as reported by Borlund (2000).
A few obvious differences between the web and traditional IR include a difference in the user base, much wider accessibility and interest, a variety of digital formats (text, hypertext, sound, video, animation) Unfortunately current web information seeking/searching studies appear, on the whole, to lack reference to traditional IR research projects, theories and models of information behaviour.This is in spite of the many useful publications that could support the theory building of web information seeking/searching studies, e.g.Ingwersen (1992Ingwersen ( , 1999)), Palmer (1999), Schamber (2000), Wilson (1999a, I999b), Vakkari (1998Vakkari ( , 1999, 200 I), 200 I), Belkin et 01. (1995), Spink (1997), Kuhlthau (1991), Dervin (1999), Marchionini (1995), and Ellis and Haughan (1997).Some of these models have already been empirically tested in general and in the web environment (e.g.Kuhlthau & Tama, 200 I ;Byron & Young, 2000), but they should be used much more extensively.
In a recent joint project, Spink, Wilson, Ford, Ellis, Lam, Burton and Foster investigated mediated information seeking.In a series of articles they report on successivesearching (Spink et 01., 2002c), uncertainty and its correlates (Wilson et 01., 2002), cognitive styles in information seeking (Ford et 01.,2002), user-intermediary interaction (Ellis et 01.,2002) and their theoretical framework and research design (Spink et 01., 2002b).These articles could also offer substantial theoretical support for web information seeking/searchingstudies.

Status quo of web information seeking/searching studies ( a brief overview)
Before planning a research agendait is essential to note past projects and progress.Sincethe early 1990s there has been a steady growth in research projects on web information seeking/searching.In 1993, Peters noted that there was still very little research on internet searching (which is in line with the fact that it was still the early days of the more general availability of the Internet).In 200 I, Jansenand Pooch published a review of web-searching studies.They list 41 sources, of which six were published in 2000.Fourie (2002) covers a much more extensive list of publications for the period 2000 to July 2002.Since the publication of her paper a number of additional articles have appeared, namely Pu et 01. (2002), Spink, Ozmutlu and Ozmutlu (2002), Pettigrew et 01. (2002), Cooper (2002) and Yu and Roh (2002).Jansenet 01.(2000) noted that there is a lack of scholarly research in particular.Jansen and Pooch (200 I) distinguish between primary web-searching studies (investigating web search engines as primary web IR systems) and secondary web-searching studies (i.e.studies providing a limited amount of web user data or that focus on a narrow aspect of web user searching).The latter includes studies of information seeking on a single web site (not a search engine), multiple web sites, searching using relevance feedback, multimedia searching, query terms, presentations on web searching, and information on the web about the web.Fourie (2002) categorises web information seeking/searching studies according to their research foci.She distinguishes research reports according to the following: single search engines, single web sites (not search engines), comparisons between similar web sites, unspecified searching of multiple websites including search engines, analysis of aspects of web searching (e.g. the searching of specific entities or media types, search queries as focus, users' perceptions, web navigation, search strategies), influencing factors (e.g.gender differences, age groups, age differences, levels of experience, cross-country and cultural differences, trust, network competence, cognitive, affective and task influences, specific occupations, disciplines or domains, usability), innovative monitoring tools, special software and search support, theory building and the development of behavioural models, reviews of trends, and longitudinal studies.refinement of research methodology, references to and comparisons with traditional IR, web spaces and web objects studied, conceptualisations, target groups, types of behaviour monitored (e.g.cognitive, affective and physical behaviour), influencing factors monitored, and intervening measures taken (e.g.software to monitor or support the search process).Fourie (2002) notices that there are still very few joint research projects (exceptions are the work for the Excite project: Jansenet 01., 2000; Spink & Xu, 2000;Spink et 01., 200 I).Resultsfrom a large-scale,international collaborative research project have, however, been recently reported by Spink et 01. (2002bSpink et 01. ( , 2002c)), Ellis et 01. (2002), Ford et 01. (2002) and Wilson et 01. (2002).Although the project does not focus on web information seeking/searching per se, it does offer valuable insights.
South African publications on web information seeking/searching In the international literature research reports concerning web information seeking/searchingalso appear in journals for computer science, informatics, education, instructional design and the natural sciences.An extensive literature search was therefore conducted on South African Studies to identify South African publications relevant to web information seeking/searching.Only a study by Icheva (1998) was identified.No publications dealing with web logging or web tracking, a key method for collecting quantitative data for web information seeking/searchingstudies, could, for example, be found.There therefore appears to be a need for an accelerated research programme in South Africa.

What can we gain from web information seeking/searching studies?
Web information seeking/searchingstudies should offer solutions to praxis, and should help us to build a sound theory.Models of information seeking/searching behaviour can, for example, offer a framework for thinking about a problem, describe activities and/or the causesand consequences of such activities, offer a source of hypotheses to be tested (e.g.relationships between stages in information seeking behaviour), and help to identify gaps in existing research (derived from Wilson,199%).The research projects reported in the subject literature highlight a number of benefits that can be gained from such studies.Only a small selection of the benefits to be gained will be briefly considered.Ĩmproved search engines or portals for designated target groups (e.g.children, elderly citizens, minority groups) The research by Bilal (2000Bilal ( , 200 I, 2002)), Bilal and Kirby (2002) and Large et 01. (2002aLarge et 01. ( , 2002b) ) shows that children need special support for their web information seeking/searching.Bilal, for example, found that Yahooligans!(a search engine designed for children aged 7 -14) is not meeting the needs of their intended target group.The children used incorrect search syntax, misspelled keywords and made extensive use of looped searchesand hyperlinks that lead to disorientation (Bilal, 2002: I 10).She suggeststhat Yahooligans!introduce a spell-checker, context-sensitive help, search and browsing instructions and examples and a natural language interface among other things.Large et 01.(2002a) used children's comments on search engines (also called portals) for younger users to make suggestionsfor the improved design of such portals.They covered Ask Jeeves for Kids, KidsClick, Lycos Zone and Yahooligans!They found that children find entertainment opportunities and lots of colour attractive.They also prefer larger fonts and animation, the vocabulary should be suitable for the intended target group and subject categories should be available along with the keywords (Large et 01., 2002a:89-91).
These studies were all conducted with children in the United States and Canada, and one would expect that South African children would also show a need for search engines that acknowledge their unique physical, affective and cognitive behaviour in searching the web.No South African search engines or portals for children could be identified.
In the United States elderly citizens show great interest in searching the web (Burwell,200 I).It might be interesting to verify if there is a similar trend in South Africa, and whether any cultural differences are reflected in dedicated portals.
Although research has been reported on the information seeking behaviour of minority groups, this has not been specifically linked to the web.This might be another research focus of interest to South Africans -especially if we can focus on possible cultural differences in order to design appropriate portals.

Improved.search features
Research projects on single search engines and comparisons of search engines have shed light on important search features that could be added (e.g.buttons for image searching to the Excite search engine) (Goodrum & Spink, 200 I;Jansenet 01., 2000;Spink & Xu, 2000).An analysisof the search queries people submit to search engines could also shed light on their use of keywords, combination of keywords, use of search operators and the length of queries (Moukdad & Large, 200 I;Ford et 01., 2002;Ross& Wolfram, 2000).Other studies that focused on search queries include the work of White and livonen (200 I), Iivonen and White (200 I), Jansen(2000), Lucasand Topi (2002), Dennis et 01. (2002), Radevet 01.(2002), Pu et 01. (2002), and Jansenet 01., (2000).Xie (2000Xie ( , 2002)), for instance, believes that in order to design IR systems to support various information seeking strategies across a variety of contexts, it is important to explore what leads to the selection of different types of information seeking strategies.This type of study is essential for businesses, & Info Sci 2003, 69(2) service organisations, companies with intranets and educational services.South African search engines could certainly benefit from such studies.

Improvement of website navigational features
This aspect is strongly linked to the improvement of search features.All website owners/masters can benefit from a better understanding of how people navigate their websites.Such studies have been reported on by Rozic-Hristovski et 0/.(2002) among others.There is also a need for studies of software that can support web navigation or the monitoring of web navigation (e.g. as reported by Danielson, 2002).Libraries and information services that make their catalogues available through the web could find such studies particularly beneficial.Cooper and Chen (200 I), for example, report on a study to predict when a person's search on the web-based catalogue would be perceived to be relevant.
Comparisons between similar websites (e.g.financial institutions, consumer sites) can also offer valuable insights.
Nicholas et 0/.(2002) report on a comparison between two consumer health websites.
Encouraging and improving the quality of information seeking in a working environment Effective information searching will become increasingly important in knowledge organisations (Choo, 200 I;Ellis & Haughan, 1997).Choo et 0/.(1999, 2000a, 2000b, 2000c) report on how knowledge workers use the web to seek information.They developed a behavioural model of information seeking/searching that allows for four modes of searching based on the work of Arguiles, namely undirected viewing, conditioned viewing, informal searches and formal searches.It also allows for six activities (starting, chaining, browsing, differentiating, monitoring and extracting) identified in earlier studies by Ellis, Hall and Cox (1989) and Ellis and Haughan (1997).Based on their findings Choo et 0/., (2002c: 186-187) suggest that portals should be designed to support undirected, serendipitous viewing, users should be trained to evaluate information and to use web notification services, and good pre-packaged search strategies should be provided.
Their findings also have implications for intranet design, which should promote information and knowledge sharing.The value of studies concerning people's work or task-related information seeking/searching behaviour is also pointed out by Yakkari (1999, 200 I), Brystrom (2002) and Brystrom and Jarvelin (1995).Their studies do not focus on the web per se, but offer valuable insights into issues to consider.Crowley et 01. (2002) used focus groups to collect information on users' perceptions of a library's web page.Such studies are essential for improving site quality and accessibility.
Finding improved methods to support people's information seeking/searching efforts A growing interest in ways to support users in their search efforts is noted in more recent research projects.Some of these are the use of constantly visible site maps (Danielson, 2002), self-organising maps (Smith & Alan, 2002) or special software for search support (Dempsey et aI., 2000), automated document clustering (Roussinov & Chen, 200 I), and information visualisation techniques (Heo & Hirtle,200 I).
Developing web information spaces allowing for cultural differences The possible impact of cultural differences is still very under-researched.Iivonen and White (200 I) and White and livonen (200 I) investigated the initial choice of search strategies and whether there are any cultural differences.They draw a comparison between Finnish and American searchers.
Although not in the context of web information seeking, Xie (2000) argues the need to design IR systems for supporting various information-seeking strategies across a variety of contexts.This requires an exploration of different types of information-seeking strategies and cognitive behaviour.With its rich cultural variety, South Africa could make a valuable contribution to the study of the possible impact of cultural variables.It may for example be interesting to note whether non-native English speakers use the Web as much as English-speaking communities, as well as how they use the web (e.g.choice of search terms, criteria for selection of web resources), how the web is used by communities with a strong oral tradition, and whether the web is equally accessible to all members of the South African community (with acknowledgement to an anonymous referee for some of the suggestions).

Supporting collaborative information seeking/searching
There is a growing interest in collaborative information seeking/searching.Fidel et 0/.(2000) and Large et 0/. (2002b) report on gender differences between grade 6 boys and girls in collaborative web searching.Collaborative learning environments are also being mentioned more frequently (Zhang, 2002;Zafeiriou et 0/., 200 I).Due to the potential of the web as an electronic learning environment, it would be worthwhile to pursue studies on collaborative information seeking/searching.The author is currently working on such a project.& Info Sci 2003, 69(2) Improved information literacy and information research skills Information literacy programmes have become very important as is clear from the numerous articles and conference papers on the topic.Information literacy is also linked to programmes on information research skills (Fourie & Van Niekerk,200 I).

SAJnl Libs
Research by Bilal (2000Bilal ( , 200 I, 2002) ) has 'shown that it is essential to understand people's information seeking/ searching behaviour on a physical, cognitive and affective level.Kuhlthau's model of the information search process (ISP model) has shed light on information searcher's affective behaviour (e.g.feelings, emotions and anxiety) in particular (Kuhlthau, 1991(Kuhlthau, , 1993)).The importance of the information seeker's cognitive state and cognitive behaviour is stressed by Ingwersen (1992); the research by Vakkari (1999, 200 I) and Bystrom and Jarvelin (1995) stresses the impact of task complexity.Savolainen (2002) argues that network competence will affect information seeking on the web.
Against this background it appears essential to improve our understanding of people's cognitive, affective and physical web information seeking/searching behaviour, so that we can adapt information literacy and information research skills programmes to allow people to operate effectively in terms of their search skills and information usage in environments typical of knowledge and learning organisations and also in collaborative work/learning.Cognitive studies can especially be important for research on different cultural groups, and people from formerly disadvantaged communities to monitor for example the impact of a digital divide.
Differentiating between expert and novice searchers Traditional IR studies have shown that search experience can impact on search behaviour and the quality of search results.Lucas and T opi (2002), Holscher and Strube (2000) and Lazonder et al. (2000) compared expert and novice searchers in order to identify the types of knowledge and knowledge structures required.Kim and Allen (2002) report on the differences in users' cognition and search tasks, while Palmquist and Kim (2000) investigated cognitive style and experience as predictors of web search performance.If we understand the impact of experience, we can accommodate it in our training programmes, the design of web spaces, interface design, and even our organisational structures (such as metadata and taxonomies) for such web spaces.
Theory and model building There are unfortunately still too few research efforts focussing on building a theory of web information seeking/searching behaviour.A need for such efforts can be linked to calls for the improvement of a theory of information behaviour and the improvement of our research methodology (e.g.Smith & Alan, 2002).In an editorial introduction to a special issue of Information Processing and Management.focusing on information seeking in context (ISIC), Kuhlthau and Vakkari (1999) stress the need for methodology leading to new approaches to research methods that address broader aspects of users' lives than formerly recognised: 'This is the early stage of what is developing into a significant body of research that holds the potential for making a major contribution to the knowledge base of information science.' Choo (200 I), Choo et al. (1999Choo et al. ( , 2000aChoo et al. ( , 2000bChoo et al. ( , 2000c) ) and Wang et al. (2000) have made some progress in contributing to such a theory.Choo and his associates have developed a behavioural model of web information seeking behaviour.It is strongly embedded in traditionallR research findings and models (e.g. the work of Kuhlthau, Wilson, Allen and Ellis).Wang and his associates made an effort to offer a holistic model for user-web interaction.We should certainly aim to add to such research efforts.

Suggestions for a South African web studies research agenda
In preceding sections the value of web information seeking/searching studies and its status quo have been briefly considered.It is clear that substantial web information seeking/searching studies are necessary to refine our knowledge of web information spaces, their design and maintenance and training-related issues.The same applies to the theoretical basis.To support such research efforts, we need a research agenda (working according to a plan to get things done) with a research framework.Although we can benefit from isolated projects, a research agenda may add greater impetus to our research efforts and understanding of the phenomenon under investigation.
In an earlier conference paper Fourie (2002) suggests a number of both practical and theoretical research topics.These include information-seeking behaviour on specific websites; information seeking/searching related to entertainment and online shopping; comparisons between similar websites; the testing of intervening measures such as training efforts, intelligent agents, filtering mechanisms, adaptive IR systems for a specific South African audience; the effectiveness of metadata and indexing practices; searching on textual as well as non-textual collections (e.g.multimedia, video, images, audio); the effectiveness of site architecture and its impact on information-seeking/searching behaviour; web-based environmental scanning and the use of current awareness services; the impact (if any) of cultural diversity (the variety of cultures in South Africa should make this a rich field for future research); the cognitive processing required by users (e.g. from different cultures); surrogate information that can support relevance feedback; the usefulness of query When selecting a research topic, researchers should take note of their own "philosophies", personalities, experience and interests, because these will have an impact on their choice of research methods and research paradigms.Ford (/999) links research approaches to the different information processing styles of individuals.
When formulating a research agenda and framework we should build on traditional IR research experience.We should also learn from past mistakes that have hindered progress in IR research.Ellis (1990:viii), for example, remarks: 'In some respects contemporary information retrieval research comprises a plethora of different approaches and techniques, each with its own adherents and opponents, experts and neophytes.The first encounter with the literature can be baffling and confusing.'A research agenda A research agenda should serve as a plan of what to do.The following are suggestions for a South African effort.It is based on the preceding sections and the conference paper by Fourie (2002).It attempts to stimulate South African based research efforts that will improve local web information spaces and training practices, and that will provide us with the research experience that is essential in designing effective intranets, portals, collaborative learning environments, etc.The agenda entails the following: • Literature reviews to establish the status quo of web information seeking/searching studies.Attempts have been made by Jansen and Pooch (200 I) and Fourie (2002).These should be supplemented with alerting services and frequent updates.• A research framework.Attempts have been made by Jansenand Pooch (200 I) and Fourie (2002).Spink et 01. (2002b) also offer useful suggestions.Suchframeworks should be revised and improved on an ongoing basisasmore experience is gained with web information seeking/searching studies, and as we build our theoretical base.• An analysisof existing web information seeking/searching models.If necessary new theoretical models can be proposed to be tested empirically with regard to South African target groups.(Fourie is currently working on a conference paper in this regard.The paper will be delivered at the 6 th Conference of the ISKO Spanish Chapter, 5-7 May 2003.)• An extensive overview of traditionallR research and findings to support web information seeking/searching studies.
It is essential to realise that although web information seeking/searching may be totally different from traditionallR, there is much to be learned from earlier theories, or at least by identifying theories worth testing in a web environment.• A review and reconsideration of the methodologies used in web information seeking/searching studies.These include quantitative and qualitative methodologies.Although much has been published on web tracking for example, there are still many problems.We need to refine the methods as well as the tracking software.Both established and new methods to gather qualitative data should also be considered, so that we can present 'information rich' results.Wang (1999) offers an excellent review article on methodologies and methods for user behaviour research -the umbrella field for information seeking/searching.• A study of the impact and suitability of the different research paradigms, namely the systems-oriented paradigm, usercentred paradigm, cognitive paradigm and the socio-cognitive paradigm.This should be mapped to the purposes of reported research projects, and their potential for future projects of interest.This could serve as a guideline to researchers when selecting research methods.• A detailed survey of web information seeking/searching and IR literature to identify other disciplines that may support our research efforts (e.g.motivational theory, Phenomenology).• A mapping of the purposes and foci of reported research projects (including the identification of web information spacese.g. a library catalogue, organisational intranet, subject portal, specific search engines) and of target groups (e.g.elderly citizens, school children, engineers, lawyers).Both the web spaces and target groups should be linked to the purpose of the research project (e.g. to collect data on students' physical, cognitive and affective behaviour in order to improve information literacy training programmes and portal interfaces, or the inclusion of web information seeking in school curricula).• An inventory of variables (e.g.influencing factors such as culture, age, gender) that may affect information seeking/ searching.These should be systematically tested in the South African context.• A list of projects that lend themselves to longitudinal and comparative studies (e.g.comparisons between children or students from different backgrounds, or the changes in the information seeking/searching behaviour of students over a three year period).• A list of projects that lend themselves to international, collaborative research efforts.Jansen and Pooch (200 I :242), for instance, believe that the framework they propose will support comparative and contrasting studies: 'Researchers utilising common terminology and following a similar framework whenever possible will improve the ability to make comparisons among research results.'In their suggestions for a research framework they allow for descriptive information, analysis presentation and statistical analysis.Fourie (2002) suggests a preliminary framework that allows for the following (more detail can be found in Fourie, 2002): • Rationale and purpose (these should be clearly spelled out and linked to decisions on research methods and the nature of the data to be collected).• Descriptive information (e.g. the background information on the web information space being investigated; specific features of the web space as an IR system; demarcation of the study and the rationale for the demarcation; size of the study in terms of the number of searchers and the queries collected; description of the target group; influencing factors considered; the experimental setting; nature of the queries [real-life or controlled, fact related, research tasks, language of the queries, language of the document collection]).• Links to related studies (especially comparative projects and traditionallR research).
• Links to research paradigms, theoretical models and information seeking/searching models.
• Research methodology and methods (e.g.type of methodology, research methods, type of data collected, research instruments, measuring units, assumptions, relevance judgements and their interpretation, guidelines given to participants.) • Analysis of data (methods and rationale).
• Interpretation of data and findings.
• Explicit correlations between findings and suggestions for practical implementation and further research.The framework should be revised as more experience is gained in web information seeking/searching studies.

Conclusion
Web information-seeking/searching studies are still at an early stage.Although some progress has been made, there are still a number of issues that need to be clarified.One of these is the link between traditional IR research and web information-seeking/searching studies, and the value of traditional information-seeking/searching models.In the South African context there are many possibilities for research projects.A major contribution could be made with regard to the impact of cultural diversity and cognitive behaviours.In order to improve web information spaces and web searching in praxis, and to contribute to an information seeking/searching body of theory, we need to work according to a research agenda and a research framework.It is hoped that a framework such as the one proposed in this article will support web information seeking/search studies, the building of models of information seeking/searching and the development of a substantial theory.
(2000) who are of the opinion that the Internet is being searched in a new way -perhaps users are even searching the web in ways that designers and IR researchers have not yet contemplated or assumed to exist.According to these authors, real-life Internet searching is changing information retrieval: SAJnl Libs & Info Sci 2003, 69(2) While Internet search engines are based on IR principles, Internet searching is very different from IR searching as traditionally practised and researched in online databases, CD-ROMs and online public access catalogs (OPACS).Internet IR is a different IR, with a number of implications that could portend changesin other areas of IRas well.
Fourie (2002) also analysesweb information seeking/searchingstudies according to trends that could be identified by making a comparison with the earlier study by Jansenand Pooch (200 I).The changesand trends concern the following and are reported in more detail by Fourie (2002): metrics, experimental situation, focus and nature of studies, research methods, shifts in research paradigms, study samples,duration of research projects, findings and conclusions, calls for the SAJnl Libs & Info Sci 2003, 69(2) SAJnl Libs& Info Sci 2003, 69(2)reformulation, query modification and relevance feedback; studies focusing on the various decisions web users have to make (e.g.their coping skills and search tactics, or the eventual use made of information); analysis of the different metrics used (e.g.search time, accuracy and search efficiency); refinement of research methodologies; failure analysis and linking it to training programmes; and web information seeking/searching in collaborative learning environments.
that serves as a point of departure for research should support a research agenda.Such a framework should help researchers to plan research and to write up their research reports.It should support comparative studies and the development of a theoretical basis and relevant research models.It should also offer guidelines on the content of research reports.
on the unique nature of the web, should be noted, it is of the utmost importance that our research efforts should build on what we have gained from traditional IR research.This point of view is supported by Jansenand Pooch (200 I) andVakkari (1999: 819)who remarks: 'A necessarycondition for this understanding is to link results from information seeking studies to the body of knowledge by IR studies.'