A Personalised Interface for Web Directories based on ... - CiteSeerX

22 downloads 389 Views 185KB Size Report
Implementing personalisation in Web Directories depends not only on .... analysis of the activity “[find] bingo software to play with friends”, may lead a participant.
A Personalised Interface for Web Directories based on Cognitive Styles George D. Magoulasa, Sherry Y. Chenb, Dionisios Dimakopoulosa,1 a

School of Computer Science and Information Systems, Birkbeck College University of London, Malet Street, London WC1E 7HX, UK {gmagoulas,dionisis}@dcs.bbk.ac.uk b Department of Information Systems and Computing, Brunel University Uxbridge, Middlesex, UB8 3PH, UK [email protected]

Abstract. Implementing personalisation in Web Directories depends not only on developing appropriate architectures and equipping Web Directories with adaptation techniques, but also on incorporating human factors considerations at an early design stage. Among a range of human factors this paper explores cognitive styles and their influence on users’ preferences. Preferences with respect to the organisation and presentation of the content, and the navigation paths are identified through a small-scale study. The findings are analysed and used to implement a prototype Web Directory Browser, gearing interface features to cognitive stylerelated preferences. Keywords. Personalisation, User Interfaces, Web-based Information Retrieval, Cognitive Styles.

1 Introduction The importance of personalisation has been demonstrated by research works in several areas, where human factors, such as level of knowledge, cognitive characteristics, purpose and goals have been shown playing import role in providing successful personalisation [1, 10]. In the context of Web-based Information Retrieval (IR), a number of previous works [2,6,7,8] have indicated that personalisation can support users performance in information seeking. Nevertheless, users of Web-based IR systems still encounter a number of challenging situations: (a) search results are not clearly interpretable and relevant to users’ individual preferences; (b) there is a lack of appropriate navigation support for users with 1

The authors gratefully acknowledge support by the UK’s Engineering and Physical Sciences Research Council (E.P.S.R.C. Grant References: GR/R57737/01 and GR/R92554/01).

different needs; and (c) search options and format presentations are not flexible enough to align with different users’ tasks, behaviours, and experience. One possible explanation is that most current search engines do not take into account human factors and serve all users in the same manner using an index of “global” importance based on the linkage structure of the web, such as in Google’s PageRank algorithm. In addition, browsing information organised in structured hierarchies, such as those offered by the Web Directories of search engines, also does not help many users: browsing is an inherently interactive activity, and relies on users constructing both a mental model of the information structure and the knowledge to be assimilated [3]. Some users face difficulties in forming a cognitive model or map of the structure and consequently they become lost in the information space [9]. Therefore, it is important to “learn” the underlying cognitive processes of the mental model in order to obtain a holistic view about the user’s behaviour [11,12]. Along these lines, this paper focuses on cognitive styles which are particularly influential in forming a mental model, as they relate to an individual’s preferred and habitual approach to organising and representing information [12]. The paper starts with a theoretical discussion on cognitive styles and then develops a new model of users’ profiling in Web Directories based on relationships identified between cognitive styles and content, structure and presentation aspects of Web Directories. Lastly, a prototype personalised interface is implemented.

2 Exploring the Preferences of the Cognitive Style Groups Cognitive styles are described as general tendencies of individuals to process information in particular ways. Within the area of cognitive styles, Witkin’s Field Dependence has been extensively studied, because it reflects the degree to which a user’s perception or comprehension of information is affected by the surrounding perceptual or contextual field: Field Dependent (FD) individuals tend to perceive objects as a whole, whereas Field Independent (FI) individuals focus on each part of the object [14]. Previous research has revealed that users’ cognitive styles significantly influence their reaction to the user interface in terms of formats [5], accessibility [4,12], and structure [3]. Results from these studies suggest that different cognitive style groups favour different interface features provided by web-based applications. Therefore, one may hypothesize: • FI users are less affected by external format/structure, whereas FD user are easily influenced by external format/structure; • FI users adopt an active approach in locating information, whereas FD users adopt a passive approach in locating information; • FI users tend to focus on detailed aspects of information space, whereas FD users tend to see a global picture of information space. A small-scale study was conducted with a sample of 57 computer science students to investigate the above hypotheses. All participants had the basic computing and Internet

skills. The research instruments included three Web Directories with different interface features, Cognitive Style Analysis to measure participants’ cognitive styles, and an Exit Questionnaire to identify users’ perceptions and attitudes towards the interface features of the examined Web Directories. Despite the fact that participants volunteered to take part in the experiment, they were extremely evenly distributed in terms of cognitive styles: 17 were FD (7 female/10 male), 21 were FI (10 female/11 male), and 19 were Intermediate (9 female/10 male). The experiment used a with-in subject design. Each participant used three specific Web Directories, namely Google, AltaVista and Lycos, in order to cover different types of Web Directories (see Fig. 1a-c). Indeed, these three directories differ in the interface features, content organisation, search results presentation and structure. Google was chosen for its simplicity in terms of interface design, whereas Lycos was chosen for its complexity. Furthermore, AltaVista adopts a design that comes in-between the other two designs. In order to identify users’ real perceptions, participants were allocated one hour to perform a practical task, designed to focus on users’ browsing through the content of each Web Directory and on content presentation. The content presented in each Directory varies, and this is reflected in the task activities shown in Fig. 1d; nevertheless all the activities are of the same nature, putting emphasis on analysis and synthesis. For example, analysis of the activity “[find] bingo software to play with friends”, may lead a participant to identify the keyword “bingo” and synthesise this keyword with the term “software” and with the requirement the software should allow multi players at the same time. The Cognitive Styles Analysis (CSA) by Riding was selected as the instrument to measure the cognitive style dimension investigated in this study because it allows Field Dependent competence to be positively measured rather than inferred from poor Field Independent capability. In addition, the CSA offers computerized administration and scoring. It measures what the authors refer to as a Wholist/Analytic (WA) dimension, considering it equivalent to Field Dependence [13]. A paper-based questionnaire was applied to collect participants’ perceptions and attitudes towards the Web Directories. Three open-ended questions were related to users’ opinions about strengths and weaknesses of each Web Directory as well as problems encountered. Thirty closed questions with a range of predetermined replies attempted to identify perceptions towards interface features; questionnaires were filled in in 15 minutes. Data collected from closed statements of the questionnaire were coded for analysis using SPSS 10.0. The participants’ cognitive style was the independent variable and their choices from a range of options were the dependent variable. The analysis used frequency tables in order to find differences among the three cognitive style groups. Data obtained from the Exit Questionnaire were used to identify participants’ most relevant preferences with respect to features of Web Directories. Among 30 closed statements, 10 items have shown significant meaning in our context and used for further processing. The preferences identified are explained in more detail in the statistical analysis presented in the following section.

(a) Google Interface Design

(b) Lycos Interface Design

Directory

(c) Lycos Interface Design

Activities - Web sites on chinese music. - Bingo software to play with friends. - Instructions on writing CVs - Description of Manchester Library and its Information Services - Personal data of Van Gogh. - Glossary of theological terms (d) Task Activities

Fig. 1. Web directories and information seeking activities used in the experiments.

3 Identifying Preferences and Relations to Interface Features The analysis of the user responses identified similarities in preferences of contents organisation, search results presentation, and navigation structure. In terms of the organisation of the information space, FD and FI users showed different preferences as far as the number of main categories and sub-categories are concerned. 11 FD users (65%) prefer contents organised in many main categories and fewer sub-categories, while FI users (N=15, 71%) favour a small number of main categories with many more subcategories. This may be interpreted as a tendency of FD users to obtain a global view of the information space; in direct matching with the holistic strategies they use. Conversely, FI users take a serialistic approach concentrating on procedural details. In addition, FI and FD users also favoured different ways for the directories’ categories organisation. FI users

(N=13, 62%) consider alphabetical order as the most effective way, whilst FD users (N=11, 65%) appreciate an organisation on the basis of relevance. This behaviour of FI users maybe indicate preference to active strategies for information seeking; deciding on their own path [14]. FD users’ behaviour, in contrast, tend to be more passive, as they rely on the level of relevance to guide them in finding out the meaningful information [14]. With respect to the presentation of search results, FI users prefer an alphabetical arrangement (N=12, 57%), considering understandable headings and sub-headings as the most important thing in finding information quickly (N=11, 52%). They are very goaloriented, and seem to exhibit strong ability in performing perceptual and conceptual tasks and actively segmenting information into relevant parts. By contrary, FD users (N=10, 59%) prefer the extra support offered when results are presented on the basis of their relevance. It was revealed that the holistic/serialistic strategies adopted by FD/FI users also influence their preferences of results presentation with respect to the sub-categories. Listing the sub-categories first, followed by the corresponding search results can help a FD user to get an overview of the relevant information space and available resources (N=11, 65%). On the contrary FI users adopt a serialistic strategy: seeing the results first followed by the sub-categories allows them to access their targets directly (N=15, 71%). In terms of navigation, the findings show that FI users prefer a depth-first navigation path, while FD users outperform in a breadth-first path. 11 FD users (65%) prefer main categories and their relevant sub-categories on different pages. Conversely, 16 FI users (76%) would appreciate the main categories and sub-categories to be presented on the same page, and 10 FI users (48%) prefer the sub-categories to be placed under the main categories. Furthermore, 15 FI users (71%) would like the subject categories are arranged vertically (from top to bottom), whilst 11 FD users (65%) favour that the subject categories are arranged horizontally (from left to right). These results imply that a FD user needs clear, planned structure and reinforcement in the use of Web Directories. Thus the system should provide them with authoritative guidance to restructure their personal information space, as the presence of massive information and the absence of external structure may delay FD users’ information seeking. However, the structure of the Web Directories does not interfere with FI users, whose cognitive skill is good in restructuring.

4 A Personalised Interface for Web Directories It has been made clear from the above analysis that FD/|FI groups’ preferences differ with regards to contents organisation. Fig. 2a presents an implementation example based on our Prototype Web Directory Browser (PWDB). Most populated second-level subcategories are placed under the main subject categories (FI users’ preference), whilst an overview of the sub-categories space is also provided through the sub-categories menu (FD users’ preference).

As far as search results presentation is concerned, one of the solutions to accommodate different cognitive style groups’ preferences is to allow the users to see both the results and sub-categories at the same time by using multiple frames. Multiple frames can provide navigation controls in one frame that always stay visible, even if the contents of another frame change. Fig.2b shows how the PWDB accommodates the needs of FI and FD users. Two frames are used to present information. The left frame lists the sub-categories for FD users to see all of the available resources, and the right frame allows FI users to examine the results directly. When the users click on a relevant result from the right frame, the corresponding web page is presented; the left frame remains unaltered in order to let users see all of the available resources. The resources shown in the right frame of Fig. 2b are arranged in order of relevance. Clicking on “View in alphabetical order” allows FI users to see the resources in alphabetical order. In terms of personalising the navigation structure, FD and FI users favour different paths to navigate through the space. In the PWDB, the pop-up window in Fig. 3 allows users’ access to the content through a sub-category’s hyperlink (this aligns with the preferences of FD users), whilst at the same time the top three resources-rich subcategories are shown under the main categories.

(a)

(b)

Fig. 2. (a) Contents organisation for FI/FD users using multiple menus. (b) Presentation of results in terms of relevance to Field Independent/Dependent users using multiple frames.

Our study also indicates that FI and FD users favour different ways to organise the subject categories. Fig. 4a shows an example of interface design that adopts alphabetical order (FI users preference). Moreover, FD users can get additional support by first rightclicking on the main category of interest and then on the “Show relevance” menu item. Subsequently, the relevant subject categories, such as for example “Games”, “Recreation” and “Sports”, are highlighted with the same colours (see Fig. 4b).

Fig. 3. Presentation of sub-categories to Field Dependent users using pop-up.

(a)

(b)

Fig. 4. Providing additional support to FD users using the “Show relevance” menu item.

5 Conclusions Web Directories let users decide their searching and browsing strategies by themselves. In this context, the user interface becomes the major channel to convey information: a well designed and friendly enough interface is thus the key element in helping users to get the best results quickly. This paper identified the interface preferences of three cognitive style groups by means of a small-scale study, and implemented a prototype Web Directory browser that accommodates their needs in terms of personalising the content, structure, and presentation.

References 1. Benyon D. and Höök K. Navigation in Information Spaces: supporting the individual.

In Human-Computer Interaction: INTERACT'97, S. Howard, J. Hammond and G. Lindgaard (Eds), London: Chapman and Hall, 39-46, 1997. 2. Borgman, C. L. Why are Online Catalogs Still Hard to Use. Journal of the American Society for Information Science, 47(7), 493-503, 1996. 3. Chen H., Houston A., Sewell R. and Schatz B. Internet browsing and searching: user evaluations of category map and concept space techniques, Journal of American Society of Information Science 49, 582-603, 1998. 4. Chen, S. Y. and Ford, N. J. Modelling User Navigation Behaviours in a HypermediaBased Learning System: An Individual Differences Approach. Int. J. of Knowledge Organization, 25, 67-78, 1998. 5. Chuang, Y-R. Teaching in a Multimedia Computer Environment: A study of effects of learning style, gender, and math achievement. Interactive Multimedia Electronic Journal of Computer-Enhanced Learning, 1, 1, 1999. [http://imej.wfu.edu/articles/1999/1/10/]. 6. Esichaikul, V., Smith, R. D., and Madely, G. R. The impact of learning style on problem-solving performance in a hypertext environment. Hypermedia, 6, 2, 101-111, 1994. 7. Ford, N. and Miller, D. Gender Differences in Internet Perceptions and Use. Aslib Proceedings, 48, 183-192, 1996. 8. Hicks, D. L. Tochterman, K. Eich, S., and Rose, T. Using Metadata to support customisation. Proceedings of the third IEEE Metadata Conference, Bethesda, MD (USA), 1999. 9. McDonald S. and Stevenson R. (1998) Navigation in Hyperspace: An evaluation of navigational tools and subject matter expertise on browsing and information retrieval in hypertext, Interacting with Computers 10(2), 129-142. 10. Magoulas G.D., Papanikolaou K.A., and Grigoriadou M. Adaptive web-based learning: accommodating individual differences through system’s adaptation. British J. of Educational Technology, 34, 511–527, 2003. 11. Nahl, D. Information counselling inventory of affective and cognitive reactions while learning the Internet. Internet Reference Services Quarterly 2(2/3), 11–33, 1997. 12. Palmquist, R. A. and Kim, K.-S. Cognitive style and on-line database search experience as predictors of Web search performance. J. of the American Society for Information Science, 51, 6, 558-66, 2000. 13. Riding, R. J. Cognitive Styles Analysis. Learning and Training Technology, Birmingham, 1991. 14. Witkin, H. A., Moore, C. A., Goodenough, D. R., and Cox, P. W. Field-dependent and field independent cognitive styles and their educational implications. Review of Educational Research, 47, 1, 1-64, 1977.