Published in the Proceedings of the Second British Computer Society HCI and Culture Workshop on “Culture and HCI: Bridging Cultural and Digital Divides”, June 2003, Karen Gunter, Andy Smith, Tim French (Eds.), 89-96 (ISBN 1-86166-191-6).

xCulture for the eEnterprise: an elementary 4-tier architecture for site localisation

Tim French Senior Lecturer Centre for Software Localisation Department of Computing and IS Faculty of Creative Arts and Technologies Luton University Park Square Campus Luton, UK Email: tim.french@luton.ac.uk

Simon Polovina Senior Lecturer School of Computing and Management Sciences Sheffield Hallam University City Campus Sheffield, UK S1 1WB Email: s.polovina@shu.ac.uk

Abstract

Cultural aspects of software localisation have hitherto been extensively investigated using a number of differing approaches both by academics, and by industrial practitioners producing web-site artefacts designed for ‘real’ clients. All too often however, we have seen a cultural divide between ‘cultural theorists’ and industrial practice. We aim to bridge this divide by not merely presenting ‘yet another model of culture’ but also critically, by describing in some detail, how such a model can form the basis of a ‘real’ commercial web-site using a novel 4-tier architecture. The aim of our work (which is formative, even speculative) is to be able to promote the deployment of localised web-sites that are not only culturally optimised, but are also inherently reusable, scalable, and explicitly derived from rich repositories and sources of cultural knowledge. We go on to describe how a rich ‘cultural layer’ might function within a 4-tier architecture model in order to be able to fulfil these ambitious aims. The particular 4-tier approach described, has the potential to deliver e-content that is matched to a set of client requirements and user cultural expectations, aligned to e-Enterprise and e-business contexts of use, as well being inherently capable of supporting a divers e set of heterogeneous client devices. Agent mediators form an integral part of the architectural model that is described below.

1 Introduction

Nowadays it has become accepted that contemporary Enterprise level applications are ‘n-tier’ in their architecture [1]. The rationale that is often put forward by its proponents being, that by sub-dividing a system into a series of logical ‘layers’, the system can be more effectively designed and maintained. This is achieved by breaking the system down into a number of horizontally related ‘chunks’ so as to make that system more manageable. In a 3 layered (or 3 ‘tier’) system for example, the focus is on 3 distinct horizontal abstractions. These are namely a) the back-end persistent store (the ‘database’), b) the middle-tier processing (or ‘business logic’), and c) the front end user interface, or ‘presentation’ layer. Each of these components is laid on top of one another accordingly (hence the term ‘layering’), with each layer/tier communicating with its adjacent layer, as seen in Figure 1 below:

Presentation

Business Logic

Database

Figure 1: A simple 3-tier architecture

Inter-tier communication is facilitated via a simple interface between each layer. This also enables the so-called ‘plug-and-play’ of layers, facilitating portability between data applications (say from Oracle to XML) or portability between a user’s client device (e.g. from PC browser to mobile phone device). The tiers represent a fundamental separation between content, made up of the business logic and the database, and its presentation. This approach enables enterprise systems to be developed that cohesively focus on what the semantics and business logic context of an item of information, regardless of how we may decide to present it to the user. The layers communicate via a simple (to maintain), well-defined interface.

To appreciate how we might make this separation, imagine how you would convey exactly the same piece of information to a blind user or a deaf user. Consider, for example, a favourite music track or artist, and how this would be conveyed to a deaf person. Take a favourite picture and likewise explain how this would be conveyed to a blind person. Quite a challenge! But it does clarify the thinking that lies behind the n-tier model (i.e. how the information content within systems might be usefully separated from its presentation). A simpler example is imagining a set of web pages, which because they have been designed to be viewed in an interactive web browser environment are hard to print out as one simple, coherent document. Clearly in separating presentation from content the issues that have to be considered may manifest themselves in many different dimensions. Nonetheless, this ‘natural separation’ must be achieved if systems are to be maintained efficiently within dynamic deployment contexts such as localised web-sites.

Contemporary systems however do not typically delve deeply into such issues. They merely examine how GUI (graphical user interface) objects (windows, dialog boxes, mouse/pen movements and so forth) present the content. Typically such presentation takes little account of the user’s cultural background, experience or disposition to the content. Put simply, there is only a superficial consideration, at best, of the semiotic (‘shared meanings’ and semantics ‘received meanings’) of the information being presented to the user. The presentation tier within the context of cross-cultural usability, should we propose, be ideally conceived as actually comprising two layers, the GUI layer and the semiotic (i.e. the cultural layer), as shown below in Figure 2.

GUI

Cultural

Business Logic

Database

Figure 2: An extended 4-tier architecture

We view the evolution of an n-tier model of the type depicted in Figure 2 above as being a prerequisite for the full evolution of the eEnterprise itself. More particularly those enterprises that essentially only exist via the Internet (such as www.egg.com) and are thus wholly reliant on computer mediated communication. We go on to outline a novel conceptual model of culture and later show how such a conceptual model may be mapped into an extended 4-tier architecture, of the type already introduced above, through the creation and instantiation of a cultural layer.

2 A Cultural Tier: some tentative conceptual foundations

One of us has previously proposed a generic conceptual model that may us help to visualise culture, as it is relates to the design and evaluation of the HCI component of a localised eEnterprise site. A series of interconnected ‘layers’ are seen to operate at different levels of abstraction [2], these being identified in Figure 3 below. We proceed to offer an explanatory commentary of the model and go on to suggest some ways in which the model can be evolved, instantiated and ultimately reside within a localised e-commerce site artefact in the form of a tangible cultural tier.

Figure 3: a conceptual model of culture

2.1 Commentary: how does the model relate to cultural research?

2.1.1 Hofstede’s work

Hofstede’s classic work on whole population cultural differences [3] has been mapped to specific countries and regions of the world using various empirical methods of verification. The value of such work is that it allows us to establish a frame of reference for a user, a web-site, a whole population or even ethnic sub-groups using a set of widely accepted meta-level cultural dimensions. It is possible to examine specific web-sites and to establish the holistic goodness-of-fit the site and these meta-level cultural dimensions through site analysis; an approach commercially exploited by Optimum-Web Ltd (www.optimum-web.co.uk) and similar specialist usability agencies. Comparative heuristic evaluation can enable a number of competitor sites within a vertical e-market segment to be examined using Hofstede’s taxonomy. The value of this kind of vertical market analysis, is that it is then often possible to relate Hofstede’s taxonomy to a particular kind of e-commerce shopping context (e-travel, e-finance, etc.) and hence to a common set of specific user goals. However exploited, Hofstede’s work represents a meta-level abstraction of the interface between culture, e-culture and site design.

2.1.2 Cultural attractors – what role do they have to play?

Any given site is likely to contain a number of culturally specific references depending on the specific purpose of the site. Previous studies have sought to classify and understand how users react differentially to such cultural references (whether explicit or implicit). Indeed the term ‘cultural attractor’ seems appropriate here, and typically refers to various visual objects embedded within a vast ‘soup’ of heterogeneous visual content, that is typically to be seen at the surface layer of a commercial web-site. Such objects typically include brand-signs, images, promotions, ‘pop-ups’, and similar features, that collectively creates a characteristic ‘look-and-feel’ of a particular site. Indeed, Fink [4] carried out one of the first empirical studies of cultural attractors in which user’s native to Australia and Thailand were seen to respond differentially to attractors according to their prior set of cultural expectations. One of us has since established that such attractors appear to reflect wider dimensions of shopping and consumer culture within Taiwan [5]. Cultural attractors are typically identified and verified through the creation of simplified prototypes, through live site audit by users or experts, as well as through wider studies of consumer behaviour and local shopping practices, including on-line branding.

2.1.3 Semiotic aspects of culture

Semiotic forms of textual analysis (in which a site is typically deconstructed into its component sign types) examines how such computer based signs comprise a kind of visual language, thereby conveying ‘shared meanings’ to users. Semiotic textual analysis is merely one way in which the notion of attractors can be examined analytically rather than empirically [6]. Clearly another approach is to carry out exhaustive empirical testing using live sites and their users, focussing on specific sign types, user groups and social semiotic contexts of use. Indeed, we can perhaps postulate that e-commerce sites can in fact act as windows within which observers can view dynamic cultural change by site audit using the semiotic paradigm.

3 Implementing a Cultural-tier: from conceptual design to a working system

Thus far our discussion has focussed on a purely conceptual model and its set of possible inputs. The discussion above has sought to establish merely that such a model does indeed appear to be potentially capable of being instantiated through a variety of analytic and empirical inputs. However, thus far our conceptual model, though useful as an analytic exercise in cultural visualisation has not as yet offered a coherent integration of these disparate sources of cultural knowledge. We propose therefore to evolve the model through implementation within our 4-tier architecture as a cultural tier. Such a system needs to be instantiated with some or all of the various inputs required (analytic, empirical, qualitative and quantitative) whilst at the same time, conform to wider principles of abstraction, polymorphism that are central to modern principles of software engineering, and more particularly reuse. A direct and open-source software artefact, once developed would offer academic researchers and practitioners alike a tangible tool within which site design and deployment is intimately connected to the theoretical and/or empirical approaches which have created it. Essentially, we propose the creation of a separate cultural tier within that explicitly seeks to express whatever model(s) of culture are being embedded within the site(s) being developed. Clearly this presents developers and researchers alike with a considerable challenge that will require equally challenging technical mediators to be become fully realised. Here are our initial suggestions, broken down into three sub-layers as shown below in Figure 4:

Figure 4: The Cultural tier, sub-layers and n-tier architecture context

4 The CULTURAL TIER : an explanatory commentary

4.1.1 Sub-layer-1: tailoring meta-level site stylistics including metaphor

Hofstede’s taxonomy needs to be supported through the development of a meta-level reference model (an extendible reusable and knowledge rich archive) that can potentially enable developers to select and dynamically tailor localised sites from templates containing appropriate metaphors, meta-level stylistics with known Hofstede cultural bindings. Goguen [7] has previously proposed that such generic meta-level HCI models can be built using novel algebraic semiotic methods and this approach may be the most promising one to adopt initially.

Suggested inputs: empirical results gathered using Hofstede’s taxonomy; library of metaphors, e-commerce shopping contexts, holistic site ratings etc.

4.1.2 Sub-layer-2: tailoring meta-level site stylistics including metaphor

It is of course necessary for the site designer to select the most appropriate mix of cultural attractors and related tangible visual objects as they appear to the user and as embedded at the surface level of the site. This selection should ideally be made through access to a reusable library of visual objects, containing known cultural bindings and notions of shared meanings, both in respect of individual signs as well as collections of signs. Clearly the object library will need to reference other layer(s) within our n-tier model explicitly (so as to match surface level object behaviours with more general system level object behaviours) but also critically, and in parallel, contain a set of rich meta-level semantics. One suggested approach might be for us to adopt and align our layer with one of the many semantic web meta-data initiatives such as XML RDF (Resource Description Format) or alternatively develop an XML derivative of our own. In any event, reusability and late dynamic run-time bindings are essential features of this layer. Navigation aspects (i.e. controlling how user goals are initiated and triggered by surface level object event handlers within the state space model) will need somehow to be incorporated too. One possibly here might be to employ a lightweight state space notation of the type proposed in the SMDF (Shared Meanings Design Framework), [8].

Suggested inputs: empirical results gathered, expert site audit(s) etc. that identify specific attractors and distractors with their known cultural bindings within key e-markets and platform contexts. Semiotic analyses of exemplar sites, taxonomies of signs.

4.1.3 Sub-layer-3: Cultural requirements visualisation

Previous layers enable designers to select material from libraries and to assemble this material in optimal way(s) so as to meet the known cultural expectations of users operating within particular e-shopping contexts. Our hope is that these layers will form a permanent part of a ‘real’ site(s). However, there is also a need to create a reference model that clearly articulates the client’s cultural requirements so designers can match these requirements with user expectations and trace how (or if!) these requirements are actually expressed through sub-layers one and two. There is also clearly a need to aggregate and document the empirical user based inputs to the layers as previously described. We propose therefore, that a cultural requirements and visualisation layer be built, so as to facilitate the separation of cultural requirements elicitation from generic requirements elicitation (such as security aspects, platform aspects and detailed site functionality). Candidate delivery methods include: the use of card sorts as early and holistic elicitation tools to uncover tacit knowledge [9], extensions to UML use-case techniques, as well as the use of rich visualisation techniques such as conceptual graphs. Indeed, we have elsewhere proposed a generic SMR (Shared Meanings Requirements) engine to carry out such tasks in a related deployment context [10].

Suggested inputs: client requirements, empirical sources, user stereotypes etc.

5 Conclusion

We have highlighted that in the valuable pursuit of n-tier enterprise systems, the presentation of information content is a much more involved affair than simply bolting on GUI objects directly onto content without considering the cultural disposition of users. Without this the true eEnterprise cannot be achieved and whilst this task is a challenging one, it also presents us with an ideal and tangible opportunity to enrich localised web-sites with cultural semantics. Semiotic approaches offer the developer who seeks this kind of enrichment to integrate these disparate inputs (as derived from cultural research and industrial best practice) within an integrative paradigm. Therein lies the challenge ahead, not only to build and deploy such a system, but also critically, to provide ways to integrate heterogeneous cultural knowledge inputs so as to better understand the elusive and challenging notion of xCulture for the eEnterprise. The role of agents is also an area we have yet to fully define or explore within the confines of the present paper. However we feel instinctively that semiotic agents of the type already proposed by Joslyn [11] might well play a critical role in facilitating the actual deployment of such a system.

References

[1] Fowler, M., (2003) “Patterns of Enterprise Application Architecture”, Addison Wesley Press.

[2] French, T., Minocha, S., and Smith, A., (2002) “eFinance Localisation: An informal analysis of specific eCulture attractors in selected Indian and Taiwanese sites”, In: “Designing for Global Markets 4” Eds. Corondao, J., Day, D., Hall, B., Procs. 4th International Workshop on Internationalization of Products and Systems., p. 9-21. ISBN 0-9722184-0-8.

[3] Hofstede, G., (1991) “Cultures and Organizations: Software of the Mind”. New York, NYM McGraw-Hill.

[4] Fink, D. and Laupase, R., (1999) “Perceptions of Web Site Attractors and their Effectiveness: An East/West Comparison”, In: Procs. 7^th European Conference on Information Systems, p. 157-170, Copenhagen Business School, Copenhagen.

[5] French, T., (2002) “eCulture texts and urban cultural change in Taiwan: a view through the semiotic lens”, Imperium On-line Journal of Post-Colonial Studies. Available on-line from: http://www.imperiumjournal.com/0pages/30001.html

[6] French, T. (2002) “What kinds of ‘interpretation' can semiotics offer to e-commerce site users and designers?”. Procs. 1st International Workshop Interpretative approaches to Information Systems Research, July 26-27^th, Brunel University, UK, p. 1-6. ISBN 1-902316-27-4.

[7] Goguen “A short overview of algebraic semiotics”, available on-line from:

http://www.cs.ucsd.edu/users/goguen/projs/semio.html

[8] French, T., Polovina, S. and Vile, A. (1999) “Semiotics for E-commerce: Shared Meanings and Generative Futures”, Proceedings of 10^th BIT Conference 1999, ISDN 0-905304-30.

[9] French, T., Springett, M. (2003) ”Developing novel iTV applications: a user centric view”, Procs. European Conference on iTV, Brighton University, April 2-4th.

[10] French, T., Polovina, S., Vile, A, Park, J “Shared Meanings Requirements Elicitation (SMRE): towards Intelligent, Semiotic, Evolving Architectures for stakeholder e-mediated communication”, Procs. ICCS Conference 2003 (International Conference on Conceptual Structures), Dresden, Germany, in press, July 2003.

[11] Joslyn, C., and Rocha, L., (1999) “Decision Structures of Socio-Technical Organizations”. A Position Paper Background and Approach: Semiotic Modelling of Socio-Technical Organizations available on-line from: http://www.c3.lanl.gov/~joslyn/nmsu/position.html

________________________________