Abstract Innovation in Business Processes An experiment using CAIN

Luis Borges Gouveia

Feliz Ribeiro Gouveia

{lmbg, fribeiro, dlamas}@ufp.ptMultimedia Study and Research Center

University Fernando Pessoa

Pr 9 de Abril, 349P-4200 Porto Portugal

David Ribeiro Lamas

Abstract

Business to consumer electronic commerce will need better tools, beyond browsers andgadgets, for the user to take full advantage of the opportunities of the electronicmarketplace. On the other side, business needs to rethink their positioning, and to dealwith this new customer using new methods and tools. We present in this paper a tool fornavigating large information spaces, such as the Web, using what is called here “roughguides”. The tool uses user preferences to perform adaptive guidance. Rough guidesbring new interactions into the electronic commerce arena: how they are produced, sold,and updated using consumer’s feedback can be done in several and different ways thanin a classic economic framework.

1. Introduction

The way people buy and sell is certainly changing with the advent of electronicmarketplaces. Current information technology already has a strong influence both onbefore-sale and after-sale techniques and systems. These systems take the form of helpdesks, call centres, logistics and distribution systems, and strong enterprise alliances tooutsource important client functions, from development to marketing. With someexceptions, namelly the UPS Tracking System (http://www.ups.com), none of them wasdeveloped with the internet in mind.

These business support systems can be seen as a visible sign of the more than probablemove to full electronic commerce. In effect, protocols like Electronic Data Interchange(EDI) or systems supporting government to citizen relationships (like the PortugueseInfocid platform) are already in place and have been used for several years. Financialinstitutions and banks have used for several years electronic transactions. This gives atrue field to implement electronic commerce both between businesses and betweengovernment and citizens.

Business to business electronic commerce is already on the move and several authorsgive figures that show its growing importance in the context of business to businessrelationships and in the global context of electronic commerce where it represents in anear future 80% of the value of transactions.

In this paper, we will concentrate in the business to consumer electronic commerce,where there is still much to be developed and where a lot of research about use anddevelopment of new methods, tools and processes is still being devised. It seems thattwo key factors are essential to leverage business to consumer electronic commerce:

tools to allow it, and consumer trust. Current tools, such as search engines, informationwarehouses, electronic mail and bulletin boards, do not scale well in a global market,and do not allow for shopping guidance. The user ends giving up either because of largewaiting times, lots of information, lost pointers, or lack of confidence.

2. The proposal

Our proposal is based in the following:

• simple, anonymous, secure, and reliable payment schemes will find the mainstream,and will eventually be adopted. Electronic wallets and micropayments are the mostpromising (Choi et al. 1997);• an information-based economy will enable consumers to be also producers(prosumers), and trading of information will take of. Information about what is onthe Web, where it is, and how much it is worth, can be a catalyst for micro-payments to take place between prosumers;• automated services, such as intelligent agents, webcrawlers, and search engines, willbe increasingly disapointing as 1) firewalls and other security features prevent themfrom entering web sites, 2) unstructured information is and will be difficult to filterand process automatically, and 3) more and more information is stored in serverdatabases and cannot be found in html pages.There are already some partial solutions to the problems outlined above: strongcryptography allows for digital cash to be a reality; internet communities exist linkedeither by interest or by other social factors and internet auctions are a success; anddirectory services, such as Yahoo, although facing the overwhelming task of manuallyindexing up to 300 million web pages, are the preferred starting point for navigation asthey are more reliable (Chakrabarti, 1999).

In our proposal, users have at their disposal available information about several themes,consisting of a compilation of internet addresses. These compilations are called “roughguides”, and they are much pretty the same as their city guides or tourist guides analogs.Rough guides can be used in a standard web browser, in a standard way, as describedlatter. Depending on the profile of the user, rough guides show different information tothe user, allowing thus to adpative navigation and guidance. The implications we see arethe following:

• rough guides can be produced and sold about a great variety of themes. The authorscan be individuals or companies. Good and bad guides will exist, but this is a marketproblem, and the market will solve it. Guides can be produced on specific themes —for example, “the 10 best places for information on History”, “buying technicalbooks”, or “best design pages” —, or very general ones, such as “complete listing ofgovernment services on the internet”;• rough guides should allow for easy updating, as internet pointers come and go. Theyshould also allow for comments from different users to be included, opening the

way to co-operative navigation. As a more advanced feature, the adaptive algorithmcould learn from different users, and modify its navigation strategy accordingly;• micropayments should help foster a community of rough guide users, producers, andreviewers. They would be willing to pay (be payed) a small fee for a serviceconsidered worthwhile. In the material, classical economy, who has never bought a“Michelin Guide”?In a new economy, with new rules and habits, there will be certainly a place and a neeedfor digital rough guides. Before describing our system, we will present some of thecurrent thoughts on how the new economy will be, and how business and consumerswill fit.

2.1. A digital economy

Don Tapscott (1996) proposed 12 themes for the New Economy — a digital economy.They are:

Theme 1 : KnowledgeTheme 2: DigitisationTheme 3: VirtualisationTheme 4: Molecularisation

Theme 5: Integration/InternetworkingTheme 6: DisintermediationTheme 7: ConvergenceTheme 8: InnovationTheme 9: ProsumptionTheme 10: ImmediacyTheme 11: GlobalisationTheme 12: Discordance

Some of these themes can be of interest in our discussion of business to consumerelectronic commerce. In particular knowledge and virtualisation seem to be key factorsto make consumers move from more traditional ways of buying what they need.A discussion of consumer needs seems to be central in establishing a valid scenariowhere a consumer can use electronic commerce not because commercial systemsimpose it, or were transformed to apply more efficient and technological approaches totheir processes, but because consumer needs are essentially different. The differencesresult from new habits, and can be seen as ongoing transformations reported by several

authors. These transformations are also the joint effect of the listed Tapscott themes forthe digital economy.

2.2. Two issues to assist the business to consumer relationship

From the knowledge and virtualisation themes, some lessons can already be learnedfrom current literature. It seems that without truly knowledge of processes, consumerneeds and consumer profiles, developing successful business to consumer systems willbe difficult.

Current research shows that:

• although initial alerts, consumers prefer brands and value them more than the priceof a product. The image of a business is an important asset, as consumers trust somemore than others, specially new entrants or unknowns (Gallaugher 1999);• new brands have to fight to enter the market, where the price is not the most relevantissue. Although it is easier to reach consumers, gaining their confidence is harder(Elofson & Robinson 1998);• there seems to be a trend to concentrate services around a main brand and/or place,most of the times calling it a portal. Again, brands are important, corporate image isimportant, and a solid reputation also;• security and trust is seen as a key factor for the consumer, giving place to someregulatory issues, such as on-line consumer protection, confidentiality, security, andprivacy, the later being one of the most important (Bakos 1996).These facts introduce the first issue: consumers still need to feel confidence and someevidence as their knowledge of the electronic market place must exist to allow them tohave an emotional relation with it (Bhimani 1996).

As what concerns the virtualisation theme, we can state the importance that the content,as collections of information items, can have to the consumer (and to the legislatoralso). Although important, content itself will not have great value for the consumerunless it provides him with a valid structure. This being so, structure is more importantthan content, in the sense that it allows the consumer to use the content. Even accesswill be more important than content as it drives the cost and the difficulty level of using(and buying) the content.

When we propose a sequence for structure, access and content, we also propose whatcan be sold as a product at different levels. An important issue in electronic commerceis what to sell and how to do it. In effect both the product and the process are on changein the electronic commerce. Some examples follow, as they show how this questions arestill open:

• some internet service providers and telecom operators are giving way computers tosell communication time;

• some added-value enterprises are giving way free access to know more about userpatterns and profiles;• some internet sites are giving way valuable information in order to get consumerattention and availability, such as reports and intelligence.These examples have also some interesting information. Its seems that we can have forfree from some, what we have to buy from others. This will lead us to the second issue;the one of the potential, the most effective asset in the current environment of businessto consumer electronic commerce.2.3. Human factors importance

Information technology now is different from the one devised several years ago. Currentproducts tend to have better human interface solutions, and thus have lead to new fieldsand problems. With the research related with Internet, and more closely with the WorldWide Web, this trend has increased.

Past experience in introducing new systems dealing with people and technology showthat technology in itself can not be seen as a generic problem solver. Two examples arevideoconferencing where besides the promise of substituting face to face meeting, it didnot happen, and expert systems, where most of their implementations were discontinuedbecause they failed to acquire, store and maintain knowledge.

These two examples are just a small part of technology systems that although theyrepresent well designed solutions to a given problem have not considered the humanfactors around.

In the business to consumer electronic commerce the same reasoning applies. This willlead the way to consider the two sides of the same equation. We need to consider theuser (client) and the process (enterprise). Successful electronic commerce technology toserve consumers must combine user needs with enterprise processes.

We will next discuss some of the technical characteristics of our proposal, and showssome results of experiments in an educational setting.

3. The web as the marketplace

From the user side, two key factors apply: the trust in the given information and the costto access and understand it. A large volume of information, such as the one availableover the Web has an inherent application potential in an educational setting.Unfortunately, besides inheriting and enhancing most of hypertext's qualities, the Webalso incorporates and stresses common hypertext problems (Brown 1990; Shum 1996;Shum & McKnight 1997).

It is the Web's disorganized nature that mainly prevents its most effective use as apedagogical tool and, as such, research is needed to develop the means that will help totransform the available data into meaningful information able to support learning orinitial research activities (Eklund & Ziegler 1996; Marshall 1995).

The Computer Aided Information Navigation project (CAIN) attempts to provideadaptive navigation support as a way of increasing the Web's value as a pedagogicaltool (Lamas 1998). CAIN's approach is a reactive one as the effort goes into providing away to deal with the vast amount of useful information available on the Web as is,rather than to try to improve or alter the Web's infrastructure in any way.

In order to help users reach their goals, CAIN provides direct guidance navigationsupport as a form of non-obtrusive weak hypertext linearization enabling the user tofollow a context specific ranked sequence of selected Web pages without ever needingto perform any search or follow any link if they do not wish to. This approach does notintend to prevent goal-oriented exploration but to provide a sound thread or guideline tohelp users to retain their focus in the way a Rough Guide does.

The system's representation of the Web, crucial to the success of its adaptive features,builds on the Dublin Core (Weibel & Lagoze 1997) and on the Resource DescriptionFramework (Lassila & Swick 1998).

The users' representation combines three user modeling techniques resulting in a hybridsolution that uses stereotypes, overlays and attribute-value pairs (Benyon & Murray1993; Brusilovsky 1996).

Navigation support is accomplished by a basic route-finding heuristic which selectscontext specific Web model items, sorts them using associated qualitative ratings andpresents them to the user, one at the time, based on the attributes of the user's model(Lamas 1998).

The next pages present and discuss the results of an experiment undertaken to test if thedirected World Wide Web navigation guidance provided by the CAIN improves theperformance of Web users under appropriate conditions. Section 4 describes theempirical study and discusses the experiment’s results. Finally, section 5 highlightssome related empirical work and section 6 discusses the validity and completeness ofthe experiment and provides future work guidelines.

4. Empirical study

It is hypothesised that contextualized direct guidance over the information available onthe World Wide Web improves comprehension under the specific condition that:• users are either inexperienced in Web use or have little expertise in the target subjecttopic or both;• users use the system pursuing learning, discovering, or initial research goals.For this purpose, a standard task was devised in which volunteers were required to usethe WWW to learn as much as they could about two given subject topics —cryptography and poetry — using either CAIN or alternative tools the users chose touse.

The experiment compared the use of CAIN with conventional tools on three measuresof performance:

• Comprehension as measured by a multiple-choice questionnaire;• Time taken to complete the tasks; and

• System usage appreciation assessed by a like/dislike rating at the end of each task.Performance was examined as function of:• Tool used, either CAIN or Free1;

• Web expertise, the Web experience of the users, either Low or High; and

• Subject expertise, the users' expertise in the target subjects topic, either Low orHigh.In order to implement the experiment, volunteers were recruited from the university'spopulation, specialist help was required to build the domain model and to provide thesubject abstracts and questionnaires and a specially designed environment was preparedto implement the experiment.

The next figure illustrates the fact that CAIN produces faster performance than Freeunder all combinations of levels of the independent variables except when the volunteerhas High Web expertise and High subject expertise, i.e. it suggests that:

• time taken may be slower with CAIN when the user is a highly experienced Webuser and a an expert in the learning assignment field, but;• CAIN is faster when users are inexperienced Web users and inexperienced in thestudy subject or inexperienced in the Web and experienced in the subject orexperienced in the Web and inexperienced in the subject.Note that, as shown in the table and figure, Web expertise alone does not slowperformance when using CAIN. In fact, highly experienced Web users performances areslightly faster under CAIN than Free. It is the particular combination of High Webexpertise and High subject expertise that slows performance using CAIN compared tousing other tools on the World Wide Web.

Web expertise

Low

High

Mean

ToolFreeCAINFreeCAIN

Subject expertiseLowHigh91.7086.5086.1768.5074.6761.5050.5075.4475.7672.99

Mean

.1077.3468.0962.97

Table 1: Web expertise vs Tool vs Subject expertise performance times contrast

1

Where Free stands for any other tool but CAIN.

Free CAIN Low, Low Low, High High, Low 100 90 80 70 60 50 40 30 20 10 0 High, High

30% when supporting learning or initial research tasks performed over the World WideWeb.

This is probably due to the way CAIN guides the user through a sequence of previouslyselected and relevant Web resources.

Though not in a very significant way, the experiment's results also suggests that the timetaken to accomplish such tasks can also be reduced when using CAIN except whenthere is no point in using it i.e. when the user is already an expert using the World WideWeb and an expert on the task's subject.

As far as enjoying the system, the results do not show any significant differencebetween using CAIN or any other approach to navigate the World Wide Web. In fact,the results show that users in general enjoy using the Web with whatever tools they use.Although encouraging, this experiment's results are not sufficient to understand all theimplications of a system like CAIN and further studies must be undertaken. In fact, thecurrent investigation only measures the adaptive guidance support leaving untouchedissues related to all the other relevant areas brought together in this work such as:• the adequacy of the metadata records and associated rating strategy;• the soundness of the implicit co-operation model;• the reliability of the relevance feedback mechanism; and• the validity of adaptivity itself.

Specifically, it fails to understand other success factors such as:

• how relevant is the order in which Web resources are presented to the user?• how do users cope with the domain building process?• are the rating criteria adequate and easy to apply?• is such Web representation successful for all its users?• how does the system cope with long term users?

• how does physical distance affect the system's collaborative nature?

This is an under-investigated area and there is clearly a need for continued studies onthe value of adaptivity in hypertext based learning environments.

In conclusion, the Computer Aided Information Navigation seams to be in the rightdirection and research will continue in order to achieve this collaborative environment'slong term goals: To guide the user while learning, memorising and forgetting things inorder to increase motivation, attention and empathy, reducing the environment's noiseand improving the knowledge acquisition process. Whether it will help users in an

electronic commerce environment it remains to be tested, but we believe, from theresults we present, that CAIN can be a valuable help.

References

Bakos, Y. “The Emerging role of Electronic Marketplaces on the internet”,Communications of the ACM, 41, 8, (1996): pp. 35-42.

Benyon, D. & Murray, D. “Adaptive systems: from intelligent tutoring to autonomousagents”, Knowledge-Based Systems, 6, 4, (1993): pp. 197-219.

Bhimani, A. “Securing the Commercial Internet”, Communications of the ACM, 39, 6,(1996): pp. 29-35.

Brown, P. “Assessing the quality of hypertext documents”, Proceedings of theECHT’90, UK, 1990.

Brusilovsky, P. “Methods and techniques of adaptive hypermedia”, User Modeling andUser-Adapted Interaction, 6, (1996): pp. 87-129.

Chakrabarti, S. “Mining the Web’s link structure”, Computer, 32, 8, (1999); pp. 60-67.Choi, S.-Y., Stahl, D., & Whinston, A. “Electronic payments and the future of electroniccommerce”, The Center for Research in Electronic Commerce, University of Texas atAustin, 1997.

Eklund, J. & Brusilovsky, P. “The value of adaptivity in hypermedia learningenvironments: a short review of empirical evidence”, On-line proceedings of the SecondWorkshop on Adaptive Hypertext and Hypermedia, [Online], available athttp://wwwis.win.tue.nl/ah98/Eklund.html, Agosto 29, 1998.

Eklund, J. & Ziegler, R. “Navigating the Web: possibilities and practicalities foradaptive navigational support”, AusWeb 96: Second Australian World Wide WebConference Proceedings, Southern Cross University, Queensland, Australia, 1996.Elofson, G. & Robinson, W. “Creating a custom mass-product on the internet”,Communications of the ACM, 41, 3, (1998): pp. 56-62.

Gallaugher, J. “Challenging the new conventional wisdom of net commerce strategies”,Communications of the ACM, 42, 7, (1999): pp. 27-29.

Lamas, D. “Adaptivity on the Internet”, University of Portsmouth research seminars,University of Portsmouth, UK, 1998.

Lassila, O. & Swick, R. “Resource description framework RDF model and syntax”,[Online], available at http://www.w3.org/Tr/WD-rdf-syntax/, January, 1998.

Marshall, A. “Developing hypertext courseware on the World Wide Web”, Educationalmultimedia and hypermedia: proceedings of Ed-Media 95, Association for theAdvancement of Computing in Education, 1995.

Shum, S. “The missing link: hypermedia usability research & the Web”, ACM SIGCHIBulletin, 28, 4, (1996): pp. 221-248.

Shum, S. & McKnight, C. “World Wide Web usability: introduction to this specialissue”, International Journal of Human-Computer Studies, 47, 2, (1997).

Tapscott, D. The Digital Economy. Promise and Peril in the age of networkedintelligence, McGraw-Hill, 1996.

Weibel, S. & Lagoze, C. “An element set to support resource discovery: the state of theDublin Core: January 1997”, International Journal on Digital Libraries, 1.2, (1997): pp176-186.