Topic Maps and the Business of Knowledge

1. Introduction

The Web and other information pools provide pentabytes of information. But today most users do not need more information - if anything, they need less, because they are already drowning in enormous quantities of it. Fulltext search and thousand of query hits are no solution. A metadata search (e.g. using Dublin Core) or a structured fulltext search in XML reduces the number of hits and makes our queries more precise, but the query result is a single information object separated from its semantic context. This is of little value for the users who are looking for intelligent navigation in the knowledge pool exploring a topic of interest and finding their way to relevant information as quickly as possible and to be able to filter out the 'noise' created by all the information for which they have no use.

Topic maps are defined as an ISO standard ISO/IEC 13250 (SGML/HyTime) and XTM (XML/XLink) designed for structuring large information pools. Topic maps are the GPS of the information universe. They offer dynamic access to the information pool. Classified navigation paths provide an easy and intuitive way to explore the pool. Topic maps bring related information together as associations - the way humans think. Querying a topic map does not lead to raw information objects, but to a presentation of the 'knowledge' about the requested topic. Thus topic maps are the solution for the mass problems of the information age. They offer an extreme power and flexibility on the one hand, and sufficiently well-defined semantics on the other.

New technologies bring us new business. Topic maps generate new business models. This paper presents the business models which cover the application side of topic maps, not the software and consulting side. Application areas of interest are: Web portals, publishing (commercial, corporate), enterprise knowledge management, eCommerce.

2. Web Portals

A Web portal collects information assets about a certain domain and connects these assets by hyperlinks. The main goal of the portal is providing the user/customer on the one hand with a complete set of information and with easy and efficient search and navigation functions on the other hand. This goal is exactly what topic maps fulfil. Topic maps will built intelligent site maps, keep the concept level (= topics and their associations) and the resource level (= Web pages) separate, and they provide an advanced search in the concept instead of a 'stupid' fulltext search in the pages.

3. Publishing Knowledge Networks

Traditional publishers (both commercial and corporate) gather, verify, assemble, and distribute information objects as publications like books, journals, loose-leaf, manuals, or service bulletins. They store and manage their information assets product-centric (= publication).

New concepts have to be information-centric. Publications refer into the information pool, reuse the same assets, which are rendered media specific. The optimization of the online publication does not only concern the layout, it also concerns the interactive access (navigation paths, queries, hyperlinks).

The interactive access is the domain of topic maps. A topic map can be seen as a transparent slide above the information objects carrying the structured and typed link network. Different slides support different views on the same data. The usage of independent link addressing (XLink/XPointer, HyTime) allows the connection of the topic map with the assets, even if the assets are read-only, are owned by someone else, or are used in various topic maps. This feature opens up the possibility of separating design, creation, and maintenance of information assets and the added values - the topic maps.

Commercial publishing and technical documentation benefit from a topic maps based data organization and navigation. But also health care data, insurance documents, or banking information will benefit as well. Topic maps can fuel the eBusiness of these industries when textual information and not numbers are in the focus of interest.

4. Enterprise Knowledge Management

The buzzword 'knowledge management' stands for gathering, storage, and retrieval of corporate knowledge. The goal is to make the knowledge which is 'in the heads' of the employees accessible for the whole company. This knowledge is about products, persons, processes, facts, documentation, research results, known problems, planned developments, tips & tricks, etc. It is about nearly everything which is of value to run a company successfully.

Topic maps are seen as a standardized base technology for knowledge representation. Therefore they could be used for the storage and retrieval of the gathered knowledge. However, it is the knowledge gathering process which is mission critical. The amount of information produced by a company or larger enterprise could be enormous. It will be more or less impossible to investigate and model all the enclosed knowledge. Thus, we need support by automatic processes. Thesaurus based fulltext indexing are an alternative and lead to useful results.

5. eCommerce

B2C eCommerce Web sites are mostly online shops or auctions selling products over the Internet. Their competitive factors compared to real shops are opening hours (24x7), shopping without leaving home, larger list of products to select from, and sometimes personalized offers controlled by customer profiles. But customers of online shops often feel like the 'lonesome cowboy' lost in the rich selection of electronic offers. If you know what you are looking for you will find it; if you do not know it, your searching will probably fail - and you will leave the shop unsatisfied and disappointed. This is true for all kinds of goods - even for products where the selection process does not need a lot of explanation and help, like books, music, videos, software. But dot.com companies also want to offer and sell products which need help from an assistant - like clothes, used cars, bank services, insurance contracts. These are the eCommerce Web sites which have to provide even more sales support. Knowledge technologies and especially topic maps can provide intelligent assistance like a human sales assistant - or even better.

Quick access of the wanted good is the key issue. Intelligent navigation is a key feature of topic maps, because it is about traversing explicitly coded knowledge structures. This could be a classification scheme or thesaurus, but it could also be a much richer representation of the knowledge about the products and the application area - the thesaurus is just a specific part of such a rich structure. Topic maps are a good choice to represent the knowledge.

Navigation helps the user to get to the product of interest by following links which represent related/associated topics. The topics model/represent the ontology - the kinds of things of the application area - and the real things/concepts of the application area. The user might start with a query getting to some topics of interest; from these topics he starts navigation the knowledge structure. The other way around - starting with navigation and continuing with a query - might also lead to the product of interest.

Example for navigation: The home page of an online Jeans House organizes the departments of the store into the categories jeans, shirts, sweats/pullovers, and accessories. Every category is represented by its own page offering further sub-categories (e.g., category jeans could be divided into original blue jeans, stone-washed jeans, colored jeans). If the customer reaches the category she is interested in - e.g. original blue jeans - she starts the query 'Levis in 30/32 (female) with straight cut'.

6. Ontology and Map Business

Topic maps generate new business opportunities independent of the application area. The definition of application ontologies or the creation of topic maps without occurrences could become business models in the near future. Or companies using proprietary formats to store their ontologies could use topic maps as vendor neutral and future save format.

6.1. Design and Reuse of Topic Map Templates

A topic map template contains the types and roles of the application domain - the ontology. Additional constraining conditions ensure the consistency of the maps created using the template. A topic map template declares like an SGML/XML DTD the structure and semantic of a class of maps. Examples of classes are: encyclopaedia, dictionary, travel guide, professional publication e.g. legal text with commentary, yellow pages, software documentation, technical manual of specific industry. A designed template can be used in various topic maps of the given class, even if the maps will be very different.

6.2. Design and Reuse of Topic Maps

A well-designed topic map is like a well-designed back-of-the-book index. A publisher can use both for different publications.

The topic map declares all topics, their names, associations between topics, and the occurrence links into the information pool. The map represents the 'knowledge' about a theme independent if the links to the information assets are established or not. The topic map is useful and valuable, whether or not there are information assets giving further information to any topic. As already stated, the same topic map can be overlaid on different information pools, just as different topic maps can be overlaid on the same information pool providing different views to different user groups. Furthermore, this separation provides the potential to be able to interchange topic maps among publishers and to merge one or more topic maps.

The design of topic maps becomes a new job for the 'Information Broker'. The information broker uses existing information assets provided by other companies/organizations ('Information Provider') and enriches them by adding metastructures - like topic maps. The broker has the knowledge about an application field. He is aware of the user needs regarding the context in which a single information makes - more - sense and regarding navigational paths. The information broker can react very quickly on changing demands of the market. A change in the topic map can be performed very easily compared to the effort for a change in the information pool.

There are two general business models for the information broker:

• Designing a topic map without occurrence links and selling it to publishers.

• Designing a topic map, contracting access to information assets owned by others, establishing the occurrence links, and selling the rendition of the map to the end users. XML, XPointer, XLink support this business model perfectly as a Web application.

These jobs will be performed by the publishers or by other companies.

7. Conclusions

The topic maps standard provides such a general model that it is applicable by a wide range of industries. The organization and navigation of continuously growing information pools is the main focus of topic maps and therefore all industries suffering with this problem are ideal candidates.

Topic maps used by commercial publishing and corporate publishing provide easy navigation in the information pools. The editorial groups could make statistical use of the topic map to figure out deficiencies in the covered (or not covered) subjects which is e.g. important for reference works publishers.

Web portals benefit from topic maps when using the possibility to divide concept level (= topics) from the resource level (= URLs) and to provide precise query results.

Enterprise knowledge management makes use of associations representing the relationships between concepts (= topics). Topic maps are used as base technology for knowledge representation. But besides the technology the organization of knowledge gathering is even more critical. AI based search techniques would help.

The design and development of topic map ontologies and topic maps open also a new business opportunity. This is possible because topic maps can be exchanged and merged with other maps and/or other information resources.

Design and Development by deepX Ltd.