Baltic Business and Socio-Economic Development 2008 Bridging the Gap between Wiki and Knowledge Networks Uwe Lämmel1 Abstract Knowledge Management is aimed at providing the desired knowledge whenever and wherever it is needed. In an organisation the complete knowledge of the organisation should be managed in such a way that it is available for everybody who needs it in order to accomplish his duties. Knowledge networks have promised to solve the problem by bringing together two concepts: Semantic networks and a graphical visualisation using topic maps. Although these networks work quite well once they have reached a certain level the amount of work for the development cannot be underestimated. Wiki systems promise an easy handling and can be well applied for knowledge management. We argue that a combination of knowledge networks and an easy-to-use Wiki will seriously improve knowledge management. Various applications are possible: in a company, in a university, in a region or across borders. Keywords: knowledge management, knowledge processing, knowledge networks, topic maps, semantic networks, wiki, semantic wiki 1. Introduction Knowledge is explicit like the one stored in databases or in documents either in the computer or in filling cabinets. The personal knowledge of all employees is part of the organisational knowledge as well. How can all the knowledge be managed in order to make the work more efficient? That is, we should get the knowledge whenever we need it. Search for information has been highly improved by google. Nevertheless there is still a serious drawback because it is solely a syntactical search. Only information will be found containing a certain keyword. Important information strongly related but missing the keyword will be left out. Beside google another concept has changed the search for information rapidly: wikipedia. The success of wikipedia or the wiki-concept at all is based on its open access. Everybody can contribute to the system and can improve the system. Semantic wiki tries to combine both approaches mentioned above. wiki pages are enriched by semantic information e.g. information about the relationships between pages or concepts. Thus, links can be defined that create a semantic network. A graphical representation of the network as a topic map can then be deduced automatically. Moreover annotations may define logical dependencies and an inference machine may be used to make implicit knowledge explicit. Since wikipedia is well known it can be expected that users are well motivated to contribute to such a system. There are plenty of possibilities for semantic wikis. Knowledge management of an organisation can be raised up onto a higher level. Therefore we propose a project on the development of semantic wiki systems within the Baltic Business Development Network. For instance the knowledge of a university expressed in a semantic wiki could lead to new co-operations: Co-operations between university and companies or between universities or regions. Regional development will profit a lot if competencies and needs can be managed together in one system. 1 Prof. Dr.-Ing., Hochschule Wismar, Wismar Business School, http://www.wi.hs-wismar.de/~laemmel in: Gunnar Prause, Tatjana Muravska (eds.), Baltic Business and Socio-Economic Development 2008, Riga, Estland, 29.09.-02.10.2008, Berliner Wissenschaftsverlag, 2009 Baltic Business and Socio-Economic Development 2008 2. Knowledge Management There are various definitions for the notion “Knowledge Management”. A standard one was given by Davenport [Davenport98]: ”Knowledge Management is a formal, structured initiative to improve the creation, distribution, or use of knowledge in an organization. Knowledge Management is a formal process of turning corporate knowledge into corporate value“ A more general definition is given in the online encyclopaedia of the PC Magazine2: “An umbrella term for making more efficient use of the human knowledge that exists within an organization. Knowledge management is the 21st century equivalent of information management.” A more simple statement has become a headline for knowledge management since the mid 90s: If X only knew what X knows! A search in the internet shows that X can be replaced by almost any organisation, company or even person: HP, Siemens, Texas Instruments, Sony, your company, we, you, I, and everyone. If we take the aim of the conference into consideration X should be replaced by “our Baltic region”: If our Baltic Region only knew what our Baltic region knows. And we go ahead and claim: If our Baltic region only knew what all Baltic regions know! This motto gives the motivation for a knowledge management in our organisations and regions around the Baltic Sea. It marks the goal of the activities as well: We all know what we all know. Why do we not know what we know? Knowledge of an organisation consists of all the knowledge stored in their files either in the bookshelves or in the computers. This is explicit knowledge. All the employees have a lot of knowledge related to their work. This is so called implicit or tacit knowledge. In order to fulfil our duties effectively and efficiently we have to consider all the knowledge within our organisation, the explicit one as well as the tacit one. To ask someone is the fastest way to gather the required knowledge, if we know who knows. Second best: we look into some documents either in the files or on the computer. As Albert Einstein referred to the problem “To know means to know where it is written”. If there is an appropriate knowledge management available all the knowledge is represented and stored in a structured way where the structure reflects the relationships among the pieces of knowledge or the knowledge assets. Unfortunately in most cases we fall back to search engines like google: Looking only for the occurrences of certain words in some kind of documents like web-pages, pdf or other text documents. Although such a search shows good results in many cases, it is sometimes not very efficient. No answers or too many results share the same effect: we have not received what we were looking for. Sometimes the knowledge we want to get is hidden because the search words are not in the documents at all. This can be true even if the search words are very close related to the topic. An illustrative example has been provided by the German company intelligent views3: Assume that we want to know something about women in leading management position in German companies. A search for “woman leading 2 PC Magazine: http://www.pcmag.com/encyclopedia_term/0,2542,t=knowledge+management&i=45843,00.asp, last visited 2008-07-07 3 intelligent views GmbH: www.i-views.de Baltic Business and Socio-Economic Development 2008 position” or “woman director” will not result in the desired information. Documents will rather contain words like: “Mrs. Anne Musterfrau, executive”. There is a need for a more powerful search which does not look for the words only but takes the relationships among words and concepts into consideration as well: a so called semantic search. Semantic search engines have been provided as part of knowledge network systems. 3. Knowledge Networks and Topic Maps Knowledge networks have been proposed for many years as an IT tool for knowledge management. In a knowledge network knowledge is stored as concepts, attributes, and relationships (links). The basic concept is rather old. It dates back to the 60s. At that time semantic networks were introduced for knowledge representation in the field of artificial intelligence. From a mathematical point of view a semantic network is a directed graph. Subjects, objects are represented in nodes. A name is associated to a node; further information about the object can be provided by properties or attributes. This approach is quite similar to the representation of entities in an entity relationship model as used in database design or similar to object-oriented programming. An edge between two nodes represents a relationship between one concept (source node) and another concept (target node). The renaissance of semantic networks became reality since new techniques for a graphical visualisation of such networks have been developed. Knowledge networks are visualised as so called topic maps. Like on a map the distance between nodes on a topic map reflects the semantic distance between concepts in the knowledge network. A semantic search proceeds now as follows: Assume we start to search for knowledge related to a notion A. First a syntactic search starts to search all documents, attributes and links included in the knowledge network. The node which fits best will be visualised as a centre node on the screen, a view onto the knowledge network is produced showing a part of the whole network. Now we get additional information about the relationship of the centre node to other concepts. They are shown on the screen as edges to other nodes. Thus we can additionally consider concepts not containing our original search word. The visualisation of knowledge networks as so called topic maps reflects the semantic distance of notions as a geometric distance between the nodes (points) in a map. In the year 2005 we ran the project ToMaHS – Topic Maps für HochschulStrukturen (Topic Maps for University Structures). The aim of the project was to experience the development process for a knowledge network. We used an application area we are all familiar with: our university. In order to reduce the complexity we focussed on two parts: 1. degree programmes in Business Informatics (Bachelor and Master programme), 2. university administration These parts represent different views: a degree programme as a “product” of a university and an administration as it may also occur in other organisations. The project could be based on the K-Infinity software provided by the intelligent views GmbH Darmstadt, Germany. As described above a knowledge network consists of notions and their relationships. Therefore the main part of the project work had to be spent on the identification of Baltic Business and Socio-Economic Development 2008 notions and relationships. The K-Infinity software distinguishes between terms, subterms, instances, relationship and extensions (roles). A small example representing the top management of the university illustrates the concepts (see picture 1): The so called Rektorat (directorate) is the top management of the university. The head of this board and therefore head of the university is the rector, usually one of the university professors. At present Prof. Norbert Grünwald is the rector of Hochschule Wismar. In general the real world has to be mapped into a knowledge network. The administrative structure of the university has been mapped into the knowledge network by the top term “Organ” and various subterms. A person can be a member of one or more such organs and does different duties: The person may have certain functions. What about a real person like Prof. Norbert Grünwald? An individual is mapped into the knowledge network as an instance of the term “Person”. An individual can have different duties in different environments, e.g. a person can act as a university teacher (professor) giving lectures. The same person may be the head of an administration organ and has management duties. How can we express different behaviour in different environments? E.g. Prof. Norbert Grünwald is a (the) rector and Prof. Norbert Grünwald is a professor. From the point of object-oriented design the individual Prof. Norbert Grünwald should be represented as an instance of a class “Professor” as well as an instance of a class “Rektor”. Unfortunately such a design causes trouble concerning the multiple inheritance4 of properties. The software KInfinity solves the problem by offering the concept of so called roles and extensions: An individual is represented as an instance of the term “Person”. An instance of “Person” can have several roles: he acts as a professor, as a member of a board, or as the rector. Picture 1 shows the individual Prof. Norbert Grünwald and his roles. Picture 1: Knowledge builder window: part of the university knowledge network A term in a knowledge network consists not only of a single name. It is the representation of the real world entity and therefore consists of a set of attributes and relationships. For every term a set of attributes has to be identified. Often a special attribute contains a link to a corresponding web-page. The web-page will be shown whenever the focus is on that term or instance. 4 „Multiple inheritance” is feature discussed in object-oriented programming languages. Baltic Business and Socio-Economic Development 2008 At the end of our project we got two small knowledge networks: The university administration network reflects the university structure and the roles of persons within the university administration. The network for our degree programmes has been developed beyond that level: The network contains terms, their attributes and relationships but all the real documents and web-pages as well. It can be accessed via: http://gauguin.wi.hs-wismar.de:3000/winet. 3.3 Conclusion The project work has shown that a knowledge network can be developed quite easily. First steps can be made quite fast. But the work should not be underestimated. Here are some conclusions: – It is necessary to start a design process: A structure of terms, subterms, instances and relationships has to be developed. The question is: Which type of real world entities is represented by which concept? – Although the K-Infinity software offers a user-friendly interface, working with the system needs serious training. The development of a knowledge network using software like K-Infinity or Ontopia requires at least basic knowledge in computer science. – A network of named nodes and edges is not a knowledge network at all. Other knowledge, documents, information have to be attached or linked to the nodes and edges. Keeping in mind that even a small knowledge network contains several hundred nodes this work is rather tedious. – A knowledge network will be used only if there is enough knowledge available and the knowledge is up to date. The network has to be maintained. The maintenance of the university knowledge network could not be solved within the project. 4. Wiki and Semantic Wiki Nowadays it is not necessary anymore to explain the word “wiki”. Wikipedia has become a source of information and knowledge not only for every day life but for education and even research. Especially in learn management systems (LMS) wiki systems have been widely used for co-operative work. In software engineering the wiki system trac has been applied as a project management tool. Meanwhile even some bigger companies5 have introduced wiki as part of their intranets. Why has the wiki approach become so widely used and famous? There are mainly two reasons: – – Wiki pages can be accessed very easily. Only a web-browser is necessary. Everybody can easily contribute to wiki pages, e.g. everybody can easily edit them. Wiki has been established as a useful tool for co-operative and collaborative work6. It has to be mentioned that there can be some stumbling blocks on the way to the first use of wiki: Most organisations and companies run a hierarchical structure. Rules exist about responsibilities: Who is responsible for a certain piece of information. Therefore some people worry about openness of the system: If everybody can change everything then 5 Wiki system at IBM: Gunter Dueck, „Bluepedia”, Informatik Spektrum 31(2008) 3, 262–269, in German. 6 Stewart Mader: Wikipatterns – a practical guide to improving productivity and collaboration in your organization, Wiley Publ. Inc. Indianapolis, 2008 Baltic Business and Socio-Economic Development 2008 misuse and malpractice will appear and will block any useful application. Experiences have shown that these are groundless fears: Since every change will be logged everybody works quite responsible. Moreover most participants not only feel their responsibility but feel a little bit proud to make own contributions. Picture 2: Wikipedia: A typical window structure of a Mediawiki application A search engine in a wiki system is still based on a lexicographical (so called syntactical) search only. Since every page in a wiki system usually contains a lot of references to other pages in the system or links into the World Wide Web we already have a good chance to get the required information quite soon. But still the problems mentioned above remain. Now the semantic wiki approach tries to combine the advantages of wiki systems and those of knowledge networks. Whereas ordinary wiki pages have untyped links to other pages links can be named and typed in a semantic wiki. These links represent the relationship between the wiki pages, e.g. between the concepts described at these wiki pages. Thus we can see a semantic wiki as a special type of knowledge network: Nodes (concepts, terms, individuals …) are represented as pages in the semantic wiki system. Edges representing relationships between nodes are implemented as annotations to the links between wiki pages. Picture 3: Semantic MediaWiki homepage The structure of pages and annotations in a semantic wiki offers new possibilities for searching the system. Queries similar to those in relational databases are possible. An Baltic Business and Socio-Economic Development 2008 additional tool the so called ontology browser allows to search and maintain the structure of the wiki system, e.g. to manage the structure of the knowledge network. In Schaffert et al. (2007) an overview of some existing semantic wiki systems has been given. We have tested the following systems: – Semantic MediaWiki – OntoWiki – IkeWiki OntoWiki offers a rather strange user interface it has been hard to do even the first steps. IkeWiki offers very powerful mechanism for the annotation of links. A drawback is that the system requires a lot of resources: cpu power as well as memory. We suspected that the system could serve more than only a few user accesses in parallel. Picture 4: Ontology browser in a semantic MediaWiki system At present the Semantic MediaWiki seems to be the one which can be applied best. Since it has the appearance of the MediaWiki used in the Wikipedia everybody is quite familiar with the window structure and can start to use the system right away. There is a large community working on the further development of the system. Installation of a Semantic MediaWiki system is straightforward and easy. We decided to run a Semantic MediaWiki system for our team: Artificial Intelligence in Business Informatics. The step ahead from plain wiki to semantic wiki is an important one: Since semantic wiki systems define nodes (wiki page) and edges (annotated links) it is possible to give a visualisation as a topic map (see e.g. Redmann and Thomas 2007). In such a way the easy-to-use advantage of wiki systems can be combined with the semantic search capability and the user-friendly visualisation of knowledge networks as topic maps. 6. Conclusion There have been reports about using a wiki system for the management of knowledge. Since “… knowledge management is about how people share and use what they know”7 a wiki system offers a good chance to get all staff members collaborating and therefore improving the knowledge management system. 7 Carl Frappaolo: „Knowledge Management“, Capstone Publ. Ltd. (A Wiley company), 2006, page 119. Baltic Business and Socio-Economic Development 2008 Semantic wiki extends the wiki approach into a knowledge network technique. wiki pages are enriched by semantic information e.g. information about the relationships between pages or concepts. Thus, links can be defined that create a knowledge network. A graphical representation of the network as a topic map can then be deduced automatically. Moreover annotations may define logical dependencies as well and an inference machine may be used to make implicit knowledge explicit. Still there are no reports that wiki systems have been used in SMEs or as a platform for the distribution of knowledge among SMEs and universities. There are several ideas for the application of semantic wiki systems: An intelligent information system in a company or an organisation. As a regional information system a semantic wiki can be a big step ahead for the distribution of knowledge or the transfer of technology between universities and business. We argue that wiki systems for these purposes should be launched in the near future. Since wikipedia is well known it can be expected that users are well motivated to contribute to such a system. We propose a project on the development of semantic wiki systems within the Baltic Business Development Network. For instance the knowledge of a university expressed in a semantic wiki could lead to new co-operations: Co-operations between university and companies or between universities or regions. Regional development will profit a lot if competencies and needs can be managed together in one system. Bibliography Carl Frappaolo: „Knowledge Management“, Capstone Publ. Ltd. (A Wiley company), 2006. Gunter Dueck: „Bluepedia”, Informatik Spektrum 31(2008) 3, 262–269, in German. Stewart Mader: Wikipatterns – a practical guide to improving productivity and collaboration in your organization, Wiley Publ. Inc. Indianapolis, 2008 Intelligent Views: „The Knowledge-Builder“, www.i-views.de/web/images/stories/tblimg/pdf/english/dbl_KB_engl.pdf, last visited 2008-08-20. Sebastian Schaffert, François Bry, Joachim Baumeister, Malte Kiesel: „Semantic Wiki”, Informatik Spektrum 30(2007) 6, 434-439, in German. Tobias Redmann, Hendrik Thomas: „The wiki way of knowledge management with topic maps”, Proceedings of the International Conference on Information Society (iSociety 2007), Merrillville, USA, October 7-11, 2007. Uwe Lämmel, Jürgen Cleve, René Greve: „A knowledge network for the university the project TOMAHS“, Wismar Discussion Papers (WDP), 19/2005, Hochschule Wismar 2005, in German.