KNOWLEDGE MANAGEMENT INSTEAD OF DATA CHAOS

Knowledge is becoming an increasingly vital resource. To make the best possible use of this resource, it is necessary for data and information to be linked together in a semantic room. Semantic technologies help to link existing information.

A few weeks ago, we published a blog article on digital technologies relating to the subject of "Content" on the EDAG-PS homepage, and have decided to explain things in more detail. In our short series covering technologies such as text mining, semantics, search engines and the term knowledge management, we will explain today what "semantics" is all about. Here, too, we have interviewed a specialist - Harald Schenda - EDAG PS' head of technical information.

EDAG-PS editor: What beneficial effect do semantic technologies have?

Harald: Even though we don't have a great deal of space here, I'll have to go into the subject in some detail. We'll confine ourselves to two fields: the Web and knowledge management applications. The idea of the Semantic Web is really quite old and originates from Tim Berners Lee, a computer scientist and physicist. Back in the 1980s, Tim had the idea which is only now gaining significance. To give you a rough idea: imagine that all ideas and concepts were linked to one another through logical relations, known as directed graphs. If this were the case, information could be channeled more effectively and tailored to the human mind. Or the other way round, human knowledge could be made machine-readable in this way.

EDAG-PS editor: That all sounds very theoretical. Can you give us an example?

Harald: EDAG PS works in the production environment. Suppose you enter the search item "robot" in the input mask. It would be good if the underlying logic "knew" that a robot is not an isolated unit that works all alone in the production plant, but that a gripper, docking station, controls, a hose kit, media supply, console etc. are also required for it to work.

Einfache Ontologie

Image: Simple ontology

EDAG-PS editor: So how do you introduce this logic into a Web application?

Harald: The logic first needs to be modelled. The word "logic" is perhaps a bit misleading. In actual fact, we mean concepts that are linked to a context. The links belonging to a robot or the different elements, or to enable the robot and a person to work together on a task, need to be created manually.

EDAG-PS editor: Why are concepts and contexts important?

Harald: Because information is only valuable in in connection with concepts and contexts. If I say "35 degrees" to you, this is information without a context. If I say "35 degrees", I, yesterday in a street café in Berlin-Moabit, then I have given the information a context.

EDAG-PS editor: Modelling concepts and contexts must be a very complex task!

Harald: If you model everything by hand in an ontology language, then it really IS complex. But software solutions are available for these models. These make it possible to import files in XML or CSV format for instance, if they contain the logical relations required.

EDAG-PS editor: OK, and what is the result?

Harald: There are a number of different possible applications. Let's assume we are working in the knowledge management domain. Assuming a DMS (Document Management System) is being used, the semantics can help to bring about a better search result. If you enter "robot", the text indexing and semantics mean that you get additional results which have previously been defined in a logical model.

EDAG-PS editor: Can you also provide certain target groups with specific types of knowledge?

Harald: Yes, you can! It does, of course, depend on the exact application. We have constructed a demo to show how the DMS, search engine and semantics interact. The user can input his role into this demo and the rules and regulations he needs to observe are then displayed in the form of documents. If this were programmed to make it is user-friendly, then the information concerning which user group the user belongs to would be transmitted when he logs in, and the information would then be provided immediately, without any entries being made.

Confluence_Demo mit iFinder-1

Image: Confluence demo with iFinder search engine integration and i-views knowledge graph

EDAG-PS editor: So the network is created via the terms, and then linked to the relevant documents.

Harald: Yes, that's right.

EDAG-PS editor: So quicker access to the individual subjects is generated via the term network. But that doesn't make the content any better, does it?

Harald: You are absolutely right there! It is not possible to draw any conclusions as to the quality of the document pages the user ends up on as a result of the semantically supported search. As a rule, we end up with contents gained from books, and and you just have to work your way through it. We can make it easier to access subjects, but this will not alter the quality of the contents. Here, I can only sing the same old song about topics: because this is way to be able to give a short, cohesive answer to a specific question.

EDAG-PS editor: Thank you for talking to us!

Have you got any questions regarding semantic technologies, knowledge management or content? Then get in touch with our Head of technical information, Harald Schenda.