University of Hamburg - to research, to teach, to educate and form, to homepage
Department of Chemistry
Departmentof Chemistry
Search

Photo: Vill

Knowledge Systems

Databases usually do not capture knowledge, but only measured values, numbers and texts. Researchers, on the other hand, are able to understand relationships (rules) between numbers and recognize analogies between facts, so that new knowledge can be gained with this knowledge. Pattern recognition and understanding similarity are thus important skills for the researcher.

If knowledge is to be conveyed by an electronic information system in addition to pure data, rules and similarity concepts must be integrated. Object-oriented databases can make this possible if properly configured. An electronic knowledge system can never replace a researcher, but it can do a much better job of providing information than a pure database is capable of. It can thus become a scientific tool with which a researcher can identify new relationships and even quantify them.

Features of such a knowledge system are:

  • A proper ("species-appropriate") data description
  • Valid, reliable data
  •  Integration of similarity models and knowledge rules, as well as inheritance rules

Classical, relational database systems are standardized by the rules of the SQL database language - but thus also regulated and "limited". The data and facts to be described must be adapted to the rules of a given set of data types (floating point numbers, texts, etc.). Data and program code are clearly separated and each database entry is independent of all others, thus without "relationship".
In an object-oriented system, on the other hand, the data description can be natively adapted to the facts to be described.

Information systems on hazardous substances are particularly important. They play a crucial role in occupational safety.

Contributions of the research group

  1. Development of the database management program SciDex
  2. Development of the liquid crystal database LiqCryst with prediction methods
    Development of the chemicals management program CLAKS
  3. evelopment of a comprehensive database of hazardous substances
  4. Creation of an overall index of all substances in the Landolt-Börnstein