Knowledge Discovery from Scientific Databases, Internet and Medical Records

Summary

There is only a limited amount of safe knowledge on the course of schizophrenic psychoses, although more than 100 years have passed since first description of "Dementia praecox" first was described.
Text knowledge enables you to find relationships in nonstructured or less structured textfiles. About 80 - 90 percent of information in scientific papers and other sources on the Internet are textual. The most valuable information is often hidden not in organized searchable tables, but in freeform text.
By using Text Mining technology it is likely that hidden links in the scientific literature will be uncovered implying new hypothesizes for further brain research.

Project aims

The main objective of this project is to create a Data Warehouse of information and use new and efficient techniques for computer structuring of large textual data on schizophrenia and related psychotic disorders. By doing that, it is likely that new insights in brain research will be discovered and that hidden links in the scientific literature will be uncovered.
Furthermore, the categorization of data will be used to define relevant documents and forward them to different project leaders of the HUBIN-project. The information extracted will also be used in a World Wide Web-project for scientists, journalists and for the general public.
The need for software tools to deal with online documents on the Internet is already large and is growing larger. Data sensing, acquisition and storage technologies have led to vast observational data sets being routinely reported in almost every aspect of biology and medicine. Unstructured data- such as text - will still become the predominant data type stored online. The problem with text is that it is not structured like the tabular information typically stored in databases.
Advanced text analysis for discovering crucial information in scientific documents are needed. To be able to mine text for information, extraction tools are needed. Tools that can for example:

Extract key information from text
Organize documents by subject
Find predominant themes in documents
Looking for missing links between documents Relation maps between keywords

This information can be used as metadata about the documents and used in turn for Data Mining and allowing the computer to generate new hypotheses.

Project leader: Jan-Eric Litton