Accepted_test

Development of a hybrid tool system for semi-automatic ontology creation
by Lebedev Mark | Savostyanov Aleksandr | Novosibirsk State University | Scientific Research Institute of Neurosciences and Medicine
Abstract ID: 722
Event: BGRS-abstracts
Sections: [Sym 10] Section “General problems in the study of cognitive processes; models of cognitive activity”

This work aims to study methods of ontology creation automatisation and compare already existing tools to help formulate requirements for the hybrid expert-automatic instrumental complex for text parsing, term and link extraction and ontology creation.

Several scientific papers on ontologies and ontology creation have been studied and a few preffered tools have been chosen. The final complex will utilise the 4-level ontology system architecture, first developed by Pal'chunov D.A. Both linguistic and statistical automation methods will be tested to find the most accurate one. Testing data will be parsed from PubMed and provided by the Scientific Research Institute of Neurosciences and Medicine.

Final requirements for the instrumental complex include:

  1. Techincal stack (Python).
  2. Integration (the ability to work with existing ontology creation and editing tools).
  3. Simplicity (made for use by non tech fluent personel).
  4. Descreteness (different modules have the ability to function on their own).

Also, the tools architechture has been developed: 4 modules with 1 being full expert, 1 fully autonomous and 2 automatic modules involving expert verification.

These preparations provide solid ground for initial development and testing which is planned for the next year.