leiloong / learningdl

This is source code of project "A Hybrid Approach to Learning DL Ontology From Text"

This is source code of project "A Hybrid Approach to Learning DL Ontology From Text"

Prerequisites:

1. Metamap version 2013. When install, put public_mm folder in the same directory with this source code directory
2. Python. To test, run python generatePositiveExamplesOriTest.py
3. Stanford Parser. After extracting the folder, make sure that you rename the outer folder to become stanford-parser (rename stanford-parser-full-20xx-xx-xx to stanford-parser). Then put that folder in the same directory with this source code directory

How to run:

1. Compile all .java files by running javac *.java
2. Run GUI class by command java GUI
3. Insert input file containing biomedical text. You can try to insert file input.in in this folder
4. Click Process and see the result

How to learn a GCI is formed by system:

User can see a set of sentences where a GCI comes from by clicking the ? symbol.
In the second box, the system will show those set of sentences.

How to validate a GCI:

User may change the role relation between two concept names.

1. To validate, click the V symbol.
2. In third box, system shows two concept names and their role in the format <first concept name> | <role relation> | <second concept name>
3. User can change role relation by writing this abbreviation:

• AM: Associated Morphology
• AM-1: Associated Morphology (inverse role)
• FS: Finding Site
• FS-1: Finding Site (inverse role)
• CA: Causative Agent
• CA-1: Causative Agent (inverse role)
• IS: Is-A relation
• IS-1: Is-A relation (inverse role)

4. and finally, click Train.