ag-sc / jlink

A Joint Entity Recognition and Linking Tool for technical Domains using Undirected Probabilistic Graphical Models with BiGram (former BIRE)

JLink

A Joint Entity Recognition and Linking Tool for Technical Domains using Undirected Probabilistic Graphical Models with BiGram (former BIRE)

NOTE: The documentation is far from being complete. If you have any issues feel free to write me an email. I will try to add more documentation by time.

Quick start:

1. Clone BIRE simplified-api branch from: https://github.com/ag-sc

2. Clone this project.

3. Import projects into your IDE-workspace (preferably eclipse).

4. Build maven. (If necessary)

5. Update config file "config/BC5_test_dataset.properties" and create folder structure in workspace project.

6. Download files from:

   6.1 CDR_Corpus http://www.biocreative.org/resources/corpora/biocreative-v-cdr-corpus/

   6.2 CTD_diseases "CTD_diseases.tsv.gz" (tsv-version) http://ctdbase.org/downloads/

   6.3 CTD_chemicals "CTD_chemicals.tsv.gz" (tsv-version) http://ctdbase.org/downloads/

   6.4 Stanford-postagger (english-bidirectional-distsim.tagger) https://nlp.stanford.edu/software/tagger.shtml

7. Copy requiered files according to your config-file. Training, Dev, and Test data can be found in the downloaded zip-folder in CDR_Data/CDR.Corpus.vxxxxxxx/*PubTator.txt.

8. Since a lot of things are loaded to memory, we need approx. 8g to run smooth. In Eclipse add -Xmx8g to VM Arguments.

9. Run the Main.java

HINT:

If you want to test only on test data, merge training and development files from CDR_Corpus into one file. Use this as training data corpus.

If you want to apply a learned model to completely new data, merge all three files (train, develop, test) into one big file and use this for training.