KnowledgeGraphIdentification

You may want to read the PSL Getting Started guide to better understand PSL: https://github.com/linqs/psl/wiki/Getting-started

iswc13: To quickly get started:

1. Dowload the dataset: bash fetchDataset.sh

2. Build your classpath using Maven: mvn dependency:build-classpath -Dmdep.outputFile=classpath.out

3. Compile the KGI model: mvn compile

4. Load the dataset: java -cp ./target/classes/edu/umd/cs/psl/kgi/:./target/classes:cat classpath.out edu.umd.cs.psl.kgi.LoadData data/

5. Run KGI: java -Xmx15800m -cp ./target/classes/edu/umd/cs/psl/kgi/:./target/classes:cat classpath.out edu.umd.cs.psl.kgi.RunKGI > out

6. Output scores (MusicBrainz): perl scripts/cal_f1_auc_j.pl out data/1000_cat.ground.txt data/1000_cat.noise.txt

7. Output scores (NELL): perl scripts/cal_f1_auc_j.pl out data/label-test-uniq-raw-cat.db.TRAIN data/label-test-uniq-raw-rel.db.TRAIN