Home

Resources

• For a regex-based approach, see https://github.com/dice-group/SPAB
• LsqSpinToArff code: https://github.com/dice-group/LsqSpinToArff

LSQ

• Website: https://dice-group.github.io/LSQ/
• Code: https://github.com/dice-group/LSQ
• Code: https://github.com/AKSW/LSQ (https://github.com/AKSW/LSQ/issues/7)
• Run: lsq -f input.log -o output.ttl -r q (Tested with part of SWDF.log, 1.1 GB, link at LSQ website)
• Run from Eclipse: Use project lsq-cli and class org.aksw.simba.lsq.cli.main.MainLSQ.
  Set CL arguments -f input.txt -o output.ttl -m sparql -r q

LSQ insights

• Input formats are defined here: default-log-formats.ttl
  (There are more formats than in the documentation)

Weka / ARFF

• Weka: https://www.cs.waikato.ac.nz/ml/weka/
• ARFF: https://waikato.github.io/weka-wiki/arff/
• Weka videos: Data Mining, More, Advanced

Weka usage

• To use J48, you may have to add the filters NumericToNominal and StringToNominal.

Weka code

• SVN: https://svn.cms.waikato.ac.nz/svn/weka/
• Maven: https://mvnrepository.com/artifact/nz.ac.waikato.cms.weka/weka-stable
• Git: https://github.com/Waikato/weka-3.8 | https://github.com/Waikato/weka-trunk

LsqSpinToArff

For multiple runs, you can create a directory containing files named as follows:

SELECT-1-pos.txt  
SELECT-1-neg.txt  
SELECT-2-pos.txt  
SELECT-2-neg.txt  
...

Create a file with prefixes using the command ls -1 DIRECTORY > prefixes.txt.
Remove the -pos.txt and -neg.txt parts.
Use the following terminal/console code to process all files:

for i in `cat prefixes.txt`; do
echo $i;

java \
 -Xms1200m -Xmx1200m    \
 -jar LsqSpinToArff-jar-with-dependencies.jar \
DIRECTORY/$i-pos.txt \
DIRECTORY/$i-neg.txt \
./results \
lsq-cli-1.0.0-jar-with-dependencies.jar

done