Experiments scripts for the benchmark used in k-HDTDiffCat
In wiki-changes. The user should have a recent precomputed Wikidata truthy HDT. See Wikidata generation.
Compile it with ./gradlew shadowJar
.
Create delta file using
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesDelta --date 2023-10-31T00:40:00Z -f simple -m 0 -S 10000
2023-10-31T00:40:00Z
is the date to rollback
It'll create a delta.df
file containing the dataset updates.
## PREPARE DATA
# generate an hdt from the delta file
rdf2hdt -config delta.df delta.hdt -multithread -color
# Split
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesCompute div delta.hdt 7
# Compute delete bitmap
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesCompute bitmap delta.hdt wikidata-truthy.hdt outbitmap.bin
## BENCHMARK
# Cat only test
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesCompute catonly diff.hdt delta.hdt
# Merge diff test
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesCompute mergediff wikidata-truthy.hdt outbitmap.bin delta.hdt 7
# Diff only test
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesCompute diffonly wikidata-truthy.hdt outbitmap.bin
# CatDiff test
java -cp wiki-changes.jar com.the_qa_company.wikidatachanges.WikidataChangesCompute catdiffonly wikidata-truthy.hdt outbitmap.bin delta.hdt