instead of excluding datasets from the deduplication process, we include only publications
codebase used to migrate to java8 the production system