It's LUBM not BSBM about lingbm HOT 4 CLOSED

Thanks for spotting this issue. @chengsijin0817 can you please fix that README.

@VladimirAlexiev In fact, we had completely defined and implemented a version of LinGBM based on the BSBM datasets. Unfortunately, once we started doing our first experiments with it, we realized that the BSBM data generator is not monotonic; that is, when generating datasets at two different scale factors (even with the same random seed), the smaller dataset is not guaranteed to be contained in the bigger dataset. As a consequence, we observed that query results of the benchmark queries were not monotonic at increasing scale factors. In other words, for the same query, when executed over increasingly-larger datasets, the result size was bouncing up and down randomly. Such a behavior is, of course, absolutely undesirable for a benchmark. So, we had to bite the bullet and start over again from scratch, by using LUBM this time.

Something else in your message sparks my interest: Where did see that "LUBM [...] can produce both RDBMS and RDF"? I believe that the original LUBM dataset generator can generate only RDF. That's why we extended it to also generate RDBs.

from lingbm.

Thanks. I have fixed the issue now (See commit #7d4c80c).
Please check it.

from lingbm.

Ah, I misunderstood. I've read you extended the LUBM gen, but I thought that only related to outputting a GraphQL schema

from lingbm.

I've read you extended the LUBM gen, but I thought that only related to outputting a GraphQL schema

Note that our extended dataset generator does not output a GraphQL schema. Instead, the LinGBM GraphQL schema has been defined manually by us, and then we have defined a mapping from that GraphQL schema to the RDB schema of the generated datasets in order to specify how the GraphQL schema has to be implemented over the datasets. (Actually, I see now that the definition of this mapping is not complete yet.)

from lingbm.