What do you think selectivity is in terms of database? Do I have to say selectivity is good or selectivity is low? Is low selectivity is good? or not?

Here’s my one penny. It seems that there are conflicting definitions of selectivity. When people(textbooks and papers related to System R) talks about ‘histogram’ or ‘query’, selectivity is # of rows returned by query. However, when people (Vendors of commercial RDBMS) say about index, selectivity is # of distinct rows divided by # of rows .

If we accept first definition, then low selectivity is good candidate for index. However, the latter definition is opposite; high selectivity is good for index.

So, I think I’d better define the selectivity equation before I talk about selectivity. Always. Even professor or Ph.d. did not know exact meaning of selectivity. They just say, ‘selectvity is good’ rather than ‘selectivity is low/high.’ That’s good for colloquial statements, but to draw a figure we need exact definition.