I noticed that if I perform a TPC search using two or more keywords connected by the OR boolean operator along with a category search, I get unexpected results and apparently the order of keywords matters.
For example, if I enter a single keyword:
"daf-2"
and select the category "phenotypic perturbation" and use the "Sentence" search scope I get back 2763 documents with 22578 matching sentences. If I instead use the keyword "Y55D5A.5" (the sequence name for daf-2) I get back 7 documents with 9 matching sentences. If I now enter both keywords like this:
"daf-2" OR "Y55D5A.5"
I get back the same 7 documents with 9 matching sentences as with "Y55D5A.5" alone. If I then reverse the keyword order:
"Y55D5A.5" OR "daf-2"
I get back 2763 documents with 22578 matching sentences (same as with "daf-2" alone). It seems that when the OR operator is applied, it is just returning results as if there is just the second keyword.
This feels like a bug so I'm reporting it. Is this the intended behavior? If so, I think it should be corrected so that a user gets back the union of both searches.
I noticed that if I perform a TPC search using two or more keywords connected by the OR boolean operator along with a category search, I get unexpected results and apparently the order of keywords matters.
For example, if I enter a single keyword:
"daf-2"
and select the category "phenotypic perturbation" and use the "Sentence" search scope I get back 2763 documents with 22578 matching sentences. If I instead use the keyword "Y55D5A.5" (the sequence name for daf-2) I get back 7 documents with 9 matching sentences. If I now enter both keywords like this:
"daf-2" OR "Y55D5A.5"
I get back the same 7 documents with 9 matching sentences as with "Y55D5A.5" alone. If I then reverse the keyword order:
"Y55D5A.5" OR "daf-2"
I get back 2763 documents with 22578 matching sentences (same as with "daf-2" alone). It seems that when the OR operator is applied, it is just returning results as if there is just the second keyword.
This feels like a bug so I'm reporting it. Is this the intended behavior? If so, I think it should be corrected so that a user gets back the union of both searches.