PubMed search matches subset of term name when there are no matches
dhimmel opened this issue · 3 comments
C566272 is Townes-Brocks-Branchiootorenal-Like Syndrome. Looking on the mesh browser this term has a frequency of 0 meaning it's never tagged a medline topic.
We're currently using the following PubMed search: Townes-Brocks-Branchiootorenal-Like Syndrome [MeSH Terms:noexp]
, which returns 117,315 results and has the message:
The following term was not found in PubMed: Townes-Brocks-Branchiootorenal-Like
Note now this doesn't include "Syndrome". So I think what's happening is that PubMed isn't finding the entire search term and is falling back to just searching Syndrome [MeSH Terms:noexp]
, which is matching 117,315 records.
So we probably have to quote the search term.
Looking up PubMed citations by the MeSH ID like C566272 [MeSH Terms:noexp]
would be far preferable since it would reduce the parsing and encoding complexity while guarding against any name changes to a MeSH term. However, I don't think this is supported. I don't see it mentioned in Searching PubMed with MeSH.
It's also possible we could download bulk data at https://www.nlm.nih.gov/databases/download/pubmed_medline.html.
I think the correct search would be "Townes-Brocks-Branchiootorenal-Like Syndrome" [Supplementary Concept]
, which quotes the MeSH label and uses the supplementary concept suffix as per #4. This query returns 0 results, which is expected, and provides the following message:
No results were found. Your search was processed without automatic term mapping because it retrieved zero results. The following term was ignored: Townes-Brocks-Branchiootorenal-Like Syndrome
I followed up on my contact with Ryan Cohen from the NCBI Helpdesk in #4 (comment):
I had another issue where I didn't properly quote the MeSH term name: #3. Do you know whether it's possible to search by MeSH ID? For example, is there a way to have a query like "C006737 [Supplementary Concept]"?
Any documentation on how to quote / escape whitespace or special characters in search terms?
Ryan replied:
For "Townes-Brocks-Branchiootorenal-Like Syndrome" [Supplementary Concept], this can be used in another NCBI database such as GTR. There are currently no citation records in MEDLINE/PubMed indexed with this MeSH SCR.
No, there is not a way to search by MeSH ID. The MeSH translation table does not include the MeSH ID.
Please see the PubMed help for a list of PubMed character conversions.
From the character conversions docs:
double quotes " - used to force a phrase search