Google Scholar

Google Scholar through HasData

Two tools. hasdata_google_scholar_scholar_getScholarSearchResults takes q and returns a page of results. hasdata_google_scholar_cite_getScholarCitationFormats also takes q, but there q is a resultId such as A7L9JolPKkoJ rather than a query. Sending a title or a DOI to the citation tool returns nothing useful, and the shared parameter name is why this goes wrong.

The two ids that look alike

Every result carries citedBy.citesId and versions.clusterId, and on some papers the two hold the same value. They are not interchangeable. citesId goes into cites to find the papers citing this one, clusterId goes into cluster to find every indexed copy of this one. The wrong id in the right parameter returns a plausible page of the wrong thing.

publicationInfo.authors gives an authorId per author Scholar has a profile for, which is how to follow one author instead of parsing a byline.

What the numbers mean

searchInformation.totalResults is Scholar's estimate and it is very rough. Treat it as an order of magnitude and never quote it as a count.

Scholar counts citations rather than quality, and it indexes preprints, theses and citing-only records alongside published work. asVis: 1 drops the citation-only entries when the task needs records with full metadata.

Widen, do not page

Deep paging gets thin. Scholar caps how far a result set really goes and starts repeating or blocking well before the estimate suggests. Raise num and narrow with asYlo, asYhi or asSdt rather than walking start upward, which also costs less because the price is per call.

scisbd set to 2 sorts by date instead of relevance, which is what a "what is new" question needs.

A BibTeX link from the citation tool is a Scholar URL carrying an expiring signature, not the BibTeX itself. Fetch it promptly rather than storing it.