1

I would like to build a local bibliography database (possibly with SQL) with the data exported from Web of Science (WOS) search results. I can export these online query results in a BibTeX file and process this file offline with my favorite scripting language.

Now, the problem is indexing the data. WOS provides each BibTeX entry with their own "WOS Unique ID" (often times, but not always, provides DOI), however, researchers' names are not uniquely identified. This is a problem because researchers often have different names in different papers and other BibTeX entries.

I really need the capability of indexing authors on my local database, to interface it with my scripting language. Ideally, WOS would allow exporting the query results datafile with some "author IDs". This would solve my problem, but can it be done for free? Is anyone aware of some other method I could explore to go about this problem? Thanks a lot!

  • Just to note that Google Scholar, ResearchGate, etc all have the issue of resolving author name ambiguity, so I don't think you're going to find a solution that easy. You can try to use ORCIDs or authors with Google Scholar profiles or such, but this is going to be a lot of work and involve a lot of cross referencing between databases. – Caharpuka Nov 18 '19 at 22:47
  • 1
    I'm not entirely clear what your end goal is that you specifically need WOS, but Zotero stores all the data about publications you import into a SQLite database (Mendeley does as well, but it's encrypted with a hidden key). – anjama Nov 19 '19 at 0:21
  • @anjama WOS is a useful and curated tool for the kind of searches I use to build a database from BibTex files. I can search all the papers from some Institution in a given year for instance. – bibliotech Nov 19 '19 at 0:29
  • @anjama Can those programs work from BibTeX files like JabRef? And can any of these programs (Mendeley, Zotero, JabRef, etc.) take the input BibTex and look for similar author names automatically, while poping up a request for the user if the such and such author names could be considered the same and "merged"? – bibliotech Nov 19 '19 at 0:34
2

Author disambiguation is a big problem in most plattforms / citation databases, and an issue in much bibliometric work. I don't think you will find a quick fix. Orcid (which WoS sometimes provides) is aimed to partially solve this, but depends on researchers actually having and using Orcid. Scopus currently has a better functionality in finding publications from standardized author records (i.e. "all" publications for an author, after their algorithm for merging authors have been applied), and this is also exported in e.g. their CSV format in the field "Author(s) ID".

See for instance these exported records from Scopus, with overlapping authors (and therefore author-IDs):

'Author(s) ID'  EID
24802234300;6602122539; 2-s2.0-84861411180
24802234300;55923304700;6603273826;6602122539;  2-s2.0-77955097159
16229637800;6602122539;24802234300;7003423871;  2-s2.0-51749084398

So if you want to build a local database from an online bibliometric source, which include some form of unique author ID, I think Scopus would be a better starting point. Web of Science has their own identifier included in exports (ResearcherID in the RI field) and Orcid IDs (OI field), but only for researchers that has those IDs, which is usually the minority (both across papers and within individual papers). One downside with Scopus is that there export does not include Orcid from what I know (at least not in the open web interface or the basic API), so it may be harder to integrate with other data sources.

Note however that all identifiers are not always included in all export formats (goes for both Scopus and WoS) so you should focus on e.g. csv-exports (which include the identifiers I mention above).

|improve this answer|||||
  • Great, thanks! I'm now a fan of Scopus. Do you know if these "Author(s) ID" are provided in the same order of the names in the "Authors" field? – bibliotech Nov 22 '19 at 19:23
  • I think so from examples I’ve seen (it is e.g true in the example data above). I’ve mainly worked with WoS data (since we have local access to it), so I’m not that familiar with all the inns and outs of Scopus data though. Generally, address data is cleaner in WoS, but Scopus has come further in this type of author data. – fileunderwater Nov 22 '19 at 19:30
  • What do you mean by "address data"? – bibliotech Nov 23 '19 at 16:56
  • @bibliotech I mean the author affiliations, e.g. “Stanford, Dept. comp sci, Stanford, USA”. – fileunderwater Nov 23 '19 at 17:26

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.