February 2006

Union catalogs from Hungary, Iceland, Israel, Portugal, Sweden, and Switzerland are now visible with links to bibliographic information and items via the Google Scholar interface. Try searching for information science and find links for some of the hits to Library Search (Sweden), if you’re coming from Swedish domain.

If, for example, you’re coming from the Swedish domain but want to check items from other union catalogs, just go to Scholar Preferences and search for the country to configure the union catalog you want. Then save your preferences.

Make the same search on information science and you will get the link Find in RERO if you have chosen Schweiz as in my example.

Read the announcement Global searches go to local libraries at Google official blog 2/20/2006.

When searching journal title abbreviation (in the version of Web of Science with archive from 1945 to present) New Engl J Med you get no hits. Truncated New Engl* J* Med* returns 81.820 hits. Searching New England Journal of Medicine gives same result.

A search on either Social science & medicine or Social science medicine returns 10.568 hits. A search for Social science and medicine on the other hand returns 0 hits, as does the journal abbreviation Soc Sci Med. A truncated search on Soc* Sci* Med* returns most 11.662.

Also worth mentioning is that search #5 in the search history was not possible to delete. Though I tried both Firefox and Internet Explorer.

Regarding the truncated search with 11.662 hits, a random check on result page 1000 shows the title: SOCIAL SCIENCE & MEDICINE PART C-MEDICAL ECONOMICS:

A search for Social Science & Medicine combined with author Kriedel as in the previous example returns 0 hits. The part A-F is not retrieved when searching Social Science & Medicine.
A search on abbreviated Lancet Neurol or BMj returns 0 hits. Truncated Lancet Neurol* or full title Lancet Neurology as British Medical Journal returns all hits.

Using the Full source titles list and browsing Social Science Medicine returns a list of all parts A-F to the journal.

The conclusion is that searching abbreviated journal titles is impossible in Web of Science and far from perfect in Scopus, but in general it’s more precise to search journal titles in Web of Science than Scopus as long as there are no subparts of the journal. Then you have to truncate or browse in Full source.

A search in Scopus on the journal abbreviation New Engl J Med in Scopus gives 3 hits.

By combining the full journal title New England journal of medicine with the title of the first article gives a satisfying hit:

By combining the full journal title with the title of the second hit gives this unsatisfying hit:

By combining the full journal title with the title of the third hit gives this strange hit:

Everything is satisfying besides the publication year which differs between this hit, with publication year 1975, and the other hit in the first search, with publication year 1974. According to Pubmed it should be 1974.
Searching the full title New England journal of medicine gives both New England Journal of Medicine and New England Journal of medicine with a lower case m in medicine. Trying to truncate New Engl* J* Med* gives all hits, even the three records with abbreviations.

Searching the Social Science & Medicine journal with the following 3 variants gives 9.322 records on all three variants: Social Science Medicine, Social Science and Medicine or Social Science & Medicine.

Searching the journal abbreviation Soc Sci Med gives 7 records. All seven records are not possible to retrieve searching the article titles combined with full source title: Social Science and Medicine. Trying to truncate soc* sci* med* returns 8,557 records on the spelling social science and medicine, and also unrelated titles because the journal subtitle is also indexed. Together 11,833 records. Here you also retrieve variants as Soc Sci Med and Part A-F.

Testing some other titles gives the following results. Searching British Medical Journal gives 78,067 hits, but searching the abbreviation BMJ gives 13,231 hits.

Searching Lancet Neurol gives 88 hits and searching Lancet neurology gives 730 hits. Truncating the abbreviated title Lancet Neurol* gives 818 hits and that includes both the abbreviated records and Lancet Neurology and Lancet. Neurology.

So what’s the conclusion with Scopus: search both the abbreviation and the full title. Truncating the journal abbreviation gives you all titles including the abbreviated journal title records and sometimes irrelevant journal titles.

But what happens if you go to Sources button in Scopus and browse journals? Searching New England Journal of Medicine returns a list of 48887 records.

Compared to 50181 records searching in source title with truncated abbreviation New Engl* J* Med*. The 48887 records are all records on New England Journal of Medicine without the variants New England Journal of medicine and New Engl J Med.

Browsing Social Science and Medicine via Sources gives the same problem. Just the 8557 records and not the variants. Lancet Neurology gives 730 records and British Medical Journal 71322 records.

The conclusion is: If you want to be sure to receive all articles for a journal don’t use the browsing option through Sources-button in Scopus.

We made a freetext search on headache in Pubmed with limits to entrez date 60 days which returned 442 records. We checked the availability of the PubMed records (in descending order) in Google Scholar until we reached the possible breakpoint and then checked the EDAT which is the date and time when the record was added to PubMed.

Google Scholar has indexed 1 of 5 articles published 2005/12/29 09:00 GMT-8. The screenshot below shows both the Google Scholar reference and a clipping from part of the PubMed reference:

From 2005/12/28 09:00 and descending one following day everything was indexed, with the exception of one article.

PMID 16375021 from 2005/12/27 doesn’t exist in Google Scholar, but the rest of the articles from the same date exist.

From 2005/12/31 09:00 and ascending at least one following date nothing was indexed by Google Scholar (2005/12/30 had no records in this search).

This means thet Google Scholar, at least in this single test 2006-02-01 11:00 GMT+1, has an update gap of more than one month. I did a similar undocumented test in 2005-10-07 which showed a latest update in Google Scholar 2005-08-17. To discover if the updates of PubMed via Google Scholar is regular or unregular requires regular tests during a longer time period.