Evidence Synthesis Methods Interest Group

Hi, Sarah!

I can't speak specifically to the Embase/Emtree issue, but I have an outstanding ticket submitted to Scopus a month ago asking how index terms are applied in the database. 

Scopus documentation says index terms are not pulled directly from the records in the original datasets (such as MEDLINE or, presumably, Embase), but that Scopus indexers apply terms themselves from existing thesauri. Excerpted from their 2022 Content Coverage Guide (section 3.3), attached: "Scopus manually adds index terms for 80% of the titles included in Scopus. These index terms are derived from thesauri that Elsevier owns or licenses and are added to improve search recall. A team of professional indexers assigns index terms to records according to the following controlled vocabularies: Ei Thesaurus ... Emtree medical terms ... MeSH ... [several others].... There is no limit to the number of index terms that Scopus can add to records. However, in the case of Emtree and MeSH terms (both terms are added to records where available), only the index terms that have a direct relation with the topic of the article are displayed and made searchable on Scopus in order to avoid retrieving irrelevant results. For Emtree, the index terms with a direct relation are the Major Focus and the mentioned index terms. For MeSH, the index terms with a direct relation are Major Topics and Minor Topics...."

A tip I read on Expert Searching last year from Krizia Tuand suggested searching TITLE-ABS() OR AUTHKEY() rather than TITLE-ABS-KEY() since the KEY() in the latter field combo includes INDEXTERMS. That way you can work around any potential indexing and just treat it as a plain keyword search approach, like [tiab] vs. [tw] in PubMed.

I hope this helps!

-Hilary

I ran into a similar issue last year with seemingly irrelevant Emtree terms being applied to articles, which exploded the results significantly (and made the search unmanageable).  From what I could tell at the time, it seemed like they would apply various neoplasm terms to any article that mentioned cancer in any capacity, even just in passing. Though I don't have anything to point to directly, from my experience they seem to do this frequently with conditions and drugs especially.

As a result, I ended up using their Major Focus option to limit the search back down to only those where cancer was a primary focus.  Since those Emtree terms showed up in Scopus as well, I did as Hilary suggested and only searched TITLE-ABS for that concept. <o:p></o:p>

During this process, I found the following two articles that looked at the effects on search sensitivity from using the Major Focus limit in Embase that were helpful.<o:p></o:p>

Searching Embase and MEDLINE by using only major descriptors or title and abstract fields: a prospective exploratory study

Pruning Emtree: Does Focusing Embase Subject Headings Impact Search Strategy Precision and Sensitivity?

I hope this is helpful!

Thanks!!

Anna

Hi, Sarah!

I can't speak specifically to the Embase/Emtree issue, but I have an outstanding ticket submitted to Scopus a month ago asking how index terms are applied in the database. 

Scopus documentation says index terms are not pulled directly from the records in the original datasets (such as MEDLINE or, presumably, Embase), but that Scopus indexers apply terms themselves from existing thesauri. Excerpted from their 2022 Content Coverage Guide (section 3.3), attached: "Scopus manually adds index terms for 80% of the titles included in Scopus. These index terms are derived from thesauri that Elsevier owns or licenses and are added to improve search recall. A team of professional indexers assigns index terms to records according to the following controlled vocabularies: Ei Thesaurus ... Emtree medical terms ... MeSH ... [several others].... There is no limit to the number of index terms that Scopus can add to records. However, in the case of Emtree and MeSH terms (both terms are added to records where available), only the index terms that have a direct relation with the topic of the article are displayed and made searchable on Scopus in order to avoid retrieving irrelevant results. For Emtree, the index terms with a direct relation are the Major Focus and the mentioned index terms. For MeSH, the index terms with a direct relation are Major Topics and Minor Topics...."

A tip I read on Expert Searching last year from Krizia Tuand suggested searching TITLE-ABS() OR AUTHKEY() rather than TITLE-ABS-KEY() since the KEY() in the latter field combo includes INDEXTERMS. That way you can work around any potential indexing and just treat it as a plain keyword search approach, like [tiab] vs. [tw] in PubMed.

I hope this helps!

-Hilary