Chem 7950 Keyword Searching in SciFinder Scholar

From SubjectGuides

Recommendation: deselect Medline unless you want to search Medline.

[edit] The Basics

Steps in a SciFinder Scholar Keyword Search

1. Enter keywords => Your knowledge of chemistry and SciFinder Scholar impacts this!
2. SciFinder search algorithm identifies "concepts"
3. SciFinder constructs a pick list of results based on "concepts" and syntax
4. Delect one, or more, results from the pick list.

Reminder: A keyword search searches a SciFinder Scholar record in the ...

• Title of the article
• Abstract – all words in the abstract are searched.
• Index Terms and their modifiers (except for Role).
• CAS Registry Numbers and their modifiers (except for Role)
• Supplementary Terms

But not in the full text of the article!!!

[edit] Concepts

The SciFinder Scholar search algorithm identifies "concepts." What is a concept? That depends on your search. Really, it depends on your search! It can be a single word or a phrase. Here are some examples.

One concept

gold

nanoparticle

gold nanoparticle

density functional theory

gold nanoparticle density functional theory

cyclic voltammetry

gold nanoparticle cyclic voltammetry

electron transfer gold

Two concepts

gold nanoparticles electron transfer

gold electron transfer nanoparticle => Very odd concepts identified!!!

Singular and plural are usually lumped together – but not always!

Lumped together:

enzyme (enzymes)

polymer (polymers)

sulfide (sulfides)

Not lumped together:

mercaptan (mercaptans)

Automatic truncation is often applied.

• oxidation finds oxidant, oxidative, oxidized, and oxidizer by truncating to oxid*
• products finds production, produce, and produced by truncating to prod*
• motor finds motion, mother, and motif by truncating to mot*

Secret dictionary connects some synonyms into the same concept

dairy finds milk

cancer finds neoplasm

Antabuse finds 97-77-8 and N,N,N',N'-tetraethyl-thioperoxydicarbonic diamide

• The most valuable part of the "secret dictionary" are synonyms for a substance.
• The most important synonym is a CAS Registry Number.

Ignored words

Some words are ignored by SciFinder Scholar. These include:

articles (a, an, the)

words that have little meaning (effect, results, etc.)

By looking at the results of a search, you can see how SciFinder Scholar treats a word.

Abbreviations

Chemical Abstracts (aka SciFinder Scholar) has a long history of using abbreviations starting back in the print era when space on the printed page was valuable. Chemical Abstracts continues to use abbreviations. Knowing about CAS Abbreviations is useful for keyword searches and to correctly read an abstract in a CAS record. Sometimes the CAS abbreviation is a better search term than the word or phrase it represents. For example, LCAO is a better search term than is linear combination of atomic orbitals. In contrast, HOMO and highest occupied molecular orbital are treated identically by SciFinder Scholar.

[edit] Concepts and Syntax

A pick list of results will be constructed based on concepts (as identified by SciFinder Scholar) and syntax (how you arrange the words in the search). It is worth making what may appear to be small changes in a search to find out what words or phrases work best. This is especially important when you need to do a comprehensive search or when you don't know the terminology used by experts in a research area. Consider combining similar searches together to get a larger and more comprehensive list of references.

[edit] Three Simple Rules

• Best terms
• "Parenthetical OR"
• Prepositions

[edit] The Best Terms

The first rule is deceptively simple. Experiment with different words to find what works best. It may take a couple of searches (or even weeks or months) to find the best terms. For example, CAS Registry Number are usually better search terms than common names (the major exception is a recent reference that has not yet been fully indexed and therefore lacks CAD Registry Numbers). Idnetifying the best terms also depends on your chemical knowledge -- as it improves in an area of chemistry, often so do your searches.

[edit] The "Parenthetical OR"

A  (B)  =  A   OR   B , especially when there are other search concepts.

In SciFinder Scholar, A (B) is much more dependable than A   OR   B

Examples:

mercaptan (thiol)

mercaptan (thiol, sulfhydryl)

metal hydrides (transition metal hydrides) => Does this expand the search? Why not?

Boolean Operators

Do not rely on Boolean operators in SciFinder! Some work partially, most do not.

What Boolean operators work?

NOT

Phenol NOT 108-95-2 => Finds unindexed articles and wherever the word phenol shows up, such as 2,6-dibromo-4-methyl phenol

What Boolean operators do not work or only work in the most trivial case?

AND

The exception is the trivial case: A   AND   B

palladium AND catalyst

OR

Exception is the trivial case: A   OR   B

phenols OR phenolic resins

Parentheses => They do not work as Boolean operators in SciFinder Scholar. Instead, they are treated as a "Parenthetical OR" operatore (see above).

[edit] Prepositions

• Increases the relevance of search results.
• Cost you nothing ... and can find better search results!
• "AND" is much weaker than a preposition.

Compare these three searches:

hydrogen storage metal hydrides

hydrogen storage with metal hydrides

hydrogen storage and metal hydrides

Now compare these two searches:

hydrogen storage with metal hydrides with low pressure

hydrogen storage with metal hydrides with low pressure (atmospheric pressure)

Can the efficiency of the last search be duplicated using "AND"s? No!!!

Prepositions invoke the "closely associated" operator, which is a powerful tool in SciFinder Scholar.

The search:   hydrogen storage with metal hydrides

will present you with a pick list with five choices. One of these choices is

• 3707 references were found containing the two concepts "hydrogen storage" and "metal hydrides" closely associated with one another (as of 11-4-09).

[edit] "Closely associated"

"Closely associated with one another" is a proximity operator. This means that in a SciFinder Scholar record that the two concepts are found together in one of five ways. They are both found in either:

1. The title of the article
2. Same sentence in the abstract
3. Same Index term (including its modifier)
4. Same CAS Registry number (including its modifier)
5. The list of Supplementary Terms

• There is no guarantee that references with both concepts "closely associated" will be the best references, but it is likely that they are.
• Do not ignore the results in which both concepts are "present anywhere". The reference(s) that you need may be in that "present anywhere" results but not in the "closely associated" results. This is especially true for references that are so new that they have not been indexed by SciFinder Scholar (they only have a title and an abstract ... no index terms, CAS Registry Numbers, or supplementary terms).
• "As entered" is a strange beast. It usually means that means that the concepts are separated by a space or the word "a". It can be used to force SciFinder to combine two concepts together, which can simplify a search by shortening the pick list or make certain searches possible.

[edit] Analyze / Refine / Categorize

[edit] Analyze

• Best if you want to know more about your search results.
• Can be very slow.
• Most useful => Index terms, Authors, Company/Organizations, Journal Title, CA Section, Year.
• Analyze by Year and sort the results "alphabetically" to see how much research has been done on a topic as a function of time. Some topics are on the rise (gold nanoparticles) while other topics have have waxed and waned over the years (Wolff reduction, Clemmensen reduction).

[edit] Refine

• Best if you know how you want to limit your search.
• Fast.
• Most useful => Publication Type for reviews.
• The keyword limit becomes an "AND" search. It is usually better to revise your search instead and use a preposition with the new concept.

[edit] Categorize

• Best if you want to see how Index terms and CAS Registry Numbers fit into categories
• This is a highly recommended "Discovery Tool". Use it to learn more about the search results.
• Fairly fast.
• Limited to less than 5,000 references.
• Generates a pick list of index terms & CAS Registry numbers.
• Especially useful to find better keywords to use in SciFinder Scholar.

[edit] Other useful features

[edit] Combine Searches

Use to combine a current search with a saved search. This is a new feature available in SciFinder Scholar 2007, but not in earlier versions. See Analyze by CA Section for an example of how to combine two searches (see below).

Use to construct comprehensive searches, such as ...

A   OR   B   OR   C  (where A, B, and C are saved searches)

[edit] Analyze by CA Section

Imagine that you want to find articles that use carbon nanotubes in analytical chemistry. By "use" I mean analytical techniques that use carbon nanotubes as part of the analytical method or procedure. Try to come up with a search that does that. Most of what you find is either analysis of carbon nanotubes or the detection of carbon nanotubes. The answer lies in a CA Section filter.

First check on the need for plural ...

Carbon nanotube (carbon nanotubes) => Both find the same "concept", therefore you can drop the singular form of the phrase and add search term for analysis.

Next, search ...

carbon nanotubes with analysis

• Select the reference set that has all three concepts "present anywhere". This finds about 13,000 references. How do these look? Are they on topic? Some, but many are not.
• You can improve this search by an Analyze feature. Select CA Section and sort alphabetically. Choose Biochemical Methods. This CA Section is just like a journal called "Biochemical Methods" (except that it contains abstracts and not articles). Also choose three other CA Sections that are relevant to your search: Inorganic Analytical Chemistry; Organic Analytical Chemistry; and Pharmaceutical Analysis. To do a really thorough search on the use of carbon nanotubes in analytical chemistry, you need to include all four these. On the other hand, you may only be interested in their application to a particular area of chemistry -- for example just biochemical methods.]
• Get the references in the four CA Sections. This should be roughly 2,300 references. How do these look? Are they on topic? Yes, they are -- just about every reference is right on target.
• Save this set as a SciFinder Saved Set labeled as: CA Section Filter.

Click on the New Task button and repeat the search carbon nanotubes with analysis.

• Select the references in which both are "closely related". Combine these references with the saved search by choosing the saved SciFinder Saved Answers file "CA Section Filter". Click on Open.
• Choose "Remove saved set from current set". How do these look? Not so good ... even though all three search terms are "closely associated". Most of the references have to do with the analysis of carbon nanotubes or the detection of nanotubes instead of their use in analytical chemistry.

One last search ... (and as it turns out, the best search).

Carbon nanotubes

• This finds over 60,000 references. What about the use of nanotubes in analysis? Analyze by CA Section and select Biochemical Methods, Inorganic Analytical Chemistry, Organic Analytical Chemistry, and Pharmaceutical Analysis." Be patient. This will take much longer than the previous searches because there are so many references to analyze (about 3-5 minutes, depending on how busy SciFinder is). However, it is worth the wait. There is no way to duplicate this terrific "CA Section" aided search. It finds over 3,000 references or about 1 in every 20 references from the initial search. How many of these are not in the previous search? Use the Combine feature to compare this search to the saved SciFinder Saved Answers "CA Section Filter. Choose "Remove saved set from current set". How do these look? Pretty good. This is a good example that of the value of a broad search that is carefully restricted (via CA Section). The broader search found over 700 more results, most of which are right on target. I know of no other way to so quickly and thoroughly limit the 60,000 references from a "carbon nanotubes" search to just those on the use of carbon nanotubes in analytical chemistry.

[edit] Explore by ...

• Author name => Try Schneller, Stu => Control A selects all names.
• Organization = Auburn works well, but most organization searches do no. Compare to SUNY Buffalo vs. University at Buffalo, which are different names used for the same university.
• Chemical structure (very useful)
• Molecular formula (sometimes useful)
• Explore reactions (same thing as chemical structure!)

[edit] Locate ...

• Bibliographic Information -- i.e., look up reference
• Document Identifier -- occasionally useful
• Substance Identifier -- very useful for common names (i.e., tryptophan)

[edit] What to do with results?

After selecting results, you can ...

• Print
• Save in text or rich text format
• Save in tagged format for reference software (EndNote, BilioTex, etc.)
• Save as SciFinder Saved Answers => Use to combine searches.

Created by Bob Buchanan © 2009

Librarian for Veterinary Medicine and Chemistry

Auburn University Libraries