Login to LibApps

CHE/MAE/MSE 593: Applied Project : Search Strategy

How to do a literature review for SEMTE applied projects.

This page contains ...

Precision vs Recall

There are two criteria by which a successful search may be judged: 

Precision
retrieving what you need
Recall
retrieving all you need


Precision and recall are inversely proportional.  A highly precise search will find few items (low recall) but most, if not all, of the results will be exactly what you need.  A search designed for high recall will retrieve much more of what you need but will also include more material that you do not (low precision).  

A literature review requires high recall results - that need for thoroughness involves more time, effort and different search strategies than the simple type of keyword searching used for precision.

Do you know what you're looking for?

Can you describe your information need  using one simple sentence or question? 

Without a clear idea of the project, you may not be able to determine which are the best resources to search, what terminology should be used in those resources, and if the results are appropriate and sufficient.    

 

Examples:

  • For a compressor cascade, what is the influence of free stream turbulence on air flow?
  • Can molecular modeling be more cost effective than lump modeling for heavy oil refining? 
  • Is solar-power feasible for running residential air-conditioning in desert climates?

PICO

If you're having difficulty getting your project described succinctly, try using a PICO chart to identify the concepts involved:

  • P is product or process,(population, problem, predicament)
  • I is the intervention or improvement
  • C is what you'll compare your intervention/improvment to, and
  • O is the outcome (or results of the comparison of I and C)

 

For a compressor cascade (P), what is the influence of free stream turbulence (I) on air flow (O)?   

Can molecular modeling (I) be more cost effective (O) than lump modeling (C) for heavy oil refining (P)? 

Is solar-power (I) feasible (O) for running residential air-conditioning in desert climates (P)?

Search Strategy

Searching within a resource is always a multi-part process:

 

A. Start with a P AND I search

  • If you retrieve too many items, narrow the search by restricting the words to the title field (not every database allows this)
  • Look for articles that are very close or exactly what you want - examine each one for the controlled vocabulary (subjects and descriptors) that the database assigned to them; also look to see if you can find synonyms
  • Look for articles that are not as closely related but still have parts of interest; again, examine each article for synonyms and for the controlled vocabulary

 

B. Expand the P AND I search to include synonyms

  • Redo your search, this time including the synonyms you discovered above
  • For "single box" search engines do several simple searches employing only one or two synonyms at a time; do not put all the synonyms for all the concepts into one search as this will confuse the engine's relevancy algorithm. 

 

C. Search using the database's controlled vocabulary for P AND I 

  • In some databases, the controlled vocabulary may be called the thesaurus, subjects or descriptors.
  • Some databases, such as Google Scholar, Library One Search and Web of Science do not use controlled vocabulary, so you'll not be able to to do this type of search.  In this databases, you can use controlled vocabulary from the other databases in your keyword/synonym searches.  


D. Where do the C (comparison) and O (outcome) come into the search strategy?

  • As you read the items you found, remember the C (comparison) and O (outcome) - this will help you decide which articles you can discard and which you should keep.
  • If the results you get from searching the P and I concepts are just too large to work with, try ANDing the "O" as another concept in the search.  Don't forget to find synonyms for this concept, too.

 

E. For the most important documents you found, use citation searching to find even more:

  • Go backwards in time by finding the references cited in the document
  • Go forwards in time by finding the articles that cite this document (using Web of Science and Google Scholar)

What's In A Name?

Scientists would rather share each other's underwear than use each other's nomenclature 

Quote From ...

Keith Yamamoto, UCSF, in ...

Dickman S (2003) Tough Mining. PLoS Biol 1(2): e48  doi: 10.1371/journal.pbio.0000048

Read, Refine, Re-search

Hours and Locations