We are just completing a methodological systematic review for the NHS
HTA programme which has depended quite heavily on pearlgrowing so I
can give you a few pointers.
There are a number of types of pearl growing although there is some
confusion of terminology - strictly speaking "citation pearl growing" is
(1) where you use the INDEX TERMS terms from a retrieved citation to help
you identify additional articles on the same or related subjects.
E.g. you do a search on MEDLINE find a relevant article and then see
how it is indexed and then use those index terms to construct a
follow-up search strategy.
Two variations however are
(2) finding a relevant article or articles and then seeing which
articles HAVE CITED THIS, seeing which of these are relevant and then
seeing which articles cited these in turn. Although strictly speaking
this is citation searching it uses a similar process of starting from
"pearls" so could be loosely labelled "prospective citation pearl growing"
(3) again finding a relevant article seeing which
articles HAVE BEEN CITED BY THIS and then using these cited references as the
pearls. Then doing a citation search on each of these.This might be
loosely termed "retrospective citation pearl growing".
The text which introduced me to citation pearl growing is Hartley RJ
et al. Online searching : Principles and practice Bowker Saur 1990.
Side note: A key interest of mine is how process 2 equates to
qualitative research with its notions of key informants and data
saturation. I am interested in how many "generations" of citations
one would have to pursue before one would reach data saturation. We
pulled out for logistic reasons after two rounds. 30-> 200 ->1000+.
Another analogy is the Delphi Technique but here your experts are
article writers rather than opinion givers. IMHO it also seems that the
concept of comprehensive searching introduced by Cochrane
Collaboration is irrelevant in this context where the aim is purposive sampling
and data saturation rather than identification of the whole universe
of studies.
Final point: For our methodological topic our yield from traditional
searching was <1% (i.e. 1 relevant article for every 100 abstracts
reviewed) whereas citation pearl growing yielded about two-thirds of
the total relevant items retrieved.
>From Draft report:
Citation Pearl Growing
Pearl growing is an "application of the method used for searching
citation indexes, in which the index terms accompanying a located
citation are used to find a new set of documents"(1) . It is
particularly appropriate for identifying a corpus of knowledge where
there are known deficiencies in indexing or terminology. The efficacy
of citation retrieval within health and related subjects has been
established in a field study that found that citation searching added
an average of 24% recall to traditional subject retrieval (2) .
Pearl growing is analogous to the "key informant technique" in
qualitative research; key documents are identified and then references
citing these documents are retrieved and reviewed for relevance. Its
limitations are similar to the key informant technique in that it
relies on the prior selection of a sufficiently diffuse sample of
records in initiating the process. It has been demonstrated with
regard to subject searching that the more cited references used for a
citation search, the better the performance, in terms of retrieving
more relevant documents, up to a point of diminishing returns(3) .
Citations to a sample of approximately thirty known relevant
references (n=34) were selected for searching using the Science
Citation and Social Science Citation Indexes. These items were
selected on the basis of centrality to the review topic (in terms of
subject content and relevant keywords) and no attempt was made to
evaluate either the extent of the contribution of a particular article
or whether it belonged to a specific school of thought. Formal
evaluation of the thirty "pearl" articles was undertaken only after
this stage of the searching had been completed. Within this context it
should be recognised that the term "pearl" is not intended to denote
any intrinsic quality of the articles themselves; to complete the
analogy, at this point in the process they equally could be annoying
pieces of grit!
1. TI: The effect of the searching environment on search performance.
AU: Buntrock-Robert-E SO: Online-. 3 (4) Oct 79, 10-13. illus. 3 refs.
2. TI: Term and citation retrieval: a field study.
AU: Pao-M-L
SO: Information-Processing-and-Management. 29 (1) Jan/Feb 93,
p.95-112.
3. TI: The performance of cited references as an approach to
information retrieval.
AU: Yoon-L-L
SO: Journal-of-the-American-Society-for-Information-Science. 45 (5)
Jun 94, p.287-99.
> Does anyone have any information about a method of identifying relevant
> articles for a systematic review called "pearl growing"?
>
> I believe that it is based on finding some papers that are highly relevant
> (pearls) and then using citation indexes to identify subsequent
> publications that have cited these papers/authors.
>
> Also, does anyone know of an on-line citation index which can be accessed
> free of charge?
>
> Thank you in advance for any information/help on this.
>
> Dianne O'Connell
>
> ==========================================
> Dianne O'Connell PhD
> Senior Brawn Research Fellow
> Discipline of Clinical Pharmacology
> Faculty of Medicine and Health Sciences
> The University of Newcastle
> AUSTRALIA
>
> Mail: Level 5, New Med 2
> Newcastle Mater Hospital
> Waratah 2298
> AUSTRALIA
>
> Phone: 61-2-49 211 293
> FAX: 61-2-49 602 088
> E-mail: [log in to unmask]
>
> ==========================================
>
Andrew Booth BA MSc Dip Lib ALA
Director of Information Resources
School of Health & Related Research (ScHARR)
Regent Court
30 Regent Street
SHEFFIELD
S1 4DA
Tel: 0114 222 5420 or 5214 Fax: 0114 272 4095
The author of Netting the Evidence:
http://www.shef.ac.uk/~scharr/ir/netting.html
and Trawling the Net:
http://www.shef.ac.uk/~scharr/ir/trawling.html
E-mail: [log in to unmask]
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|
