metric proceduressuch as cluster analysisand similarity measures(Topic B5),
and may use sequential searchesor hierarchical search trees.
A sequential search of a spectral library, for example, involves the compar-
ison of every part of a sample spectrum with library spectra, and is suitable only
for small libraries. A hierarchical search involves comparing groups (families) of
spectra having the same set of key features as the sample spectrum, enabling
large libraries classified in a tree-like structure to be searched very efficiently.
Searching spectral libraries may involve the use of inverted lists. These
consist of each characteristic absorption band or emission line along with a list
of corresponding numbered library spectra that include that particular band or
line. A list for the spectrum of an unknown analyte can then be rapidly checked
against the library lists. An example of part of an inverted list for an infrared
spectral library is shown in Figure 1. It includes spectrum No. 66 among those
listed with an absorbance band at 1220 cm−^1 and spectrum No. 105 among those
listed with an absorbance band at 2730 cm−^1.
Correlation coefficientsthat define the quality of a match (a perfect match
corresponds to a value of 1.000) between an unknown and a library spectrum or
structure) are used to compile a hit listthat places possible identities in
descending rank order. A hit list shows the five to ten most probable identities
of an unknown based on the selected search criteria, but it is limited by the
content of the library and may include unlikely or impossible identifications.H4 – Data enhancement and databases 335
4000350030002500200018001600140012001000800600400AbsorbanceWavenumber (cm–1)66
ID Library spectra
4000350030002500200018001600140012001000800600400AbsorbanceWavenumber (cm–1)105
Key Inverted list
4000 cm–12730 cm–1 12 ... 45 ... 105 ... 5171220 cm–1400 cm–1... 66 ... 309 ...Fig. 1. Part of an inverted list for an infrared spectral library.