LIS 601: Week8
Key Concepts concerning Citation Indexing

General Concepts

1. Writers, in general, refer to (cite) works that fall within the subject area of their expositions. For example, English professors refer to the works of Northrop Frye, while Physics professors refer to the works of Albert Einstein.
2. By examining the similarities in citation usage it is thought that groups of similar works (i.e. works about the same thing) can be brought together (ideally without human intervention).
3. Clustering based upon citation similarities can be used to determine if new fields are emerging. This could indicate the need for new indexing terms, classifications, and/or new publications. This is the idea behind Garfield's Research Fronts.
4. Citations can be used to search for similar works forward through time.
5. Citation patterns can be analysed to determine the importance of a work (the impact) as determined by other scholars.
6. Citation lists are put together by the experts in the field; thus, whatever groupings are created using citation analysis methods are considered to be an artefact of the experts in the field in question and not those of an arbitrary indexer, taxonomist or librarian.

Some Reservations

Using citations to determined subject groupings can be problematic be citation creation is not consistent. Some problems include:

• invisible (or assumed) references (i.e., classic works in the field become "common wisdom" and are not cited)
• self-citation
• fraternal citation (citing of colleagues)
• reference list padding
• inconsistencies in citation styles (i.e. spelling variants, incomplete citations, etc.: hard on automated systems).
• missing citations

Bibliographic coupling

Two (or more) works are said to be bibliographically coupled if they have references to one (or more) other works in common. The strength of the coupling is said to increase as the number of common references increases.

Example: (see graphic)

Work A cites Works U, X and Z.
Work B cites Works U, Y and Z.
Work C cites Works T, U and W

Bibliographic coupling strength (AB) = 2.
Bibliographic coupling strength (AC) = 1.
Bibliographic coupling strength (BC) = 1.

Thus, Works A and B are most likely to be about the same subject.

Comments:

• time invariant (i.e., once coupled the coupling strength never changes)
• coupling strengths can be normalized to take into account the variations in the lengths of reference lists
 

Co-citation links

Example: (see graphics)

Work A cites Works U, X and Z.
Work B cites Works U, Y and Z.
Work C cites Works T, U and W.

Co-citation link strength (UZ) = 2
Co-citation link strength (UX) = 1
Co-citation link strength (UT) = 1
Co-citation link strength (XZ) = 1
(etc.)

Thus, Works U and Z are most likely to be about the same subject.

Comments:

• time variant (i.e., co-citation link strengths vary over time):
•   good for monitoring trends
• link strengths can be normalized to take into account the number of times that the works are not cited together