As with any other field, bibliometrics has its own terminology.
Presented below, are some starting-points as well as a couple of basic terms that are good to know if your intentions are to follow the ongoing discussions about bibliometrics.
The data forming the basis of bibliometric analyses are scientific publications (or rather, their bibliographic representations) indexed in multidisciplinary or subject-specific databases. In some databases, besides traditional bibliographic data, such as author, title and abstract, publications' reference lists and author addresses are also presented. This facilitates an analysis of publications with respect to the number of obtained citations as well as, on the basis of the addresses, attributing these citations specific units. Citations are an essential component in the bibliometric methodology as they are used to indicate the impact of published documents which, in the case of evaluative analyses, are considered to be the measurable aspect of quality. In this context the following terms are essential:
- Unit of analysis. The object of a bibliometric study. Different aggregation levels may be of interest, such as research group, institution, university, or country. Units can also be defined in terms of subject areas or specific journals.
- Impact. Usually refers to a measure of the frequency of citations regarding an analyzed unit. Put more simply, the higher the frequency of citations, the greater the impact is ascribed to the unit being analyzed. However, impact is not equivalent with quality, but it is often used as a partial indicator for this term which is quite hard to define.
- Normalization. Since different fields have different publishing traditions the frequency of citations or publishing between, let us say, two research groups, can not be compared with each other without due consideration to this fact. Furthermore, the frequency of citations is dependent on what year a publication is published (citations are accumulated by time) and publication type. Normalization means that comparisons are made between objects of similar nature in order to obtain a normalized value that forms the basis of fair comparisons between given units, i.e. independent of time, publication type and field.
- Self-citation. By this is usually meant that the author cites his or her own work. In some studies this type of citations are not counted, although sometimes this variable is ignored if a sufficiently large number of publications is analyzed since the amount of self-citations tends to be relatively equally distributed between the different units. However, a self-citation can be defined in several ways, for example by the first author being the same person in the cited as well as in the citing publication.
- Citation window. When the number of citations a journal has received is calculated, either an open time span or a limited time span is used. Using the latter means that citation values are comparable over time, for instance, a three-year interval would show that, for an article published in 2005, citations are measured for the publishing year and 2006, 2007 and 2008. However, an open time interval is to recommend if the purpose is not to compare changes over time. In such cases, citations are measured from the publishing year to the year the study is made.
- Fractionalization. This means that the number of citations or publications are in relation to the number of authors. For a publication that is written by n researchers and cited x times, each author is attributed x/n citations. Example: two authors who have written an article, which in turn is cited by ten other articles, will be attributed five citations each (that is: 10/2=5). Fractionalization regarding the number of publications is done in a similar way. Fractionalizations may also be carried out for higher aggregation levels, such as institutions, universities or countries.
- Indicator. The result of bibliometric analyzes is one or several bibliometric indicators. The choice of words reflects the fact that bibliometric anzalyzes can only measure certain aspects of the complex and multidimensional research activity. Indicators can be constructed to measure a variety of different things: if the focus is on evaluation, quality indicators are used, while structural indicators are used in order to study different relationships between authors, publications or other analyzed units.