_lognostics maps



lognostics maps

lognostics home

lognostics tools


The methodology

This page outlines the procedure for generating the maps that you will find on the _lognostics maps pages.

First we identify the literature that we want to analyse. In these maps, the raw data is usually a list of papers taken from the VARGA database. Normal practice is to eliminate from this list bibliographies, monographs and theses, which tend to contain unusual citation patterns. This elimination leaves us with a set of research papers that cover the relevant topic or time span.

Next we make a list of all the authors cited in each paper. Each author is listed separately, and co-authors all receive the same weight. Authors citing themselves are not penalised. From this author list, we can construct a list of co-citations - i.e. a list which identifies pairs of authors who are cited in the same work.

After this, we count the number of times a co-citation appears in this list, and eliminate the co-citations which appear only rarely. This simplifies the displays generated by the next procedure. The best results seem to emerge when the maps are not too dense. We therefore normally set the threshold for inclusion so that about 100 authors appear as nodes in the maps. We also eliminate the weaker co-citiation links so that the resulting map contains about 1000 edges. For the very early work, where fewer sources are available, a lower threshold has been set. The threshold deatails are included with the notes for each map.

Following this preparatory work, the list of eligible co-citations is submitted to GEPHI (http://gephi.org). GEPHI performs a cluster analysis on the co-citation data and generates a map which shows the relationships between the clusters. Each cluster consists of a number of authors who are frequently cited alongside each other. The clusters represent "invisible colleges" in the research community - a group of people who share common research interests. The specific focus of each cluster can usually be established by identifying the cluster members who figure in the largest number of co-citations for that cluster.

Finally we note the salient features of each map, and interpret their significance. Each map is accompanied by a short account which notes the principle clusters it contains, and who make these clusters up. The notes suggest how the clusters might be interpreted, and identify the significant researchers in each cluster.

Some computer programs which facilitate the collection of co-citation data can be found on the _lognostics tools page. Look for the Bibliometric mapping tools.

(c) 2014 Paul Meara