Figure 1.
Karate club graph with overlapping communities of two nodes (red and violet).
Figure 2.
Karate club graph with overlapping node communities induced by three hard link clusters.
Figure 3.
Fuzzification of hard clusters.
Karate club graph with overlapping communities from two hard clusters.
Figure 4.
Three topics and their overlaps in a network of 492 bibliographically coupled papers. Topics assigned manually to papers by inspection of their keywords, titles and abstracts. The nodes' colours correspond to four sets: (i) green to -index, (ii) blue to bibliometrics without
-index and without webometrics, (iii) red to webometrics without bibliometrics, and (iv) violet to the overlap of webometrics and bibliometrics. Transparent nodes are papers dealing with other information-science topics, mainly with information retrieval and information behaviour.
Figure 5.
Stability over size of all MONC branch communities.
Stable communities corresponding to our three topics in information-science papers 2008 are marked: bibliometrics (blue), webometrics (red), -index (green).
Figure 6.
Plots of MONC communities of the three topics in information-science papers 2008: -index, bibliometrics, webometrics.
Coloured lines mark corresponding thresholds.
Figure 7.
Stability over size of all 5004 HLC branch communities.
Stable communities corresponding to our three topics in information-science papers 2008 are marked: bibliometrics as blue, webometrics as red, and -index as green point, respectively.
Figure 8.
Stability over size of branch communities.
Stable Ward clusters corresponding to our three topics in information-science papers 2008 are marked: bibliometrics as blue, webometrics as red, and -index as green point, respectively.
Figure 9.
Membership distribution of -index topic in information-science papers 2008 determined by the fuzzification algorithm.
The red step curve represents the initial hard cluster, the violet curve the members after fitness maximisation, and the green curve the grades of membership in the final fuzzy community.
Table 1.
Topic matches by algorithms.
Figure 10.
The -index communities constructed by the three algorithms in information-science papers 2008.
Saturation of points correlates with membership grade. Colours of circles denote manually determined topics (green to -index, blue to bibliometrics without
-index and without webometrics, red to webometrics without bibliometrics, and violet to the overlap of webometrics and bibliometrics).
Table 2.
Community matching between algorithms.
Table 3.
Fuzzy of communities.
Table 4.
Values of and
of HLC communities calculated in the bipartite graph of papers and cited sources.