Figure 1.
An overview of the network that consists of 1,000 researchers and 40 research topics reconstructed in this study.
Each link is weighted by the average number of web search hits for a search query of (research topic) + (researcher's name) + “research" (left), or (researcher's name 1) + (researcher's name 2) + “research" (right).
Figure 2.
An example of “visibility boost" calculation.
This figure shows how to calculate the visibility boost by research topic “network" for the researcher A in the middle (blue).
Figure 3.
Reconstructed network of 1,000 researchers.
(A) Visualized network. Each node represents a researcher. (B) Distribution of total link weights of nodes, plotted as a complementary cumulative distribution function (CCDF).
Figure 4.
(A) Comparison of the numbers of nodes (researchers) that appeared at least once in 40 times of selection trials. Yellow: Top 20% nodes selected based on their original search hit counts for each of the 40 research topics. Red: Top 20% nodes selected based on their visibility boosts by each of the 40 research topics. Blue: Random selection of 20% nodes repeated 40 times. (B) Smoothed histograms of average shortest path lengths among the selected 20% nodes in the researchers' network (N = 40 for each histogram; each sample point corresponds to one measurement of average shortest path length among the selected 20% nodes). To calculate path lengths, the reciprocals of link weights were used as edge distances. The average shortest path lengths among the top 20% nodes selected based on their visibility boosts were significantly smaller than random counterparts (p<0.05 by standard t-test), showing that researchers who share high visibility boosts by the same topic tended to come closer to each other in the network.
Figure 5.
Correlations between the visibility boost of each research topic and a researcher's overall popularity (total topic hit, A) and individual-level interdisciplinarity (topic hit entropy, B).
Upward or downward moves of topics from A to B by 20 or more places in the ranking are indicated by solid and dashed arrows, respectively.
Figure 6.
Correlation between each research topic and a researcher's social-level interdisciplinarity.
(A) Effective degree. (B) Total normalized incoming link weights. (C) Betweenness centrality. (D) Closeness centrality. Network-related topics are highlighted.
Figure 7.
Two dimensional maps summarizing each research topic's correlations with a researcher's individual-level interdisciplinarity (vertical, topic hit entropy) and social-level interdisciplinarity (horizontal, three centrality measurements).