Cutting Edge: A network approach to mixing delegates at meetings
- Fondazione Edmund Mach, Italy
- U-Hopper, Italy
- University of Cambridge, United Kingdom
- Waseda University, Japan
- King’s College London, United Kingdom
Figures
Figure 1
Data from initial survey sent to delegates.
Prior to the meeting, we asked each delegate to choose, from a predetermined list, which methods they were familiar with, and which methods they wanted to learn more about. (A) Methods known by delegate. Each row represents a method and each column represents an individual delegate (names omitted for reasons of confidentiality). Delegates were most familiar with microscopy (both quantitative and advanced), cells (both cancer and stem) and mathematical modelling. This information was used to calculate the knowledge distance between delegates in Figure 3. (B) Methods that delegates wanted to learn more about. The most popular methods were biophysical methods and microfluidic devices.
Figure 2
Speed dates increase network density.
(A) The collaboration network before the meeting: some delegates already knew 10 or more other delegates, whereas others knew just one or two. (B) After the first round of speed dates, 20 new connections (shown in red) had been added to the network. (C, D) The network after three (C) and five (D) rounds of speed dates; α = 0.9.
Figure 3
The distance between delegates.
(A) Matrix showing the inverse of the ‘acquaintance distance’ between all pairs of delegates, who are arranged horizontally and vertically as in Figure 1. (B) Matrix showing the ‘knowledge similarity’ between all pairs of delegates: similarity increases with the number of methods known in common by both candidates, and decreases with the number of methods that are only known by one. The inverse acquaintance distance (A) and the knowledge similarity (B) have both been normalized to have zero mean and unit variance so that they can be added together, suitably weighted by α. (C–F) The sum of inverse acquaintance distance and the knowledge similarity for pairs of delegates for different values of α. Colder colours such as blue indicate low similarity and high distance (see colour bar); hotter colours such as red and brown indicate the reverse; white squares not on the diagonal represent delegates who have previously collaborated. The similarity measure that we used (the Rogers and Tanimoto similarity measure) penalized pairs with low overlap, which led to the pronounced vertical and horizontal blue lines observed for the delegate familiar with 26 methods in B.
Figure 4
Increasing α increases new knowledge gained.
(A and B) Average shortest path and new knowledge gained (i.e., the total number of new methods delegates were told about in a given round) vs number of rounds of speed dating for four different values of α. The shortest average path (A) fell quickly with the number of rounds of speed dating, almost irrespective of the value of α. However, new knowledge gained (B) increased with the value of α: this is to be expected because higher values of α mean that delegates are more likely to be paired with someone who can tell them about new methods.
Download links
A two-part list of links to download the article, or parts of the article, in various formats.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
Cutting Edge: A network approach to mixing delegates at meetings
eLife 3:e02273.
https://doi.org/10.7554/eLife.02273
