Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Quantifying the role of supernatural entities and the effect of missing data in Irish sagas

P. MacCarron

Abstract

For over a decade, complex networks have been applied to mythological texts in order to quantitatively compare them. This has allowed us to identify similarities between texts in different cultures, as well as to quantify the significance of some heroic characters. Analysing a full mythology of a culture requires gathering data from many individual myths which is time consuming and often impractical. In this work, we attempt to bypass this by analysing the network of characters in a dictionary of mythological characters. We show that the top characters identified by different centrality measures are consistent with central figures in the Irish sagas. Although much of Irish mythology has been lost, we demonstrate that these most central characters are highly robust to a large random removal of edges.

Quantifying the role of supernatural entities and the effect of missing data in Irish sagas

Abstract

For over a decade, complex networks have been applied to mythological texts in order to quantitatively compare them. This has allowed us to identify similarities between texts in different cultures, as well as to quantify the significance of some heroic characters. Analysing a full mythology of a culture requires gathering data from many individual myths which is time consuming and often impractical. In this work, we attempt to bypass this by analysing the network of characters in a dictionary of mythological characters. We show that the top characters identified by different centrality measures are consistent with central figures in the Irish sagas. Although much of Irish mythology has been lost, we demonstrate that these most central characters are highly robust to a large random removal of edges.
Paper Structure (7 sections, 3 figures, 1 table)

This paper contains 7 sections, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Data and methods
  3. Results
  4. Core network properties
  5. Communities
  6. Simulations
  7. Conclusions

Figures (3)

  • Figure 1: (Colour online) The directed network with 3 communities fitted using the Girvan-Newman community-detection algorithm. The ten characters with the highest betweenness are named. Of these, four are supernatural characters, two are kings, two are leaders of the settlers of Ireland and the others are the two major heroes.
  • Figure 2: (Colour online) Panel (a) shows the complementary cumulative degree distribution $P_k$ for the in-degree (circles) with a fitted truncated power law, and out-degree (squares) with a fitted log-normal distribution. Panel (b) shows the probability mass function for the shortest path length $p_\ell$ with a fitted Poisson distribution.
  • Figure 3: (Colour online) Panel (a) displays the Jaccard index for the top 5, 10 and 25 nodes ranked by betweenness $J_b$, here we see even removing 500 (almost 25%) of the edges, the Jaccard stays above 0.5 showing that more than half the characters in the top 25 do not change. Panel (b) shows the Jaccard index when removing characters strategically is not as robust.