TrajanNet

Extraction and analysis of a Trajan-related social network

• Copyright 2019-2020 Vincent Labatut

TrajanNet is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation. For source availability and license information see licence.txt

• Lab site: http://lia.univ-avignon.fr/
• GitHub repo: https://github.com/CompNet/TrajanNet
• Data: https://doi.org/10.6084/m9.figshare.13143104
• Contact: Vincent Labatut [email protected]

---

Description

This set of R scripts aims at analyzing a historical dataset describing the relationships between the Roman emperor Trajan and his entourage. It does the following:

1. Extracts various networks based on some tabular data containing individual and relational attributes.
2. Computes a number of statistics and generates the corresponding plots, for both the tables and graphs.
3. Performs a sequence analysis of certain chronological attributes.

Data

The raw dataset was manually elaborated by Gaëtane Vallet during her Master's thesis in ancient history. See her thesis (in French) for more information [V'18,V'20]. The files produced by the scripts (graphs, plots, tables...) can be obtained by executing them, but they are also directly available on Figshare.

Organization

Here are the folders composing the project:

• Folder data: contains the data used by the R scripts, as well as produced by them.
  • Folder tables: input data presented as the following CSV tables
    • trajan_attributes.csv: individual description of each historical character in the dataset.
    • trajan_careers.csv: careers of the character, described as sequences of positions.
    • trajan_positions.csv: list of professional positions a character can hold.
    • trajan_relations.csv: connections between the characters.
    • trajan_typical_careers.csv: ideal careers, described as sequences of positions.
  • Folder nets: networks procuded by the scripts, and the associated plots and tables.
    • Folder all: network containing all the types of links at once (multiplex signed network).
    • Folder family: network with only the family ties (uniplex signed network).
    • Folder friend: network with only the friendship ties (uniplex unsigned network).
    • Folder pro: network with only the professional ties (uniplex signed network).
    • Folder unknown: network with relationships whose exact nature is unknown (uniplex signed network).
    • Folder signed: signed network with no distinction between relationship types (signed collapsed multiplex network).
      • Folder na-as-positive*: network obtained by considering the links whose sign is unknown as positive ones.
      • Folder na-as-ignored*: network obtained by discarding the links whose sign is unknown.
      • Folder *-closure: closure of the signed network obtained using strong structural balance.
      • Folder *-closure-poly: closure of the signed network obtained using weak structural balance.
  • Folder attributes: descriptive results obtained for the individual attributes.
  • Folder sequences: descriptive results obtained for the individual attributes.
    • Folder withNAs: gaps in careers are explicitly represented as missing values.
    • Folder withoutNAs: gaps in careers are not represented at all.
• Folder src: contains the R source code.

Installation

You just need to install R and the required packages:

1. Install the R language
2. Download this project from GitHub and unzip.
3. Install the required packages:
   1. Open the R console.
   2. Set the current directory as the working directory, using setwd("<my directory>").
   3. Run the install script src/install.R.

Use

In order to extract the networks from the raw data, compute the statistics, and generate the plots:

1. Open the R console.
2. Set the current directory as the working directory, using setwd("<my directory>").
3. Run the main script src/main.R.

The scripts will produce a number of files in the subfolders of folder nets. They are grouped in subsubfolders, each one corresponding to a specific topological measure (degree, closeness, etc.).

The verification.R was used to check the consistency of the raw data. The rest of the scripts are just secondary functions called by main.R.

Dependencies

• igraph package: build and handle graphs.
• signnet: analysis of signed graphs.
• graphlayouts: plot graphs.
• ggraph: plot graphs.
• TraMineR: sequence analysis.
• SDMTools: misc.
• scales: color conversion.
• circlize: circos-type plots.
• plot.matrix: matrix plots.
• alluvial: alluvial diagrams.
• cluster: cluster analysis.
• dendextend: dendrogram-related features.

To-do List

• Signed nets:
  • Consider the evolution of the relationships
  • Synchronous closure, separate the different steps until complete graph
• Structural similarity between spaniards / the rest (and other attributes)
• Multiplex plot of the different types of links (didn't find an appropriate tool)

References

• [V'20] Vallet, G. Les princes et les hommes : carrières et réseaux dans l’entourage de Trajan de 98 à 118 après J.-C., Second part of the Master's thesis, Avignon University, Human and Social Sciences Faculty, History Department, Avignon, France.
• [V'18] Vallet, G. L'entourage de Trajan : étude prosopographique de l'entourage du prince de 98-117 apr. J.-C., First part of the Master's thesis, Avignon University, Human and Social Sciences Faculty, History Department, Avignon, France.