Main page Tropes, Semantic Text Analysis - Online Reference Manual 
  info@semantic-knowledge.com 
Home | News | Reference | Support | Download | About 

Gephi GEXF exports in Tropes software

 

Since version 8.4, Tropes can generate GEXF XML export files. These files are reusable with Gephi, an open-source data analysis and visualization software.

You must download and install Gephi (gephi.org), before using this function

To export these files, open one or preferably several texts with Tropes, then use the [Tools][Exports GEXF] menu. Choose the output folder and press the [Export] button.

Three GEXF files are generated:
1 - TropesRelations (Relations, extracted from the current Scenario)
2 - TropesOntology (co-occurrences between documents and elements of your Scenario)
3 - TropesMatrix (co-occurrences between documents)

The first file is the only relevant output for a single document analysis.

If you checked the [Show results] option, Gephi is automatically started with the TropesRelations file, at end of the process.

With Gephi, you will be able to produce graphs like this:

Tropes with Gephi - Alice in Wonderland
This graph is issued from Alice in Wonderland text analysis, after applying a “Force Atlas”
layout in Gephi. Tropes’ Relations are extracted from a custom Scenario.

Unlike Tropes, Gephi displays all the data transmitted to it. If you use the default Scenarios with several documents, then the output results can exceed the maximum capacity of Gephi. It is therefore necessary to check the [Automatic reduction of large data structures] option to avoid saturating the software. When this option is selected, the reduction is carried out by gradually increasing the thresholds of frequency of occurrence and eventually filtering the link weights (<edge weight>).

Aftermath: During the reduction process, Tropes may discard some interesting variables. Preferably, reduce yourself manually the number of groups of your Scenario, by choosing the most relevant elements for your analysis. A finely tuned Scenario, focused on what you want to see, is still more methodologically acceptable.

Characteristics of the GEXF generated files:

GEXF valuesTropesOntologyTropesRelationsTropesMatrix
<node label>Elements of the Scenario
and Filenames
Elements of the ScenarioElements of the Scenario
<attvalue for="frequency">Frequency of occurrence
or mean frequency (files)
Frequency of occurrenceFrequency of occurrence
<attvalue for="domain">Label of the first level
in the Scenario
Label of the first
level in the Scenario
Label of the first
level in the Scenario
<attvalue for="level">Scenario depth levelScenario depth levelScenario depth level
<edge weight>Word occurrences,
in each document
RelationsNumber of documents
<viz:color>RandomRandomRandom
<viz:position>Frequency and Scenario
depth level
Frequency and Scenario
depth level
Frequency and Scenario
depth level
<viz:size>Related to FrequencyRelated to FrequencyRelated to Frequency

The TropesOntology file is an unusual mix of different types of nodes: document names and elements of the Scenario. The frequency specified for documents is the average frequency of occurrence of the elements of the Scenario - these values are necessary for the Gephi layout algorithms. You can change the values in the Gephi [Data Laboratory] tab, when selecting the #documents variable (in the right column).

Attributes <viz:color> visualization and <viz:position> are generated to facilitate the management of data in Gephi. They are not significant. The default colors (<viz:color>) are generated automatically for each branch of the Scenario, and unique to each subset of the ontology.

Gephi can export all the graphs in PDF format, see: AliceInWonderland-Tropes-Gexf-Sample1.pdf

To learn more about how Gephi, see: http://ww.gephi.org

 


First page Previous Next Last page

Copyright Acetic and Semantic Knowledge, all rights reserved
www.semantic-knowledge.com