Topic network in french periodicals

This visualization of topics and periodicals is a force directed graph created using the algorithms Fruchterman Reingold and Force Atlas 2 and the modularity method from Blondel et al. (2008) with the analysis and visualization software Gephi 0.9.2.

The goal was to represent a network of topic modeling results (topics) and periodicals with their manually assigned keywords. Therefore, there are two node types: periodical nodes and topic nodes. The periodical nodes are pie charts showing in how many periodical issues a keyword was assigned to. The same keyword-color correlation was used for each periodical. To see the color legend and numerical data for the pie charts, select Pie charts data under the Menu button .

To make the pie charts more readable, only keywords that were assigned a minimum of 100 times in the whole Spectators corpus (which are 22 keywords) were included, due to the otherwise too large amount of colors. The size of a periodical node (pie chart) indicates whether the number of analyzed periodical issues from the topic modeling set is large or small in comparison to other periodicals. Note that numerous issues does not mean the same as a large amount of text, since some issues can be very long and others very short. Also, the number of the analyzed issues is not always the same as the whole amount of issues in a periodical in the Spectators edition, since issues without manually assigned keywords (often, for example, the tables of contents) were left out from the topic modeling analysis.

The size of the topic nodes indicates whether a topic has a high or low possibility in the analyzed set of texts. The same results are shown as heatmaps and bar charts in the section Topics in periodicals under Topic modeling results. The Topic modeling data is provided in the Download section under the Menu button (a ZIP file containing two TXT files), together with this visualization in PNG format and the Gephi data as a GraphML, but the topic keywords are also listed under Topic keywords in the Menu. Nodes with the same color belong to the same community. This means that the densities of the edges within these nodes are higher than from these nodes towards the rest of the network. But, since this is a small network where all topics occur in all periodicals to some degree, the weighted modularity of this network is low and the community structure is not perfectly clear. Nevertheless, it is possible to detect topics that often co occurred in certain periodicals.

Because this is a directed graph, the edges are round and in clockwise direction from the topic nodes to the periodical nodes. They are higher weighted (thicker) if the likelihood of a topic in a periodical is higher. The weight of a topic in a periodical is documented with the keywords in the Topic modeling data and Topic keywords.

See also: