Topics in French periodicals
In this view you can gain the firts impression of the topics in the periodicals through a heat map, bar charts and word clouds of topics in the periodicals on the levels of individual periodicals, individual topics and the whole collection.
Use the navigation buttons on the right or scroll to the visualisations.
Download all result files
See also:
Topic prevalence over timeTopic network
Analysis results of other collections
Heat map
After using topic modeling on segmented periodical issues, we calculated the mean probability values of topics for each periodical and represented the topics with their top three keywords. Use the bokeh tools positioned over the heat map to navigate the image or open it in a seperate window for a better view by clicking the blue arrow in the upper right corner. Take a look at the word clouds above for more insight in the topic keywords.
Word cloud
The first word cloud represents the most frequent words from all periodicals, generated after lemmatization and token reduction (excluding the 100 most frequent words and stop words).
In the second word cloud you can see the word probability per topic. Each topic consists of all lemmas (keywords) from all texts, with a higher or lower probability. How we interprete a topic depends on its top keywords, which are usually thematically related. Use the arrows to scroll through all topics.
Word probability per topic
Bar chart
While heat maps (as the one obove) are a good representation of the overall topic modeling results, bar charts enable a more focussed view on idividuall periodicals topics. Zoom in with your mouse scroll or click the toggle full page icon to better see the labels.
Probability of topics in a periodical
The same results can be seen from another perspective: Each bar chart below shows the probability of one topic in all periodicals.