The graph below was created using the tool LDAvis and provides an interactive view of the main themes present in Ricoeur's documents in this decade. To the left of the visualization, the Intertopic Distance Map displays the similarity of topics based on how many words they have in common. The size of the circle representing each topic is proportional to the prevalence of the topic, with topic 1 being the most prevalent. Regarding the axes, they try to provide a bi-dimensional space showing how close the most prevalent words of each topic are. The idea is to give a sense of groups of closely related topics in contrast with some very different topics. To do this, the tool uses a technique called dimension reduction, as each topic is different from all others in multiple dimensions (basically, each word it contains). This specific dimension reduction algorithm is called Principal Component (PC) Analysis, hence the terms PC1 and PC2.
To the right, the Top-30 Most Salient Terms shows the most salient terms across all topics, with the blue bar indicating the overall term frequency. To view the Top-30 Most Salient Terms for a specific topic, click on a circle in the Intertopic Distance Map. The blue bar indicates the overall term frequency, while the red bar shows the estimated term frequency within the topic. Clicking on a term shifts the size of the topic circles to show the probability of that word in each topic.Â
"Words representing a given topic may be ranked high because they are globally frequent across a corpus. Relevancy score (λ (lambda)) helps prioritize terms that belong more exclusively to a given topic, making the topic more obvious." (Ruchirawat, 2020).Adjusting λ (lambda) at the top right changes the words shown in the Top-30 Most Relevant Terms bar chart to improve the interpretability of the topics. Lambda values closer to 1 rank the terms in order of decreasing topic-specific probability. Thus, terms may be less topic-specific with higher lambda values. Lambda values closer to 0 display rarer, topic-specific terms.Â