Machine Learning Research Abstracts
Interactive thematic analysis of six extracted themes, their prevalence, and the structure of pairwise co-occurrence across the abstract corpus.
Themes identified
Largest theme share
Top 3 theme share
Strongest pairwise score
Theme prevalence ranking
Ranked view of prevalence from the extracted theme summary.
Prevalence landscape treemap
Area encodes theme share; hover exposes the definition and top words for each theme.
Theme co-occurrence matrix
Diagonal cells are self-values; off-diagonal cells show pairwise structure from the co-occurrence matrix.
Strongest pairwise relations
The most pronounced off-diagonal scores ranked by absolute magnitude.
Cross-theme network
Node size tracks prevalence, while thicker links represent stronger pairwise structure among the top relationships.
Relationship notes
The notes below are populated from the same embedded JSON and matrix data used to render the charts.
How to read the dashboard
The bar chart and treemap summarize prevalence from themes_summary.json. The heatmap, ranked pair chart, and network use values parsed from theme_cooccurrence.csv.
Because the off-diagonal matrix values are negative in this extract, darker warm colors and thicker links indicate stronger separation between two themes, while the diagonal remains the self-value of each theme.
Hover any mark to inspect the exact score, definition, or keyword list.
Theme directory
Definitions and keyword signatures pulled directly from the extracted theme summary.
| Theme | Prevalence | Top words | Definition |
|---|