Graph Explorer

What is a community? A group of keywords that co-occur on the same papers much more often than expected by chance (Louvain algorithm).
How to read: Node size = paper count. Color = dominant taxonomy domain. Lines = co-occurrence within the community.
Insight: Communities reveal how research subfields naturally cluster — they often cross taxonomy boundaries (e.g. NLP + ML keywords in the same community).

How to read: Root keyword at top, specializations below. Size = paper count. Abbreviations shown in badges.
Example: "neural network" → children: "graph neural network" (GNN), "recurrent neural network" (RNN), etc.
Insight: Shows how broad concepts break into niche sub-areas. Click a leaf to expand further.

Each bubble = a research domain (14 total). Size = number of papers.
Lines connect domains that share papers classified in both areas. Thicker = more shared.

"Bridge" keywords appear in papers spanning many research domains.
Brighter color = connects more domains. These are the most interdisciplinary methods.

Slide through years to see how research focus shifted.
Each bar = a keyword from the selected year. Bar length = papers published that year. Press Play for a smooth race with counter updates every 0.1s.

"DNA" = what makes this conference unique. Keywords scored by TF-IDF specificity.
Bright/hot = highly characteristic of this conference. Dark = common across all conferences.