Nikita Rokotyan

Exciting news — Market Map has been nominated for @TheWebbyAwards in the Best Data Visualization category! This means it's one of the five best data vis projects submitted, hooray 🎉 marketmap.one

Claude Code dependency graph visualization built by Claude & Cosmograph. run.cosmograph.app/public/dfb673f…

So great to see Cosmograph cosmograph.app in action, very smooth!

Cosmograph is still the only web-based framework capable of visualizing network graphs and ML embeddings with 1+ million points and way more than a million links. More about the release here: cosmograph.app/docs-general

Finally launching 🪐 Cosmograph 2.0! Work with larger datasets, use SQL, open Parquet files, save & share your graphs, integrate with Python or React, ... All thanks to open source tools like DuckDB, Mosaic, SQLRooms and our cosmos.gl → cosmograph.app

Explore the landscape of 5787 (!) @NeurIPSConf'25 papers. → mondal.dev/neurips-embedd… - Embeddings (of titles + abstracts) with Qwen3-Embedding-8B - 2D projection with t-SNE - Visualization and search over titles, abstracts, keywords, or authors with Cosmograph

Happy to see that the majority of the new features are community contributions! This is why I love open source

I never thought that this small experimental library could grow to be among such titans of high-performance visualization tools as deck.gl and kepler.gl. It's a real honor! Blog: openjsf.org/blog/introduci… GitHub: github.com/cosmosgl/graph

cosmos.gl was created to overcome bottlenecks when working with huge network graphs and machine learning embeddings. Version 2.0 comes with a significant performance boost over the previous version and implements new features like point dragging and clustering.

Big news for cosmos.gl — the core layout and rendering library behind cosmograph.app — it is going completely open source (MIT) and joining the OpenJS Foundation @openjsf ✨

@DennisAdriaans @nuxt_js Many of the components clearly come from Unovis unovis.dev, for which @zernonia had kindly contributed the Vue integration. A mention would be appreciated 🙂

Lots of fun building a Vue chart library for @nuxt_js

✴︎ Rectangular Selection – Select points of interest and use them in your analysis; ✴︎ Play/Pause Force Graph Simulation – Freeze node positions when needed; ✴︎ ... and more Check out this colab if you want to give it a try: #scrollTo=I9WzV2S7bBuK" target="_blank" rel="nofollow noopener">colab.research.google.com/drive/1fK7SLso…

Last year we released 🪐 Cosmograph Python Widget and have the first round of updates! ✴︎ Interactive Timeline – If your data has a temporal dimension, select a time range to highlight relevant points; 🧵

@The_Technocrat 🤩 Can you tell more about this graph? What does it show?

The interactive visual after submitting Deep Funding weights is slick (Cosmograph by @ernaem).

To keep entertained until the 15th, checkout the interactive Cosmograph (visualization courtesy of @ernaem ): cosmograph.app/run/?data=http…

I’ve just realized it’s Unovis’ 2nd birthday today 🥳 It’s amazing to see how much the project has grown—it now has 20K weekly downloads on NPM. Half of the users are @vuejs developers, thanks to the integration contributed by @zernonia. Huge thanks to the community!

@phil_renaud @AskNewsApp I don't know the exact logic, but it looks like the terms that appear in multiple articles get connected. I don't think the edge strength is being used here. We've added the support to the Cosmograph library but not into the app. We are working on a big update.

@ernaem @AskNewsApp Neat! what defines an edge? is there relative edge strength?

A month's worth of news visualized as a network graph by @AskNewsApp & 🪐 Cosmograph. I really like looking at how topics change over time. A bit noisy, but it still gives you understanding of what events have happened and how they are connected.

Physics simulation is the most enjoyable part of creative coding / data vis (at least for me). It brings so much joy to the dev process. (Working on a node clustering algorithm for https//cosmograph.app)

💡 We need more tools to work with textual datasets because we'll have unprecedented amount of text generated in the next decade – @ernaem #Outlier2024

Dimensionality reduction techniques come to the rescue #Outlier2024 by @ernaem