Learning What’s in a Name with Graphical Models
Presented at VISxAI 2023
Made in 2023 with D3, React, and Python
An in-depth account of graphical models and their use in named entity recognition — identifying names of organizations, locations, people, etc., from text sequences.
A custom-built network visualization helps bring many important concepts to life. Various panels containing interactive control elements allow readers to explore alternative configurations and test their understanding of the material.
![Tiled scatter plots](/static/443589a1d1b69812c211426c72d24bd0/532b7/sampling.png)
![Emission paths of a Hidden Markov Model](/static/22ee33d32796daf96273cac2017936f9/8cde9/hmm-emissions.png)
![Transition paths of a Hidden Markov Model](/static/863e5ad7b88821f239776c1f99700032/8cde9/hmm-transition.png)
![Named entity predictions from HMM, MEMM, and CRF models](/static/750ba34b30da1acd0a5479ece4ced0ed/41dc1/prediction.png)
![Breakdown of how to calculate the transition probability in a Maximum-Entropy Markov Model](/static/8f2a46d7e70f7f93fbb2a268efc0e32f/f264e/memm-transition.png)