Smrity Mallik · NLP demoCode on GitHub

German text analysis · Bundestag energy debates

Paste German text. See the energy terms and a guess of the party.

Two small models from the Energiewende project run here, in your browser. Nothing is sent anywhere. The text never leaves your device.

How far to trust it: the term finder is a word list (it only finds words it knows, and cannot tell for from against). The party guess comes from word choice only and is right about half of the time on speeches it has not seen, against one third to one half for always naming the biggest party. It is a teaching demo. Do not use it to label real people.

1. Your text

2. Energy terms found

technology policy The list has 14 stems (for example windkraft, wasserstoff, atomkraft, klimaschutz). A stem also matches compounds that begin with it, like Windkraftanlagen or Klimaschutzgesetz. This is the same list as the spaCy pipeline in the repository, which found 3,337 mentions in the 1,161 speeches, against 2,315 for exact-word matching.

3. A guess of the speaker's party

The bars are the model's probabilities for six parties. Short texts give weak, flat results.

Words the model leans on, by party

PartyStrongest word signals