Try / Inspect / Change
Small inputs. Visible results.
Text similarity, lexicon-based sentiment, information extraction and setup commands. Each example exposes its method and boundaries.
Lab
Interactive NLP example
What do two texts have in common?
A word-based comparison with TF–IDF and cosine similarity. Inspect the terms behind the result.
Shared terms
How is the result calculated?
Locale-aware lowercasing → tokenization → stop-word removal → TF–IDF weighting → cosine similarity.
The score reflects shared words; it does not evaluate synonyms, context or factual accuracy. Text is processed on this page.
Lexicon-based example
Which words change the tone?
Inspect positive and negative words with simple negation rules.
Results use a small lexicon and simple rules. Irony, context and word inflections can change interpretation.
Pattern extraction example
Separate information from a text.
Extract email addresses, links and hashtags locally.
This example finds text patterns; it does not verify whether addresses or links exist. Input stays in your browser.
Command generation from an allowlist
Define your tools. Take your profile.
Choose your tools to generate a reusable set of Winget commands. Review the commands before using them in your own environment.
This example generates commands. Installation starts when you run them on your own computer.
Explore project: AppAnvil