Corpus Q&A tool

Ranks paragraphs with TF-IDF, matches patterns with KMP and tries, then feeds the best context to an LLM.

C++TF-IDFKMPTriesLLMs

Code kept offline (coursework). Happy to walk through it.

What it does

Ask a question about a large text corpus. Paragraphs are ranked with a TF-IDF scoring scheme, patterns are found with KMP plus tries and heaps, and the top context is sent to ChatGPT or Hugging Face models within strict token limits. The “Ask about my work” box on this site uses the same ranking idea.

Size

630 lines of code.

Next project

Rollerball AI bot

Keep going