Videlicet

What is Videlicet?

Videlicet™ is an Information System that can execute and measure the accuracy of one to many Prompts using one to many Artificial Intelligence (AI) Large Language Models (LLMs) resulting in an archive that is searchable via semantic search.

Why Videlicet™ is unique:

1. Word searchable archive
   • AI enabled Semantic Search (US Patent pending)
   • Keyword Search
2. For better accuracy, uses an AI LLM (not OCR) for transcription
   • Transcribes both printed and cursive text - OCR cannot accurately recognize cursive
   • Transcribes Early Modern English which OCR did not do well
   • Transcribes documents of questionable preserved quality
3. Ability to extract whole sections such as advertisements and poems
4. Uses Levenshtein Distance as a measure of transcription accuracy

How does Videlicet™ work?

1. Documents are collected, found, and scanned into the Videlicet archive
2. Prompt engineers develop prompts to accurately transcribe the document with the LLM
3. A subset of the documents are transcribed manually by a human
4. Evaluations are performed to compare the documents and prompts processed by the LLM with human transcripted documents resulting in a Levenshtein score
5. Researchers iterate this process until the Levenshtein score shows accurate processing of the document
6. The prompt is run against the full archive of documents and the results stored in the database for future researchers
7. Videlicet has the ability to create distinct prompts for cursive handling or to extract text out of a document (poems, advertisements)

Semantic versus keyword search:

1. I want to search for an enslaved man name “Paris” or “Parris” who ran away multiple times
2. Keyword search will find every instance of “Paris” including the city of Paris.
3. Semantic search will find documents for only the enslaved person named Paris.