Harry Howard's Journals

Using AI to Make Big Data More Accessible

Primary Tools

React logo

React

Next.js logo

Next.js

Node.js logo

Node.js

TypeScript logo

TypeScript

Tailwind CSS logo

Tailwind CSS

ChatGPT API logo

ChatGPT API

Prisma logo

Prisma

Supabase logo

Supabase

Project Image

Harry Howard's Journals
Link logoLive SiteGitHub logoSource Code

Summary

I'm proud to introduce Harry Howard's Journals, an advanced web application that redefines the way we explore historical journal entries.

Using Next.js, Python, and AI tools, the app processes and analyzes entries from the 1930s and 40s. These entries are enhanced by AI and made easily accessible, putting a treasure trove of historical data at your fingertips.

Features

  • Harry has been brought to life with state-of-the-art AI technology. Ask Harry detailed questions about the journals, or just chat! (powered by ChatGPT)
  • Read journal entries one page at a time or easily skip around
  • Search function for journal entries
  • AI-generated topics for each entry allow users to understand an entry at a glance and link together stories woven through multiple entries
  • Sign in with Google to keep track of pages read, save special passages and make posts to discuss entries with others

Other Tools

Uses Python library OpenCV to preprocess imagesRuns CRAFT AI locally to split images into linesApplies Google Vision API on individual lines of textUses GPT-4 to reconstruct journal entry from Google Vision outputGenerates metadata on journal entries with GPT-3.5 APIStores data with Google Cloud StorageRuns python scripts with Google Cloud Run