Norwalk Chronicler

How it works

What happens behind Find, Ask, the scans and the map — and where each one can mislead you.

The text

Every newspaper page was scanned and run through optical character recognition (OCR) by the Connecticut Digital Archive; the books were scanned and OCR’d by the Internet Archive. OCR on nineteenth-century type is rough: “Norwalk” can come out as “Norwa1k” or “N orwalk”, columns can run together, and ads can bleed into news. The Chronicler shows you that text as it is — it does not silently “correct” it.

Find

Find is a full-text search over every page. With fuzzy spelling on, each word also matches the common OCR misreadings of it, which is how a search for “oyster” also finds “0yster”. Put a phrase in “quotes” for an exact match, or end a word with * to match its beginning. When you search for several words, results are marked together when the words fall in the same stretch of text (likely one article) and apart when they only share a page.

Ask

Ask first finds the passages most likely to answer your question, then hands only those passages to an AI language model running on our own computer, which writes an answer and cites the passage behind each claim as [1], [2]… The model is told to use nothing but those passages and to say so when they don’t answer the question. It can still misread OCR text or join facts that don’t belong together, so treat an answer as a pointer to the clippings, not as a finding — open the cited pages and read them. Use the filters to limit Ask to a neighborhood, years, a single paper, or to the books or newspapers alone.

The scans

Every newspaper page opens beside its original scan from the Connecticut Digital Archive, with your search words boxed on the image where the OCR found them. The books link to their scans at the Internet Archive.

The map and neighborhoods

Street addresses printed in the papers (“10 Haviland Street”, “on Water Street”) were extracted from the text and placed on the map, keeping only streets that really exist in Norwalk. Neighborhood outlines are the City of Norwalk’s own; historic names that are no longer on the map, such as Whistleville, are shown as approximate circles.

Privacy

There are no accounts and no ads. Your Ask conversation is kept only in your own browser.