We've been working on extracting information in templatized PDFs for the last couple of years, leveraging the best of LLMs and classical data extraction techniques.
Our latest technique, TWIX, has the best of all worlds: beats Azure DI, AWS Textract, or LLM-based approaches by over 25% in precision AND recall, and is multiple orders of magnitude cheaper and faster. TWIX is now open-source, with an API plus an optional front-end for you to tweak the extraction results.
Try it out on your gnarliest datasets and let us know what you think!
Led by Yiming Lin, w/ Yash Jain, Mawil Hasan, @alvinkcheung
@MoharirAdvait Thank you. I don't have STATA or a National Data Archive log-in. I was able to run your Parquet --> CSV notebook. Is it correct to assume that this a cleaned version of the raw data dump on the archive servers?
@_MayankSaxena has left us, suddenly. Here is to the poet and sensitive mind behind मोर्चे पर कवि . Every weekend along Bandra Sea Face, Mumbai a bunch of poets and writers would gather and relay their observations and visions. He started it. He will be remembered by all he met.
#WATCH | On Supreme Court's order to release NewsClick founder and Editor-in-Chief Prabir Purkayastha in the UAPA case, Advocate Arshdeep Khurana says "Supreme Court has held the arrest and the remand proceedings to be illegal and has directed the release of Purkayastha. We have been directed to furnish the bail bond before the trial court. This is a major and big relief because we have been maintaining from the beginning that the entire proceedings against him were illegal and the manner of arrest was illegal which has now been upheld by the Supreme Court..."
Prabir Purkayastha, my former colleague and friend, has been wrongfully jailed for his decades of science activism, critical reporting and excellent, path breaking online news channel @newsclickin
Release Prabir.
https://t.co/RinBsNfhF1
Thrilled to share that my story for @techreview on how ASHA workers use WhatsApp to save pregnant women has been shortlisted for the True Story Award!
A perfect end to 2023.
Here's the story:
https://t.co/CKdJNyoN4l
@truestoryaward
Best of PARI 2023
During the last 12 months, we archived hundreds of reports and surveys, and thousands of words were archived in the PARI Library.
Each one of these documents supported and validated issues of justice and rights. ⬇️
https://t.co/7nifICXnHd
We completed 9 years of reporting!
And as 2023 draws to a close, we've slated a series of year-end reviews! Every day, for the rest of the year, we will publish our editors’ choices of the Best of PARI.
We begin with the year in focus 👇
https://t.co/T2f5S1B43M
“I keep thinking about what may happen if the climate changes & there is no longer enough water to drink or grass to feed my animal”
With temperatures rising in Ladakh, yak herders in Zanskar Valley find it difficult & unprofitable to maintain their herd https://t.co/h6ZjJkmfVL
Wasim Ahmed's shop was among the many the Madhya Pradesh government bulldozed in Khargone after last year's communal riots.
The state, supposedly, was punishing "stone pelters".
But Wasim lost both his hands in an accident in 2005.
My ground report
https://t.co/Z2Iq7ESNeM
Maneesha, a trans woman works as a fish auctioneer in Cuddalore. She says:
“Boat owners look upon me as a lucky hand. I auction fish better than others. They don't reject me. They just want me to sell their fish. Only my family hesitated to accept me, not the fishermen.”
🧵👇
Life in North Chennai through the lens of young residents: An exhibition of photographs, Dec 8-10
Please extend your valued support to this call:
https://t.co/o5YrYRps5u
The kids are mentees of @chempkumar photojournalist @PARInetwork
@vfsglobalcare@vfsglobalcare Your support team is unfortunately not empowered or trained to handle the software errors in your new website. They don't know what to do. Literally what OFC staff said to me. Look into it. The poor helpdesk folks can't do much. Clearly understaffed and untrained.