Stephan Burlot sburlot

Extracting financial disclosure reports and police blotter narratives using OpenAI's Structured Output

tl;dr this demo shows how to call OpenAI's gpt-4o-mini model, provide it with URL of a screenshot of a document, and extract data that follows a schema you define. The results are pretty solid even with little effort in defining the data — and no effort doing data prep. OpenAI's API could be a cost-efficient tool for large scale data gathering projects involving public documents.

OpenAI announced Structured Outputs for its API, a feature that allows users to specify the fields and schema of extracted data, and guarantees that the JSON output will follow that specification.

For example, given a Congressional financial disclosure report, with assets defined in a table like this:

	#!/bin/bash

	# Creates a ramdisk and start Xcode with the DerivedData stored in ramdisk. Also deletes the ramdisk and reset Xcode prefs.
	# xc_ramdisk.sh
	# - creates a ramdisk, set Xcode DerivedData to this disk and start Xcode.
	# - umount a ramdisk, set Xcode DerivedData to default
	# Stephan Burlot, Coriolis Technologies, http://www.coriolis.ch
	#
	# based on Alex Shevchenko xcode_ramdisk.sh script (https://gist.github.com/skeeet/2367298)
	# based on Diego Freniche xc-launch.sh script (https://github.com/dfreniche/xc-launch)

	<?php
	// https://en.wikipedia.org/wiki/Category:Coffee_brands
	$covfefes = [
	'alterracoffeeroasters',
	'angiangcoffee',
	'angelinus',
	'autocrat',
	'barcaffe',
	'batdorfbronson',
	'bewleys',

	# !/usr/bin/env python3
	# Nicolas Seriot
	# 2022-06-10
	# https://seriot.ch/visualization/truchet_simple.png

	import cairo
	import math
	import random

	NB_COLS = 8

	# !/usr/bin/env python3
	# Nicolas Seriot
	# 2022-06-14
	# https://seriot.ch/visualization/truchet_2.png
	# loosely inspired by https://www.fxhash.xyz/gentk/slug/dirty-interlaced-truchet-tiles-18

	import cairo
	import math
	import random