CAP Lawsuit Summarization

Export Extraction Results

Download AI extraction results for import into Airtable or analysis tools.

Download Data

Export contains 5 extracted cases using the Two-Step extraction method (Qwen3 VL OCR + Llama 4 extraction).

Download TSV (for Airtable) Download CSV Download JSON (full data)

How to Import into Airtable

1
Download the TSV file
TSV (Tab-Separated Values) works best with Airtable. Click the blue button above.
2
Open your Airtable base
Go to the base where you want to import the data (e.g., your LELU database).
3
Import the file
Click "+ Add or import" in the top-left, select "CSV file" (works for TSV too), and upload the file.
4
Map the fields
Airtable will show a preview. Match each column to your existing LELU fields. Multi-value fields (allegations, themes, etc.) are separated by pipes (|).
Tip: For multi-select fields in Airtable, use Find & Replace to change pipes (|) to commas after import.

Exported Fields

Field Description Example
case_id Unique case identifier owens_troy
summary AI-generated narrative summary "On December 9, 2015, the police officer defendants..."
allegations LELU allegation categories False arrest | Excessive force
force_types Types of force alleged Non-weapon physical force | Tight handcuffs
themes Thematic categories Race and/or ethnicity | Mental health
locations Where the incident occurred Street | Precinct
defendant_types Types of defendants named NYPD | City of New York
officers Officer names identified John Smith | Jane Doe
num_pages Document page count 12
total_cost_usd Processing cost 0.0384
recall Accuracy vs ground truth 0.75 (75%)