Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/holzerjm/civichacks-demo/llms.txt

Use this file to discover all available pages before exploring further.

The JusticeHack dataset (justicehack_ma_justice.txt) is a synthetic criminal justice reform progress report for Massachusetts, 2025. It contains detailed metrics on incarceration rates, racial disparities, pretrial detention, recidivism, policing data, juvenile justice, and access to legal representation.

​
Dataset overview

File: data/justicehack_ma_justice.txt
Scope: Massachusetts Criminal Justice Reform Progress Report, 2025
Size: ~31 lines, 14,500+ characters
Format: Plain text, structured report

​
Data sections

​
Prison and jail populations

  • State prison population: 6,847 (down 18% from 2019)
  • County jail population: 4,291 (down 12% from 2019)
  • Incarceration rate: 152 per 100,000 (lowest in US, national avg: 350)
Racial disparities persist despite 2018 reforms:
  • Black residents incarcerated at 7.9x the rate of White residents
  • Hispanic/Latino residents at 4.2x the rate of White residents
  • No improvement since 2018 Criminal Justice Reform Act

​
Sample queries

Here are the three built-in queries for the JusticeHack track: 
python scripts/demo_step2_rag.py justice 1

​
Query 1: What racial disparities exist in pretrial detention in Massachusetts?

This query surfaces:
  • Black: 38% of pretrial detainees (population: ~9%)
  • Hispanic/Latino: 31% of pretrial detainees
  • White: 27% of pretrial detainees (population: ~70%)
  • 62% held on bail ≤$5,000
  • Average detention: 94 days
  • 71% of low-level detainees in Suffolk County ultimately released without further incarceration
  • Pretrial detention increases guilty plea likelihood 2.5x

​
Query 2: How effective are reentry programs at reducing recidivism?

This query identifies:
  • Overall 3-year recidivism: 32%
  • Prison education participants: 28% lower recidivism
  • Employment within 30 days: 47% lower recidivism
  • 68% report employment difficulties
  • 41% face housing instability within 6 months
  • Geographic disparity: Suffolk 14 programs, Berkshire 2
  • $28M budget serves only 35% of those released annually

​
Query 3: What does the data reveal about policing patterns in Boston?

This query highlights:
  • 14,827 FIO stops in 2025
  • Black individuals: 63% of stops (23% of population)
  • 89% of stops result in no action
  • Most common reason: “suspicious behavior” (47%)
  • Use of force: 72% involve Black/Hispanic individuals
  • 94% body camera compliance
  • Internal affairs: 7.5% complaint sustain rate, 278-day avg resolution

​
Key metrics reference

MetricValueContext
State prison population6,847Down 18% from 2019
County jail population4,291Down 12% from 2019
MA incarceration rate152 per 100KLowest in US (national: 350)
Black incarceration rate7.9x WhiteNo improvement since 2018 reforms
Hispanic/Latino rate4.2x WhiteNo improvement since 2018 reforms
MetricValueContext
Daily pretrial population3,41248% of total jail population
Average detention length94 days—
Held on bail ≤$5K62%—
Cost per person per day$182Suffolk County
Black detainees38%—
Hispanic/Latino detainees31%—
White detainees27%—
Low-level → no further incarceration71%Suffolk County
Pretrial → guilty plea multiplier2.5xvs. released defendants
Cash bail usage-23%Since 2018 reforms
Dangerousness hearings+45%Alternative detention path
FactorRecidivism rateImpact
Overall 3-year32%Baseline
Property crimes41%+9 pts
Drug offenses38%+6 pts
Violent offenses24%-8 pts
Prison education programs—-28%
Employment within 30 days—-47%
Barriers:
  • 68% difficulty finding employment
  • 41% housing instability within 6 months
  • FY2025 budget serves only 35% of releases
MetricValuePopulation %
Total FIO stops14,827—
Black individuals stopped63%23% of Boston
Hispanic/Latino stopped18%—
White individuals stopped15%—
Stops with no action89%—
Reason: “suspicious behavior”47%—
Use of force incidents387Down 8% YoY
UoF involving Black/Hispanic72%—
Body camera compliance94%—
MetricValueContext
DYS commitments186Down 42% from 2019
Average commitment length14.2 months—
Black youth41%—
Hispanic/Latino youth33%—
White youth22%—
Mental health diagnosis73%Only 48% received treatment
Community alternatives-19% re-offensevs. facility commitments
Suspended/expelled → justice contact3.1xWithin 2 years
Black BPS suspension rate4x WhiteSchool-to-prison pipeline

​
Using this dataset in the web app

When you select JusticeHack in the Gradio app (Step 3), the interface displays:
  • Header: “⚖️ JusticeHack — Analyze Massachusetts criminal justice reform, incarceration disparities, and policing data”
  • Example questions: All three queries above as clickable examples
  • Chat responses: AI answers grounded in the specific metrics from this dataset
Try asking follow-up questions like “Why is the pretrial detention rate so high?” or “What explains the school-to-prison pipeline?” to explore the data interactively.

​
Querying from code

Here’s how the RAG pipeline loads and queries this dataset: 
from llama_index.core import VectorStoreIndex, SimpleDirectoryReader, Settings
from llama_index.llms.ollama import Ollama
from llama_index.embeddings.huggingface import HuggingFaceEmbedding

# Configure local AI
Settings.llm = Ollama(model="llama3.1", request_timeout=120.0)
Settings.embed_model = HuggingFaceEmbedding(model_name="all-MiniLM-L6-v2")

# Load the JusticeHack dataset
data_file = "data/justicehack_ma_justice.txt"
documents = SimpleDirectoryReader(input_files=[data_file]).load_data()

# Build vector index
index = VectorStoreIndex.from_documents(documents)
query_engine = index.as_query_engine(streaming=True, similarity_top_k=3)

# Query
response = query_engine.query(
    "What racial disparities exist in pretrial detention in Massachusetts?"
)
response.print_response_stream()
See Step 2: RAG with civic data for the full implementation.

Build docs developers (and LLMs) love

Get started for freeTalk to us