1 of 6

HACKATHON PROJECT • TRACK 1: PUBLIC DATA APIs

Expert Evidence

Graph

From a litigation question to evidence-backed expert leads.

DEMO SEARCH

forensic accounting lost profits business valuation

For lawyers and litigants • Live prototype • Public court records • Traceable evidence

A lawyer should be able to ask “who fits?”—and then verify why.

EXPERT EVIDENCE GRAPH / HACKATHON

2 of 6

01 / THE PROBLEM

Expert discovery is a research problem—not just a directory problem

Built for lawyers and litigants. A relevant name is only the beginning; they need evidence that explains the match.

No open expert search

Court sites aren’t built to search by expertise, and dedicated expert-research tools are closed, commercial platforms.

Only marketing

A lawyer may land on an expert’s own website, and not every expert has one. That’s their marketing, not real court data.

Hard to verify quickly

Lawyers must manually connect the person, the dispute, and the supporting passage.

The gap: move from “search results” to “a candidate + the case evidence behind the match.”

EXPERT EVIDENCE GRAPH / HACKATHON

Expert witness: a specialist a court allows to give opinions at trial. Choosing the right one can shape a case.

3 of 6

02 / THE SOLUTION

A source-backed path from query → candidate → evidence

Ranks potential experts, visualizes connections, and links every match to its court record, which directories and Google don’t.

ASK

Describe the dispute

→

MATCH

Rank potential experts

→

TRACE

Inspect cases & evidence

POTENTIAL EXPERT

Carrol Gatlin

Forensic review accounting • Business valuation

MATCH

Evidence strength: STRONG • 1 case

EXPERT

CASE

ISSUE

PASSAGE

Expert

Quinn v. Palmer

Accounting / valuation

Every promising match should lead back to inspectable source evidence.

EXPERT EVIDENCE GRAPH / HACKATHON

Product score and evidence label shown as reported by the current prototype; they are not a legal opinion or guarantee of suitability.

HOW THE SCORE WORKS Named in the case • Specialty match • Cases found • Trial won • Appeal won

A transparent heuristic, not a verdict. Win and appeal outcomes are context, not proof of expert quality, and we plan to recalibrate them.

4 of 6

03 / CASE WALKTHROUGH

One opinion reveals expert roles—and the evidence behind them

Illustrative live-search result: F. Patrick Quinn III v. Edwin M. Palmer & Decatur Hotels, L.L.C. (La. App. 4 Cir. 2024)

WHAT THE OPINION SAYS

Carrol Gatlin

Accepted as an expert in forensic review accounting and business valuation.

$1.331M

Duba lease transfer loss

The appellate court affirmed the trial judgment and denied the writ.

Total judgment against Palmer: $3,873,217 for Decatur.

THE CONNECTION TO SHOW

Carrol Gatlin

Quinn v. Palmer

Accounting evidence

Person → case → source passage

Accuracy note: the court rejected a separate lost-profits theory as speculative. This example demonstrates traceability—not that Gatlin won that theory or that outcome predicts future performance.

EXPERT EVIDENCE GRAPH / HACKATHON

Source: Louisiana Court of Appeal, Fourth Circuit, No. 2023-CA-0181, decided Feb. 5, 2024. https://law.justia.com/cases/louisiana/fourth-circuit-court-of-appeal/2024/2023-ca-0181.html

5 of 6

04 / TECHNICAL IMPLEMENTATION

A thin, inspectable pipeline—not a black-box chatbot

The current prototype uses a web UI, a server-side API proxy, CourtListener records, and a ranking/evidence presentation layer.

01 USER QUERY

Query + jurisdiction

from React UI

02 EDGE FUNCTION

Supabase function

keeps API token server-side

03 COURTLISTENER

REST search API

returns opinion records

04 RANK + EXPLAIN

TypeScript processing

score, graph, evidence panel

→

Search construction

Builds a CourtListener query and applies a court filter where supplied.

Source transparency

Returns record links and snippets so users can inspect source material.

Honest fallback

Demo data keeps the workflow usable when live search is unavailable.

Next technical milestone: entity resolution across cases + calibrated, auditable scoring with evidence-level citations.

EXPERT EVIDENCE GRAPH / HACKATHON

RESPONSIBLE DATA USE Public opinions only • API token kept server-side • Source link on every result • Not legal advice

6 of 6

05 / WHY IT MATTERS

Make expert discovery faster to investigate—and easier to verify

A narrow first use case gives us a measurable starting point: forensic accountants in business-damages disputes.

WHAT THE PROTOTYPE SHOWS

Need → candidate

A query returns a ranked shortlist, not just a generic answer.

Candidate → case

A graph makes the relationship between a person and litigation visible.

Case → evidence

The user can inspect the source instead of trusting an unexplained claim.

WHAT WE’LL VALIDATE NEXT

Precision

Do lawyers agree that the top results are relevant to the matter?

Evidence quality

Can every score be explained with a source-backed reason?

Generalization

Can the approach extend to more expert specialties and jurisdictions?

Expert Evidence Graph

Find the lead. Follow the evidence.

LIVE DEMO

EXPERT EVIDENCE GRAPH / HACKATHON

Demo app: https://ubiquitous-disco-vprxx97jw667hpgg4-5173.app.github.dev/