pa-workflow-litigation-transcript-search-q-and-a-indexing

Category: Design Risk: Unknown ★ 3.9 · Rating 3.9/5 (8) sboghossian/mini-claude-for-legal MIT

Rating is derived from the repo's GitHub stars and shown for reference.


name: pa-workflow-litigation-transcript-search-q-and-a-indexing
description: Use when a litigation team needs to index and search deposition transcripts, trial hearing transcripts, or arbitration session transcripts for witness preparation, cross-examination planning, or factual reconstruction. Builds a Q&A-indexed, topic-clustered, searchable corpus from raw transcripts; detects intra-witness and inter-witness contradictions; and powers the real-time trial assist and deposition binder workflows.
license: MIT
metadata:
id: pa-workflow.litigation.transcript-search-Q-and-A-indexing
category: pa-workflow
practice_area: Litigation
jurisdictions: [US, UK, DIFC, ADGM, UAE, KSA, LB, EG]
priority: P1
intent: [transcript, deposition, Q-and-A, indexing, witness-prep, cross-examination, search]
related: [pa-workflow-litigation-witness-contradiction-finder, pa-workflow-litigation-deposition-binder-builder, pa-workflow-litigation-real-time-trial-assist-api, pa-workflow-litigation-expert-witness-prep-memo, pa-workflow-litigation-discovery-first-pass-tagging]
source: Louis — HAQQ Legal AI (github.com/sboghossian/mini-claude-for-legal)
version: "1.0"

Litigation — Transcript Search and Q&A Indexing

Purpose

Raw deposition and hearing transcripts are unwieldy: a complex commercial arbitration can generate thousands of pages across dozens of witnesses. This workflow transforms raw transcript text into a structured, searchable Q&A index — enabling counsel to find every statement a witness made on a given topic in seconds, cluster testimony by subject matter, and immediately surface potential contradictions across witnesses or between sessions.

Inputs

Input Required Notes
Transcript files Yes PDF, plain text, Word — single or batch
Transcript type Yes Deposition / trial hearing / arbitration session / regulatory investigation
Witness name(s) Yes For attribution and per-witness indexing
Dates of transcript(s) Yes Critical for contradiction timeline analysis
Matter summary Recommended Drives topic clustering
Prior-session transcripts (same witness) Optional Enables intra-witness contradiction detection
Opposing witness transcripts Optional Enables inter-witness contradiction detection

Processing Pipeline

Step 1 — Ingestion and parsing

  • Accept transcript in any standard format
  • Identify and tag: question-asker (examining counsel, opposing counsel, arbitrator/judge), witness, objections, non-answer responses ("I don't recall," "I'm not sure," "that's not accurate")
  • Assign a sequential Q&A number to every question-answer pair
  • Flag: interrupted answers, objections sustained / overruled, documents shown to witness during examination

Step 2 — Q&A indexing

For each Q&A pair, produce:

{
  "qa_id": "DEP-AhmedAlSayed-2023-03-14-Q247",
  "witness": "Ahmed Al-Sayed",
  "date": "2023-03-14",
  "transcript_type": "Deposition",
  "page": 47,
  "line": 12,
  "examiner": "Plaintiff's counsel",
  "question": "Did you review vendor contracts over ,000 before they were signed?",
  "answer": "Yes, I personally reviewed and approved every vendor contract over ,000.",
  "topics": ["contract_approval", "authorization", "vendor_management"],
  "exhibit_shown": null,
  "non_answer": false,
  "flag": null
}

Step 3 — Topic clustering

Automatically group Q&A pairs into topic clusters using the matter summary as a guide:

  • Contract formation / approval
  • Payments and financial transactions
  • Corporate authority and authorization
  • Communications with specific parties
  • Knowledge of specific events
  • Timeline of events

Output: per-topic index with all relevant Q&A pairs from all transcripts, sorted chronologically.

Step 4 — Contradiction detection

Intra-witness (same witness, multiple transcripts/sessions):

  • Compare statements on the same factual proposition across deposition session 1, session 2, and hearing
  • Flag: direct factual contradiction (same question, different answer), timeline inconsistency (date stated differently), denial of knowledge when prior session indicates knowledge

Inter-witness (across witnesses, same matter):

  • Compare statements on the same factual proposition across different witnesses
  • Flag: conflicting accounts of the same event; different characterizations of the same document; contradictory claims about who attended a meeting or made a decision

Contradiction severity:

Level Definition
HIGH Direct factual contradiction on a material issue — usable for impeachment
MEDIUM Inconsistency that requires explanation — may be innocent (faulty memory, different vantage point)
LOW Tone or characterization difference — unlikely to be significant

Step 5 — Search interface

The indexed corpus supports:

  • Full-text keyword search ("wire transfer," "approval," "contract No. 4")
  • Per-witness search (all Q&As by Ahmed Al-Sayed on topic X)
  • Date-range search (all testimony from after September 2022)
  • Topic search (all Q&As on "authorization" across all witnesses)
  • Contradiction search (all HIGH-level contradictions for witness X)

Output

Per-witness testimony profile

## Testimony Profile — Ahmed Al-Sayed
**Sessions indexed**: Deposition 2023-03-14 (87 pages), Deposition 2023-05-22 (53 pages)
**Total Q&A pairs**: 641

### Key Admissions
- Q247 (3/14/23): Confirmed personal approval of all vendor contracts over 
- Q312 (3/14/23): Confirmed he signed Exhibit 22 (contract approval form)
- Q089 (5/22/23): Confirmed receipt of the memo dated 2022-09-01

### Contradictions Found
| ID | Topic | Prior statement | Later statement | Severity |
|---|---|---|---|---|
| C-001 | Contract review | "I reviewed all contracts" (Dep 1, Q247) | "I never reviewed that document" (Dep 2, Q156) | HIGH |

### Non-Answer Instances (41 total)
Topics where witness claimed lack of recall: [list]

Matter-wide topic index

For each topic cluster: a consolidated table of all Q&As from all witnesses, enabling counsel to compare accounts side by side.

Contradiction master log

All contradictions across all witnesses, ranked by severity, with transcript references.

Integration with Other Workflows

This workflow's output feeds directly into:

  • [[pa-workflow-litigation-deposition-binder-builder]] — per-witness document packages
  • [[pa-workflow-litigation-witness-contradiction-finder]] — full contradiction analysis
  • [[pa-workflow-litigation-real-time-trial-assist-api]] — the indexed corpus is the trial database

Jurisdictional Notes

  • US: Court reporters produce verbatim transcripts. Errata sheets (changes by witness post-deposition) must be captured and noted — changed answers are themselves impeachable.
  • UK / DIFC / ADGM: Written witness statements serve as direct examination; oral examination is cross-examination and re-examination. Index witness statements as "transcripts" and cross-examination as separate sessions.
  • International Arbitration: Procedural orders may limit transcript availability to approved parties. Ensure proper handling of confidential arbitration transcripts.
  • UAE onshore / KSA: Court sessions are not typically transcribed verbatim by a court reporter. Session minutes (da'wa records) are prepared by the court clerk. These are the transcript equivalent for indexing purposes and are in Arabic. Ensure Arabic-language parsing is enabled for MENA court session notes.
  • Lebanon / Egypt: Similar to UAE — court session records (procès-verbaux) are the indexable source. Bilingual (Arabic/French for Lebanon) processing may be required.
  • [[pa-workflow-litigation-witness-contradiction-finder]]
  • [[pa-workflow-litigation-deposition-binder-builder]]
  • [[pa-workflow-litigation-real-time-trial-assist-api]]
  • [[pa-workflow-litigation-expert-witness-prep-memo]]
  • [[pa-workflow-litigation-discovery-first-pass-tagging]]