sanjeevafk/kalam-speech-corpus

Dr. A.P.J. Abdul Kalam Lifetime Speech & Lecture Corpus

Dataset Summary

The Dr. A.P.J. Abdul Kalam Lifetime Speech & Lecture Corpus is a comprehensive, machine-readable dataset of speeches, addresses, convocations, and scientific lectures delivered by Dr. A.P.J. Abdul Kalam across his career as a rocket scientist, 11th President of India (2002–2007), and global statesman (2007–2015).

• Total Canonical Events: 912 unique speech events
• Total Raw Sources Tracked: 1,028 source records
• Total Word Count: 1,822,794 words
• Coverage Period: 1963 – 2015
• Verified Presidential Core: 1,024 site records (785 unique canonical speeches)
• Primary Data Formats: Parquet (speeches.parquet, sources.parquet, duplicates.parquet), JSONL (speeches.jsonl, sources.jsonl, duplicates.jsonl)

Dataset Structure

kalam-speech-corpus/
├── README.md                          # Hugging Face Dataset Card
├── RIGHTS.md                          # Legal rights & licensing assessment
├── data/
│   ├── speeches.parquet               # Canonical speech events (Primary Parquet file)
│   ├── sources.parquet                # Raw source records Parquet (provenance layer)
│   ├── duplicates.parquet             # Candidate duplicate mappings Parquet
│   ├── speeches.jsonl                 # Canonical events JSONL
│   ├── sources.jsonl                  # Raw source records JSONL
│   └── duplicates.jsonl               # Candidate duplicate mappings JSONL
├── metadata/
│   ├── qc_report.json                 # Quality control audit report
│   └── coverage_by_year.csv           # Annual coverage metrics
└── scripts/
    ├── schema.py                      # Data models (Dataclasses)
    ├── ingest_presidential.py         # Presidential archive ingest script
    ├── discover_lifetime.py           # Pre/Post presidency discovery module
    ├── dedup_and_qc.py                # Pipeline execution & export script
    ├── export_hf.py                   # Hugging Face hub dataset packager
    └── upload_hf.py                   # HF dataset uploader script

Schema & Features

Each row in speeches.parquet represents one canonical speech event with the following fields:

Field Name	Type	Description
speech_id	string	Unique canonical speech identifier (e.g. KALAM-2007-0006)
canonical_event_id	string	Canonical event ID linking duplicate/multi-source records
title	string	Normalized title of the address
date	string	Date string (DD.MM.YYYY or DD-MM-YYYY)
year	int	Event year
location	string	Extracted venue location (e.g., New Delhi, Hyderabad, Strasbourg)
venue	string	Venue / Host institution
period	string	Era (pre-presidency, presidency, post-presidency)
event_type	string	Category (presidential_address, convocation, distinguished_lecture)
speaker	string	Dr. A.P.J. Abdul Kalam
tldr	string	Extractive key summary (3-4 sentences preserving original wording)
transcript	string	Full cleaned transcript in Markdown
transcript_status	string	Status (official_transcript, institutional_transcript, video_only)
transcript_confidence	string	high, medium, low
source_url	string	Official source URL
source_type	string	official_presidential, institutional_archive, video_recording
source_domain	string	Source web domain
rights_status	string	Rights assessment (public_domain_gov, educational_fair_use, copyright_restricted)
sha256	string	SHA-256 hash of the transcript text

Quick Start / Loading Dataset

from datasets import load_dataset

# Load canonical speeches (default)
ds = load_dataset("sanjeevafk/kalam-speech-corpus")
print(ds["train"][0])

# Load raw source records (provenance layer)
sources_ds = load_dataset("sanjeevafk/kalam-speech-corpus", "sources")
print(sources_ds["train"][0])

Quality Control & Methodology

1. No Silent Rewriting: Original transcript wording, headings, lists, and quotes are strictly preserved.
2. Double-Layer Provenance: Preserves raw source records (sources.jsonl) linked to canonical events (speeches.jsonl).
3. Automated QC: Checks for missing titles, empty transcripts, site boilerplate, and encoding corruptions.

Citation & Usage

@dataset{kalam_lifetime_corpus_2026,
  author = {Kalam Speech Archival Project},
  title = {Dr. A.P.J. Abdul Kalam Lifetime Speech and Lecture Corpus},
  year = {2026},
  publisher = {Hugging Face},
  howpublished = {\\url{https://huggingface.co/datasets/sanjeevafk/kalam-speech-corpus}}
}