Datasets:

rafmacalaba
/

fcv-corpus

license: cc-by-4.0
configs:
  - config_name: reliefweb
    data_files:
      - split: train
        path: reliefweb.jsonl
  - config_name: refugee_pads
    data_files:
      - split: train
        path: refugee_pads.jsonl
  - config_name: fcv_pads_east_africa
    data_files:
      - split: train
        path: fcv_pads_east_africa.jsonl
  - config_name: jdc_operational
    data_files:
      - split: train
        path: jdc_operational.jsonl

FCV Corpus

A corpus of Fragility, Conflict, and Violence (FCV) related documents, standardized to a common JSONL schema. Every record has a pdf_url field pointing to the source PDF, so the corpus can be used directly for document inference.

Configs

Each config is one source corpus:

config	records	source
`reliefweb`	1,696	ReliefWeb metadata (UNHCR/OCHA/IFRC reports, etc.)
`refugee_pads`	196	World Bank refugee project documents (PADs)
`fcv_pads_east_africa`	21,338	World Bank FCV project documents (East Africa + regional)
`jdc_operational`	65	WB operational project documents (JDC operational set)

Schema

All records share pdf_url and title. The remaining fields are source-specific (standardized to snake_case; multi-valued fields are JSON lists).

reliefweb: organization, report_url, posted, originally_published, country, themes, disaster_type, language
refugee_pads: abstract, document_date, report_no, document_type, authors, country, disclosure_date, disclosure_status, language, major_sector, sector, topics, themes, region, project_id, landing_page_url
fcv_pads_east_africa: project_id, doc_id, document_type, document_date, language, disclosure_status, country
jdc_operational: project_id, country, document_type

Usage

from datasets import load_dataset

reliefweb    = load_dataset("rafmacalaba/fcv-corpus", "reliefweb")
refugee      = load_dataset("rafmacalaba/fcv-corpus", "refugee_pads")
fcv          = load_dataset("rafmacalaba/fcv-corpus", "fcv_pads_east_africa")
jdc          = load_dataset("rafmacalaba/fcv-corpus", "jdc_operational")