data/README.md

data/

Data files needed for running agents and evaluation in XBCP.

Directory structure

data/
├── crosslingual/
│   ├── lang_assignment.json               # docid -> language (per-document)
│   ├── crosslingual_corpus.jsonl          # Full corpus, Pyserini format (BM25 indexing)
│   └── crosslingual_corpus_tevatron.jsonl # Full corpus, Tevatron format (FAISS runtime)
└── multilingual/
    ├── lang_assignment.json               # docid -> language (per-query-group)
    ├── query_lang_assignment.json         # queryid -> language (per-language eval, tq+tp)
    ├── crosslingual_corpus.jsonl
    ├── crosslingual_corpus_tevatron.jsonl
    └── queries_translated.tsv            # Translated queries (tq+tp experiment)