Music in European Parliament Plenary Discourse: Analysis, Validation, and Processed Corpus (v1.1.0)

This record contains the data and reproducibility materials for the revised study “Music in European Parliament Plenary Discourse: A Validated Full-Corpus Screening Workflow for Concept-Based Analysis”. The release is divided into two archives. 1. europarl_music_q1_analysis_v1.1.0.zip This archive contains the derived analysis datasets, the broad full-corpus screening output, restrictive post-filtered subsets, the blind validation sample, the normalised single-expert annotation, bootstrap validation metrics, multilingual and temporal diagnostics, false-positive and false-negative review materials, paper tables and figures, provenance metadata, manifests, checksums, source code, documentation, and a record-level data dictionary. 2. europarl_music_q1_corpus_v1.1.0.zip This archive contains the processed SQLite parliamentary corpus, the database schema, corpus documentation, provenance metadata, checksums, and the official European Parliament MEP term files used for speaker enrichment. The principal analysis view contains 446,208 parliamentary speech events from 1,732 plenary sitting dates, covering 15 April 1996 to 9 February 2026 and parliamentary terms EP4–EP10. Important interpretation: The 5,681-case dataset is the output of a broad screening pass over the complete analysis corpus. It is not a validated true-positive corpus. A stratified blind expert audit of 480 speeches estimated weighted precision of 0.052, weighted recall of 0.816, and weighted F1 of 0.098. The validation-adjusted estimate is approximately 364 truly music-related speeches in the corpus, with a bootstrap 95% interval of approximately 250 to 487. The restrictive 117-case subset is intended for qualitative interpretation. Its estimated weighted precision is 0.846 and its estimated weighted recall is 0.499. The validation was completed by one expert human annotator working blind to the model prediction and evidence. No statistic of agreement between independent annotators is reported. Code, prompts, and documentation are versioned in Codeberg: https://codeberg.org/perotinus/europarl-music-presence Source release:v1.1.0-q1-revision Source commit:0bfab350ed5753c31247cbe1bb9d020b766a1ec7 Screening model:qwen2.5:7b-instruct-q4_K_M The MIT License applies to the original project software and documentation. Reuse of European Parliament source records and derived data remains subject to the conditions specified in this Zenodo record and in the included DATA.md file.