This document contains practical information for using the data. **1. NAMING CONVENTIONS** - **document corresponding to a movie's subtitle**: name of the movie in camel case, hyphen, add the iso code of the language: NameOfMovie-de (Avatar-en) - **documents corresponding to serie's subtitles:** name of the show in camel case (consisting of name, season number and episode number), hypen, add the iso code of the language: ShowS05E10-de: breakingBadS01E01-fr - **name of linking file:** name of the serie/movie, hyphen, iso code of source language, underscore, iso code of target language: breakingBadS01E01-en_nl - **Portuguese and Spanish subtitles:** added LA for Latin American to specify whether the files are European or Latin American subtitles: - howIMetYourMotherS01E01-spLA.srt (from Latin America) - howIMetYourMotherS01E01-sp.srt (from Spain) - HowIMetYourMotherS01E01-ptLA.srt (from Brazil) - HowIMetYourMotherS01E01-pt.srt (from Portugal) - **Relation names (relations linking sentences in source and target languages)**: - 1-to-1 relation: - sentence 5 in English = sentence 5 in French - many-to-1: - sentence 5 in English = sentence 5 in French - sentence 6 in English = sentence 5 in French - 1-to-many - sentence 5 in English = sentence 5 and sentence 6 in French **RULES:** 1- never change the English sentence IDs! - if 1-to-many: we merge the sentences in the target language, output = 1 line (target merge) - if many-to-1: we repeat the line in the target language, output = n of lines in source language - (target repeat) - if no match in target language: we do not have to do anything - if no match in source language: we append the last sentence **SCENE ANNOTATION:** - defined based on a single, coherent time unit (+ same location) - add speaker and scene ID in merged tab delimited format