README.md

metadata

language:
  - ca
  - cs
  - da
  - de
  - el
  - en
  - es
  - eu
  - fa
  - fr
  - gl
  - hu
  - it
  - nl
  - pl
  - pt
  - ro
  - ru
  - sv
  - tr
task_categories:
  - text-classification
pretty_name: OpenVoiceOS Localize — Intent Classification Dataset
license: apache-2.0
size_categories:
  - 100K<n<1M
tags:
  - ovos
  - voice-assistant
  - intent-classification
  - nlu
  - multilingual
source_datasets:
  - OpenVoiceOS/OVOSGitLocalize-Intents

OpenVoiceOS Localize — Intent Classification Dataset

Multilingual intent classification corpus exported from OpenVoiceOS/ovos-localize.

Each row is a single expanded utterance labelled with the OVOS skill and intent file that produced it. Templates are fully expanded (bracket alternation resolved); {slot_name} placeholders from .intent files are kept verbatim so models can learn the slot-carrying pattern.

Schema

Column	Description
lang	BCP-47 locale code (en-US, pt-PT, …)
domain	OVOS skill id (ovos-skill-hello-world, …)
intent	Source filename (HelloWorldKeyword.voc, HowAreYou.intent, …)
sentence	Expanded, cleaned utterance text

Languages

• ca-ES
• cs-CZ
• da-DK
• de-DE
• el-GR
• en-US
• es-ES
• eu-ES
• fa-IR
• fr-FR
• gl-ES
• hu-HU
• it-IT
• nl-NL
• pl-PL
• pt-BR
• pt-PT
• ro-RO
• ru-RU
• sv-FI
• sv-SE
• tr-TR

Data source

Generated daily by the ovos-localize data pipeline from live OVOS skill repositories.

Collection

Part of the OpenVoiceOS intent-classification-datasets collection.