Skip to main content

A yes/no answer parser for OpenVoiceOS

Project description

ovos-yes-no-plugin

PyPI version License Build Tests

A heuristic yes/no answer classifier for OpenVoiceOS.

Given a question and a spoken or typed response, it returns True (yes), False (no), or None (unclear). It handles common variants (yeah, sure, nah), double negatives (not a lie → yes), and language-specific word lists for 20+ locales.

Role in OVOS

OVOS skills ask yes/no questions through OVOSSkill.ask_yesno() in ovos-workshop:

# ovos_workshop/skills/ovos.py
def ask_yesno(self, prompt: str, data=None) -> Optional[str]:
    resp = self.get_response(dialog=prompt, data=data)
    engine = self._get_yesno_engine()
    answer = engine.yes_or_no(question=prompt, response=resp, lang=self.lang)
    if answer is True:
        return "yes"
    elif answer is False:
        return "no"
    else:
        return resp   # pass-through when answer is ambiguous

_get_yesno_engine() loads the plugin named ovos-yes-no-plugin by default. This package provides that plugin. The active engine can be overridden per-skill in settings.json or system-wide in mycroft.conf:

# mycroft.conf
{
  "skills": {
    "ask_yesno_plugin": "ovos-yes-no-plugin"
  }
}
# skill settings.json  (overrides the system default for this skill only)
{
  "ask_yesno_plugin": "my-custom-yesno-plugin"
}

Install

pip install ovos-yes-no-plugin

Usage

Inside an OVOS skill

class MySkill(OVOSSkill):
    def handle_intent(self, message):
        answer = self.ask_yesno("confirm.action")  # plays TTS, waits for reply
        if answer == "yes":
            self.speak_dialog("confirmed")
        elif answer == "no":
            self.speak_dialog("cancelled")
        else:
            self.speak_dialog("didnt.understand")

ask_yesno returns "yes", "no", or the raw utterance when the engine returns None.

Direct use

from ovos_yes_no import HeuristicYesNoEngine

engine = HeuristicYesNoEngine()
engine.yes_or_no("Do you want to continue?", "yes", "en-us")          # True
engine.yes_or_no("Do you want to continue?", "no way", "en-us")       # False
engine.yes_or_no("Do you want to continue?", "beans", "en-us")        # None
engine.yes_or_no("Do you want to continue?", "it's not a lie", "en-us")  # True  (double negative)
engine.yes_or_no("Do you want to continue?", "yes, but actually, no", "en-us")  # False (last word wins)

Via ovos-plugin-manager

from ovos_plugin_manager.agents import load_yesno_plugin

cls = load_yesno_plugin("ovos-yes-no-plugin")
engine = cls()
result = engine.yes_or_no("Shall I set a reminder?", "please do", "en-us")  # True

Configuration

No mandatory configuration. An optional config dict is accepted by the constructor. The only supported key is lang, which sets the default language when yes_or_no is called without an explicit lang argument:

engine = HeuristicYesNoEngine(config={"lang": "de-de"})

Algorithm

Source: HeuristicYesNoEngine.yes_or_noovos_yes_no/__init__.py:71

The engine scans the normalised response against a per-language locale/<lang>/yesno.json resource file containing four word lists:

Key Role
yes Unambiguous affirmatives: yes, yeah, affirmative, ...
no Unambiguous negatives: no, nah, disagree, ...
neutral_yes Soft affirmatives counted only when no no word is present: sure, please, ...
neutral_no Soft negatives counted only when no yes word is present: wrong, mistake, lie, ...

Decision rules (in order):

  1. Last word wins. When both a yes word and a no word appear, the one at the higher character index is taken as the user's final intent.
  2. Double negatives. A no-category word immediately followed by a neutral_no word (e.g., not + lie"not a lie") is interpreted as affirmative.
  3. Neutral words. If neither a yes nor a no word was found, neutral_yes and neutral_no words are considered as weak signals.
  4. Default. No recognised word → return None.

Language matching uses langcodes.tag_distance to find the closest available locale, so en-AU silently falls back to en-US.

Supported Languages

an, az, ca-ES, cs-CZ, da-DK, de-DE, en-US, es-ES, eu-ES, fa-IR, fr-FR, hu-HU, it-IT, nl-NL, pl-PL, pt-BR, ru-RU, sv-SE, tr-TR, uk-UA

Adding a Language

Create ovos_yes_no/locale/<lang-TAG>/yesno.json with the four word lists described above:

  • neutral_yes: mild agreement words that are positive but not direct synonyms of "yes" (e.g., French "bien sur", Portuguese "claro").
  • neutral_no: words that imply disapproval indirectly (e.g., French "mensonge", Portuguese "errado").
  • Double-negative structures vary widely between languages — test them explicitly.

Limitations

  • No sarcasm or idiom detection.
  • Vocabulary is limited to the words in the resource files; unlisted slang is ignored.
  • Complex nested negations beyond one level may yield incorrect results.
  • Missing or incomplete resource files cause the engine to return None for that language rather than raising an error.

License

Apache 2.0

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

ovos_yes_no_plugin-0.3.0.tar.gz (18.2 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

ovos_yes_no_plugin-0.3.0-py3-none-any.whl (18.2 kB view details)

Uploaded Python 3

File details

Details for the file ovos_yes_no_plugin-0.3.0.tar.gz.

File metadata

  • Download URL: ovos_yes_no_plugin-0.3.0.tar.gz
  • Upload date:
  • Size: 18.2 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for ovos_yes_no_plugin-0.3.0.tar.gz
Algorithm Hash digest
SHA256 f13f40b347fec4fca5ffd9c56bf564dcfa94c26b09545f3f3db0f720c8338c43
MD5 51fff53b268903869e2e04b79a761547
BLAKE2b-256 fc7cd115deed612b05d117ed6848e86dfeaf338440dd90e9bdc3082fb570b637

See more details on using hashes here.

File details

Details for the file ovos_yes_no_plugin-0.3.0-py3-none-any.whl.

File metadata

File hashes

Hashes for ovos_yes_no_plugin-0.3.0-py3-none-any.whl
Algorithm Hash digest
SHA256 36b2330b8c8fe55cc7f2770459b82acda0582542f164fa6eea75340aa0ff02b6
MD5 f3fa892c3d653d865dd66cf014087a1c
BLAKE2b-256 583d209cd9813b7b896112751624ecc0df2a29900ee64eb58c211cd98c856011

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page