Prototype: Standardized Sentiment and Emotion Output Schema#22
Open
aadi-joshi wants to merge 6 commits intoruxailab:mainfrom
Open
Prototype: Standardized Sentiment and Emotion Output Schema#22aadi-joshi wants to merge 6 commits intoruxailab:mainfrom
aadi-joshi wants to merge 6 commits intoruxailab:mainfrom
Conversation
Define a unified JSON schema for sentiment and emotion analysis results across text, audio, and facial modalities. Includes field descriptions, label normalization mappings, example outputs for each modality, and compatibility notes for the RUXAILAB frontend and Firestore storage.
Introduce sentiment_normalization package containing: - schema.py: Pydantic models (StandardizedOutput, ResultEntry, Segment) that enforce the v1 output schema with type validation - normalizer.py: converter functions for text sentiment (BERTweet), audio pipeline outputs, and facial emotion percentages Label normalization maps model-specific labels (POS/NEG/NEU, Angry/Happy/etc.) to a consistent lowercase vocabulary. Scores from 0-100 ranges are automatically converted to 0.0-1.0.
25 tests covering schema validation, label mapping, text sentiment normalization, audio pipeline conversion, facial emotion conversion, edge cases (unknown labels, long text truncation, error utterance skipping, percentage-to-ratio conversion), and label map completeness.
JSON file showing how a multi-utterance audio transcript sentiment result looks when converted to the v1 standardized schema, with segment timestamps, normalized labels, and a task identifier.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
This PR proposes a simple standardized output format for sentiment / emotion results.
Right now different parts of the ruxailab stack return slightly different structures. for example the text sentiment endpoint returns POS/NEG/NEU + confidence, while the facial sentiment API returns emotion percentages. that makes it harder to store or compare results across modalities.
This adds a small prototype for a shared schema that could represent outputs from text sentiment, audio transcript sentiment, and facial emotion analysis.
Included in this PR:
docs/emotion_output_schema_v1.md
proposal for a common JSON structure + field descriptions
normalization/schema.py
pydantic models for validating the schema
normalization/normalizer.py
helper functions to convert existing outputs into the schema (text sentiment, audio pipeline output, facial emotion percentages)
examples/audio_sentiment_standardized_example.json
example output showing segment-level audio sentiment
tests/unit/test_normalization.py
tests for the schema + normalization helpers
All changes are additive and existing endpoints are untouched.
Mainly opening this as a starting point for discussing how sentiment/emotion outputs could be standardized across services. If there's already a preferred structure for outputs like this in the project, happy to align this with it.