HARMONI Sentiment

Sentiment analysis service to get whether a text has a positive or negative sentiment using the huggingface transformers library.

You can run this module in the harmoni_full container.

Usage

The following documentation refers to the sentiment request.

The API for Sentiment has:

  • Request Name: ActionType: REQUEST

  • Body: data(str): input text that you want to run the sentiment analysis

  • Response:

    • response (int): SUCCESS, or FAILURE

    • message (str): response from the sentiment detector (e.g., )

You can run the service as follows: roslaunch harmoni_sentiment sentiment_service.launch

Parameters

Parameters input for the sentiment service corresponds the the $instand_id_param which includes:

Parameters

Definition

Value

model_name

model name of the huggingface library for sentiment analysis

str; “roberta-base”

Change the model_name in config/configuration.yaml to desired model. Default is ‘roberta-base’.

Testing

Before running rostest ensure the model has been downloaded. The local test will save a wav file of the speech specified in the sentiment.test test_sentiment_input parameter in the temp_data directory.

Use the command rostest harmoni_sentiment sentiment.test which will detect hte sentiment from a sentence in the test input.

References

Documentation