Skip to main content
Version: 7.0.0

Analytics: Sentiment Analysis

Sentiment analysis reveals how customers feel about products, services, or other topics by classifying text as positive, negative, or neutral. This feature relies on BERT, a pretrained language model that captures contextual relationships in text.

Model Description​

This is a fine-tuned downstream version of the bert-base-uncased model for sentiment analysis. It is not intended for further downstream fine-tuning for other tasks. The model is trained on a classified dataset for text classification.

How To Use the Sentiment Analysis Component​

One of the unique features of the formsflow.ai framework is Sentiment Analysis. It can analyze sentiment from forms based on specific topics specified by the designer during form creation.

  • A form designer can drag and drop the Text Area with Analytics component and associate it with the corresponding workflow. This activates the Sentiment Analysis component.

    Sentiment Analysis

  • Based on the input responses of the user, formsflow.ai process sentiment is associated with each user's responses and the response will be patched to submission data by the bpm listener.

    • Refer section for more information on API used.

  • sentiment Analysis Flow is the workflow associated with sentiment analysis. You need to add the Java class in listeners as org.camunda.bpm.extension.hooks.delegates.FormTextAnalysisDelegate

    Refer to the sample shown below:

    Sentiment Analysis

note

Refer here for model creation and training procedures.

Data Analysis API​

The Data Analysis API is used to analyze customer sentiments, identifying whether they are positive, negative, or neutral.

POST API for sentiment analysis​

POST

{{DATA_ANALYSIS_API_BASE_URL}}/sentiment

HEADERS

Authorization Bearer {{token}}
Content-Type application/json

BODY

{
  "applicationId": "{valid applicationId}",
  "formUrl": "{valid formUrl}",
  "data": [{
    "text": "bad service",
    "elementId": ""
  }]
}

RESPONSE

{
  "overallSentiment": "NEGATIVE"
}