Sentiment Analysis

Azure Cognitive Service for Language provides sentiment analysis and opinion mining as features, a collection of machine learning and artificial intelligence algorithms in the cloud to develop smart applications involving written language. These features help you discover what people think of your brand or topic by looking for clues of positive or negative sentiment in the text, and can associate it with specific aspects of the text.

Its function provides opinion tags (such as negative, neutral, and positive) based on the highest confidence score found by the service at sentence and document level. This function also returns confidence scores from 0 to 1 for each document and sentences within it for positive, neutral, and negative sentiment.

Text Analytics consists of the qualitative extraction of information from a text by using computational systems employing technologies such as machine learning. As companies began to generate large amounts of data from different business areas, they saw how much of this information was unstructured, as is the case with complaint forms and suggestions, emails from customers, post-attention surveys, among others.

Then, the question of how to use this data in a useful way encountered old theories and mathematical algorithms that had been archived for decades waiting for an increase in computing power. Thus begins the exploitation of Text Analytics as an indispensable tool for companies that want to take their customer experience to the next level.

Supported Languages

Language	Language Code	v3.x Compatibility	Starting from the version of the model v3
Simplified Chinese	zh-hans	✓	01-10-2019
Traditional Chinese	zh-hant	✓	01-10-2019
Dutch	nl	✓	01-10-2019
English	en	✓	01-10-2019
French	fr	✓	01-10-2019
German	de	✓	01-10-2019
Hindi	hi	✓	01-04-2020
Italian	it	✓	01-10-2019
Japanese	ja	✓	01-10-2019
Korean	ko	✓	01-10-2019
Norwegian (Bokmål)	no	✓	01-04-2020
Portuguese (Brasil)	pt-BR	✓	01-04-2020
Portuguese (Portugal)	pt-PT	✓	01-10-2019
Spanish	es	✓	01-10-2019
Turkish	tr	✓	01-04-2020

Integrating with Lynn

This functionality is integrated by adding a new extension module, which must add the parameters required for its correct functioning

Configuration Parameters of the Extension Module

• Expression to evaluate: You can add a text between single quotes or an entity with the text to evaluate.

• Language: language of the text to be evaluated

• Subscription Key: Enter the subscription key of the service

• Endpoint: Endpoint of the service

• Intent to execute in case of error

Operations on the Extension Module

• Create
• Update
• Delete

Testing

Service response (JSON)

    {
        ConfidenceScores:
        {
            Positive: 0.12,                
            Neutral: 0.23,
            Negative: 0.65
        },
        Length: 58,
        Offset: 0,
        Opinions: [],
        Sentiment: Negative,
        Text: I'm very upset with the service, the internet is very slow.
    }