AnalyzeSentiment Processor

Part of the GCP Natural Language processor family

This processor takes in an input string and returns information about the sentiment of the text. Sentiment can be thought of as the connotation of the message. This processor will analyze the sentiment of individual sentences as well as the text overall.

The resulting Sentiments are reported with two values: score and magnitude. Per the official documentation:

score ranges from -1.0 (negative sentiment) to 1.0 (positive sentiment) magnitude indicates the overall strength of the emotion (both positive and negative)

This means that, for example, a sentence with a sentiment that has a score of 0.8 and a magnitude of 1.7 means it's a very strongly positive message. Similarly, a sentiment with a score of -0.2 with a magnitude of 0.2 is likely a "lightly" negative message.

The example output is a great example of the strength / sentiment. The message is overall a very positive message, and the output reflects as such.

Properties

All of our Natural Language processors also include these common properties.

This processor does not have any unique properties outside of the common ones.

Data Output

Output Format

Field Name	Data Type	Description
sentences	array of Sentence	The list of all sentences in the message
language	string	The language code of the language the message is in.
overallSentiment	Sentiment	The sentiment of the message as a whole

Relevant Data Structures

Sentence

Field Name	Data Type	Description
text	string	The raw string of the sentence
sentiment	Sentiment	The sentiment of this sentence
beginOffset	int	The number of characters from the beginning of the input string to the start of this sentence

Sentiment

Field Name	Data Type	Description
score	float	A number ranging from -1.0 (negative sentiment) to 1.0 (positive sentiment)
magnitude	float	A number ranging from 0 to +inf representing the absolute magnitude of the sentiment (regardless of score). This number can be thought of as the strength of the emotion (e.g., something being very positive, or slightly negative)

Example Output

{
   "output":{
      "overallSentiment":{
         "score":0.5,
         "magnitude":1.7
      },
      "sentences":[
         {
            "sentiment":{
               "score":0.6,
               "magnitude":0.6
            },
            "text":"James Adams is a fantastic developer who has been on the scene for over ten years."
         },
         {
            "sentiment":{
               "score":0.9,
               "magnitude":0.9
            },
            "text":"His excellent attitude and steady work ethic contribute to his character."
         },
         {
            "sentiment":{
               "score":0.2,
               "magnitude":0.2
            },
            "text":"May he stay with our company for many years to come."
         }
      ],
      "language":"en"
   },
   "raw-input":"James Adams is a fantastic developer who has been on the scene for over ten years. His excellent attitude and steady work ethic contribute to his character. May he stay with our company for many years to come."
}