Analyze Customer Engagement

Analyzes customer service and support conversations such as transcripts of a conversation between a call center agent and a customer. There are 7 tones that can be identified:

Sad, Frustrated, Satisfied, Excited, Polite, Impolite, Sympathetic

Important:

Each conversation analyzed may contain at most 50 utterances.

Each utterance may contain at most 500 characters.

Properties

Common

• DisplayName - The display name of the activity.

Credentials

• APIKey - The API key generated for your Tone Analyzer instance on IBM Cloud.
• URL - The URL generated for your Tone Analyzer instance on IBM Cloud.

Input

• ConversationJSON - The conversation to analyze as a stringified JSON object having the following format:

  { "utterances": [ { "text": "Hi, I can't get your
            software to work.", "user": "customer" }, { "text":
            "OK, what seems to be the problem?", "user": "agent" }, ...
            ] }

• ConversationTable - The conversation to analyze as a DataTable with columns Speaker & Utterance. The Speaker column may contain only Agent & Customer. For example:

  Speaker	Utterance
  Customer	"Hi, I can't get your software to work."
  Agent	"OK, what seems to be the problem
  ...	...

  Important: Note that ConversationJSON and ConversationTable represent the same input but in different formats. The two are mutually exclusive.

Misc

• InLanguage - The language of the inputted text.
• OutLanguage - The language in which the outputted tones are written.
• Private - If selected, the values of variables and arguments are no longer logged at Verbose level.

Output

• Utterance Tones - An Array<SentenceTone> variable holding analyses of each individual utterance in the submitted conversation. SentenceTone variables contain:
  • SentenceNumber - The index (starting at 0) at which the described utterance appears in the conversation.
  • Text - The isolated text of the utterance.
  • ToneName - The name of the tone found in this utterance.
  • Confidence - A value between 0 and 1 indicating the Tone Analyzer's confidence that the tone selected accurately describes the utterance.