analyzed-conversation-viewer

analyzed-conversation-viewer

analyzed-conversation-viewer is a UI component that hel you display a conversation analysis through multiple aspects like the conversation sentiment, conversation personally identifiable information (PII), and conversation summary

the target of this component is to gather the analysis resulting from the different services in one place to give the user the needed information for each sentence and the overall conversation.

Props

Name	Type	Description	Default value
result	{ sentences: SentenceType[] }	the main object to display
labels	{ onText?: string; offText?: string; hidePii?: string; piiDescription?: string; sentimentHeader?: string; transcriptHeader?: string; sentimentDescription?: string; }	Optional - pass the labels you want to be displayed instead the default ones	{ onText: 'On', offText: 'Off', hidePii: 'Hide PII', sentimentHeader: 'Sentiment', transcriptHeader: 'Transcript', piiDescription: 'Sensitive information across several pre-defined categories is detected and redacted from the transcript.', sentimentDescription: 'Sensitive information across several pre-defined categories is detected and redacted from the transcript.' }

## SentenceType

Name	Type
text	string
speakerTitle	string
redactedText	string
entities	NamedEntity[]
sentiment	SentimentTypes
duration	{ minutes: number; seconds: number; }

## NamedEntity

Name	Type
text	string
offset	number
length	number
category	string
subcategory	string
confidenceScore	number

## SentimentTypes

SentimentTypes = 'positive' | 'negative' | 'neutral'

Examples

Here is a usage example:

const conversationResult = {
  sentences: [
    {
      sentiment: "positive",
      speakerTitle: "Speaker 1",
      duration: { minutes: 0, seconds: 2 },
      text: "Hi thank you for calling Bank of Contoso. Hello my name is Mary. How can I help you today?",
      redactedText: "Hi thank you for calling Bank of Contoso. Hello my name is ****. How can I help you today?",
      entities: [
        { category: "Person", confidenceScore: 0.9, length: 4, offset: 59, text: "Mary" },
        { category: "Organization", confidenceScore: 0.9, length: 15, offset: 25, text: "Bank of Contoso" },
      ],
    },
    {
      entities: [],
      sentiment: "neutral",
      speakerTitle: "Speaker 2",
      duration: { minutes: 0, seconds: 2 },
      text: "I would like to know my remaining balance in my checking account.",
      redactedText: "I would like to know my remaining balance in my checking account.",
    },
  ],
};

return <AnalyzedConversationViewer result={conversationResult} />;

and the view will be like:

the nice annotations are made by the annotator library

Related topics

• Conversation sentiment analysis
• Conversation summarization
• Conversation PII
• react-text-annotator