Sentiment

Sentiment service can be used to analyze sentiment score of the conversation and manage the conversation according to latest sentiment score of the end user.

The steps to use the service:

1. Create a sentiment service integration by using UI (Integrations/Services/+New)
2. Connect that service integration to a project
3. Optional: Use "SentimentScore" variable in flow, if you want to change flow accordingly
4. For ended conversations, check Dashboard to get some insights about average sentiment of the conversations

To integrate sentiment analysis, it is sufficient to select the Sentiment service from the Integrations/Services section. There is no need to provide additional information; the service will be connected automatically. Once the service is integrated into the project, sentiment calculations will be activated for that specific project. This means that sentiment analysis will be enabled for the project, and sentiment scores can be calculated for user interactions without requiring any extra input or configuration.

Sentiment value can be tracked on;

1. Utterance level
2. Conversation level.

Utterance-level Sentiment Score: This refers to the sentiment value derived from analyzing the sentiment of user utterances within a conversation. It is calculated by evaluating the sentiment of each individual utterance and is stored as a system variable called SentimentScore.

The utterance-based Sentiment Score is calculated by analyzing the sentiment values of each user utterance. Here is a summary of how it is calculated:

• At the beginning of each session, the utterance-based Sentiment Score is set to 0.
• For each user utterance, the sentiment analysis service is used to determine its sentiment value.
• The sentiment value of the utterance is then added to the current Sentiment Score, updating its cumulative value.
• If the sentiment analysis service is not available or fails to provide a response, the Sentiment Score remains as null for that particular utterance.
• The Sentiment Score reflects the cumulative sentiment of all user utterances within the ongoing conversation.

This variable allows the scenario designer to create different designs based on the sentiment score. For example, they may choose to route a user with a negative sentiment score to a live agent or offer different promotions to improve their experience.

Conversation-level Sentiment Score: The conversation-level sentiment is used for displaying information on the dashboard. It represents the overall sentiment of the entire conversation and is often used to provide an overview or summary of the sentiment trend within a conversation. This aggregated sentiment score can be useful for monitoring and analyzing the sentiment patterns and trends across multiple user utterances within a conversation. It allows for a high-level view of the sentiment dynamics and can be utilized for reporting, visualization, or decision-making purposes on the dashboard.