The Dialog Manager dashboard provides analytics on the performance of the Sofi Virtual Agent.
The dashboard enables administrators to review the performance of the natural language model model by reviewing the chat logs generated by users interacting with the Sofi Virtual Agent.
The Chat log show the following information:
- Utterance - the initial utterance that the user entered.
- Response - the Initial response from Sofi
- Intent Match - the highest scored Entry Point
- User - the name of the user that entered the utterance
- Score - the matching score provided by the NLU model.
- Match - If the utterance successfully matched a Entry Point - ie had a score above the minimum threshold (80% by default).
A String based filter can be set (top right of Chat log list) to filter the list of utterances.
The chat logs only show the initial utterance that the user entered. It does not list the entire conversation. This information can be requested by submitting a request to Servicely support.
There are various gauges (Filter controls) that control the filter that is applied to the chat log. By selecting these filter controls, you create Filter that is visualised by the Filter Breadcrumb on the top left of the dashboard.
To reset the filter, click the Reset Filter (section of the Breadcrumb).
- Word cloud - A word cloud highlighting the terms most frequently used keywords from utterances
- Threshold - Percentage of utterances above or below the configured threshold
- Intent cloud A word cloud highlighting the Intents most frequently matched
- User cloud - A word cloud highlighting the most active users
- Date Range - located on the top right of the dashboard is a drop down list that provides various options to filter results based on date range
In addition to the Chat Log there are a number of other gauges that provide insights into the performance and usage of the Virtual Agent.
- Term breakout - A pie chart showing the relative frequency of words used in utterances
- Chat interactions/period - The number of chat conversations initiated per period
- Intent breakout - A pie chart showing the relative frequency of matched intents / entry points
- User breakout - A pie chart showing the relative usage by user
- Unique users/period - The number of unique users initiating chats per period
Updated about 2 years ago