Entity extraction with system entities


This content is archived and is not being updated. For the latest documentation, see Microsoft Power Platform product documentation. For the latest release plans, see Dynamics 365 and Microsoft Power Platform release plans.

Enabled for Public preview General availability
Admins, makers, or analysts, automatically - This feature is released. Dec 2, 2019

Feature details

Power Virtual Agents can recognize, extract, and act on dozens of common entities in a user’s responses. For example, colors, currencies, ages, dates and times, names, telephone numbers, durations, cities, states, addresses, ZIP codes, email addresses, languages, weights, speeds, temperatures, organizations, percentages, points of interests, and links can all be recognized.

The system uses natural language capabilities to match a user’s responses to recognized entities. For example, if a bot asks a user to provide a city entity, it understands that the response, “I live in the emerald city” maps to the city of Seattle.

Moreover, the system validates these entities. For example, when the bot asks the user for a color, and the user gives an unexpected answer, the bot will repeat this question.

See also

Use entities and slot filling in Power Virtual Agents bots (docs)