What are Entities?

---

Entities play a major role in language understanding. To perform an action on a certain user query you not only need to understand the intent behind it but also the entities present in it. E.g., if someone says "flights from Berlin to London", the intent here is flight-search and entities are Berlin and London, which are of type city.

In a given piece of text, entities can be anything from names, addresses, account numbers to very domain-specific terms like names of chemicals, medicines, etc. Essentially any valuable information that can be extracted from text.

Entities can also be seen at a more granular level. In the above example Berlin can be from-city and London can be to-city. A very domain-specific use-case could be, e.g., "I need 8 paracetamol tablets", where 8 is number, and paracetamol is medicine-component, and tablets is medicine-form.

Supported Entities on NeuralSpace

To be able to extract a wide variety of entities we provide five different types of entities.

Pre-trained Entities

We have a catalog of over 30 entity extractors like, person, date, number, geo-location, etc. that you can use off-the-shelf.

• 💻 Learn how to use our pre-trained entities
• 👉 Checkout our pre-trained entities catalog here

Regular Expressions

To extract entities that follow a certain pattern, you can use our in-build regular expression entity extractor.

• 💻 Learn how to use regular expression entities

Lookup Entities

If you have a finite list of things you want to extract from a given piece of text, e.g., list of languages, location landmarks, company departments, etc. you can use lookup entities. These are essentially a list of words or phrases that you want to extract from a given text.

• 💻 Learn how to use lookup entities

Synonyms

Same entities can sometimes come in different forms. E.g. New Delhi and Delhi, New Your City and NYC. Cases where you are performing an action based on the entity value, you will have to make them homogeneous. That means, you would want your NLU model to return a single value (Delhi) for both New Delhi and Delhi. This can be done using synonyms.

• 💻 Learn how to use synonym entities

Trainable Entities

If you want AutoNLP to learn to extract your custom entities, you can do that too.

• 💻 Learn how to use trainable entities

Tip

Make sure to try all the above methods to extract entities as they are more deterministic. AutoNLP trains an AI model to predict entities and that can be nondeterministic.