Keyword entities

The Keyword entity is the most commonly used custom entity type. It is based on common words that are manually categorized as needed.
Keyword entities are used primarily to store synonyms of particular terms or topics.

🪄

You can use AI to generate synonyms directly from the panel.

Name is general and points to a category of objects. It lets you refer to the whole categories directly in the flow.

Improved support in Polish language 🇵🇱

In Polish NLU words that are added as example words for keyword entity are subject to lemmatization. As a result other inflection forms of given word are also detected as this keyword entity. For verbs, this involves inflection by gender, person, number, tense, and mood, and for nouns by number and case. On the backend side it uses Morfeusz that creates lemmas, you can verify what forms are detected for user utterance in online demo.

To receive a result like on the image below, user has to add only one synonym to keyword entity (e.g. mieć).

There are some system's limitations, such as the lack of inflection for abbreviations (e.g., RJ45, rjtka, rjtce), the lack of inflection for proper nouns (e.g., Horizon, horizonem), and the lack of aspectual inflection for verbs (perfective and imperfective). What does this mean? If we add verbs to a keyword entity, we can add them in the infinitive form only, but we must remember to include both the perfective and imperfective forms of the verb.

To receive a full inflection of polish verb 'czytać' (eng. read), like on the image below, it is necessary to add those verb in perfective and imperfective form.