# How detect number entities written as words?

**URL:** <https://forum.rasa.com/t/how-detect-number-entities-written-as-words/14382>\
**Category:** Rasa Open Source\
**Created:** [August 15, 2019, 3:58pm UTC](https://forum.rasa.com/t/how-detect-number-entities-written-as-words/14382 "2019-08-15T15:58:15Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![bhargavat](https://avatars.discourse-cdn.com/v4/letter/b/90ced4/32.png) [@bhargavat](https://forum.rasa.com/u/bhargavat)\
**Post date:** [August 15, 2019, 3:58pm UTC](https://forum.rasa.com/t/how-detect-number-entities-written-as-words/14382/1 "2019-08-15T15:58:15Z")

</div>

I’m trying to detect time period and dollar amount entities (separately) regardless of format. If I have ninety five days, my entity model only recognized the “five days” part. Additionally, I would like to detect dollar amounts in the following formats: $4 thousand, $4.3k, $4 million, forty six million, $4M, $4 billion, $4.6B. How do I detect something like this that comes in many formats.

---

<div class="post-metadata">

**Author:** ![gcgloven](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/gcgloven/32/3915_2.png) [@gcgloven](https://forum.rasa.com/u/gcgloven)\
**Post date:** [August 16, 2019, 4:01am UTC](https://forum.rasa.com/t/how-detect-number-entities-written-as-words/14382/2 "2019-08-16T04:01:17Z")

</div>

You can try out duckling entity extractor in your pipeline

> **[Components](https://rasa.com/docs/rasa/components/)**

These are the type of entties duckling supports

 ![Screenshot%20from%202019-08-16%2011-57-36](https://europe1.discourse-cdn.com/flex013/uploads/rasa/original/2X/9/93882add8dce6b10e4dfc5c5c96d5c03b5ada185.png)

---

<div class="post-metadata">

**Author:** ![gcgloven](https://dub1.discourse-cdn.com/flex013/user_avatar/forum.rasa.com/gcgloven/32/3915_2.png) [@gcgloven](https://forum.rasa.com/u/gcgloven)\
**Post date:** [August 16, 2019, 4:02am UTC](https://forum.rasa.com/t/how-detect-number-entities-written-as-words/14382/3 "2019-08-16T04:02:38Z")

</div>

Worst case come to worst. You may write your own regex to extract the specify entity format … or you can develop a sophisticated story logic to read such entities as slots.