# Does textcat use NER entities as features?

**URL:** <https://support.prodi.gy/t/does-textcat-use-ner-entities-as-features/2319>\
**Category:** Uncategorized\
**Tags:** ner, textcat, spacy, solved\
**Created:** [December 14, 2019, 1:03pm UTC](https://support.prodi.gy/t/does-textcat-use-ner-entities-as-features/2319 "2019-12-14T13:03:07Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![edward](https://avatars.discourse-cdn.com/v4/letter/e/90db22/32.png) [@edward](https://support.prodi.gy/u/edward)\
**Post date:** [December 14, 2019, 1:03pm UTC](https://support.prodi.gy/t/does-textcat-use-ner-entities-as-features/2319/1 "2019-12-14T13:03:07Z")

</div>

Simple question, I hope.

Say I use a base model with NER to train a textcat compent, e.g.

`prodigy textcat.batch-train my_dataset en_vectors_web_lg --output my_model`

The resulting pipeline will be something like:

`["tagger","parser","ner","textcat"...]`

Will the `textcat` component use the extracted `ner` entities - e.g. `PERSON`, `NORP`, etc. - as features in any way? Or will it just use the text/syntax/vectors the same as if the `ner` wasn't even there?

---

<div class="post-metadata">

**Author:** ![ines](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/ines/32/3_2.png) [@ines](https://support.prodi.gy/u/ines)\
**Post date:** [December 16, 2019, 11:44am UTC](https://support.prodi.gy/t/does-textcat-use-ner-entities-as-features/2319/2 "2019-12-16T11:44:04Z")

</div>

> [@edward](#):
>
> Will the `textcat` component use the extracted `ner` entities - e.g. `PERSON` , `NORP` , etc. - as features in any way? Or will it just use the text/syntax/vectors the same as if the `ner` wasn't even there?

By default, spaCy's pipeline components are independent and don't share any features between them ([also see this docs section](https://spacy.io/usage/processing-pipelines#pipeline-components-order) on the pipeline component order). So whether the entity recognizer is there or not won't make a difference.

---

<div class="post-metadata">

**Author:** ![gitclem](https://sea2.discourse-cdn.com/flex020/user_avatar/support.prodi.gy/gitclem/32/2095_2.png) [@gitclem](https://support.prodi.gy/u/gitclem)\
**Post date:** [April 20, 2021, 7:27pm UTC](https://support.prodi.gy/t/does-textcat-use-ner-entities-as-features/2319/3 "2021-04-20T19:27:54Z")

</div>

What if I really want spaCy to use NER to improve text categorization?

E.g. what if I want these two sentences to be categorized differently? Obviously the NER recognizing GPE would be helpful:  
"We live in Denmark" vs. "We live in fear".

How can I get TextCat to take NER into consideration?