# How to add an inner field as a source field in ML inference pipeline?

**URL:** <https://discuss.elastic.co/t/how-to-add-an-inner-field-as-a-source-field-in-ml-inference-pipeline/328323>\
**Category:** Elasticsearch\
**Tags:** elastic-stack-machine-learning, ingest-pipeline\
**Created:** [March 23, 2023, 8:44am UTC](https://discuss.elastic.co/t/how-to-add-an-inner-field-as-a-source-field-in-ml-inference-pipeline/328323 "2023-03-23T08:44:52Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![848bb15cf6e891bef7ba](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/848bb15cf6e891bef7ba/32/118656_2.png) [@848bb15cf6e891bef7ba](https://discuss.elastic.co/u/848bb15cf6e891bef7ba)\
**Post date:** [March 23, 2023, 8:44am UTC](https://discuss.elastic.co/t/how-to-add-an-inner-field-as-a-source-field-in-ml-inference-pipeline/328323/1 "2023-03-23T08:44:52Z")

</div>

Hi,

I am trying out vector search. While creating a Machine Learning Inference Pipeline, I see that some fields in the index are not recognised. Only the top level fields are shown.

Consider the below example with field mappings:

```auto
{
  "mappings": {
    "properties": {
      "core": {
        "properties": {
          "propertyA::keyword": {
            "type": "keyword",
            "normalizer": "lowercase_normalizer",
            "fields": {
              "exact": {
                "type": "keyword"
              }
            }
          },
          "propertyB::keyword": {
            "type": "keyword",
            "normalizer": "lowercase_normalizer",
            "fields": {
              "exact": {
                "type": "keyword"
              }
            }
          }
       }
      "propertyC::keyword": {
        "type": "keyword",
        "normalizer": "lowercase_normalizer",
        "fields": {
          "exact": {
            "type": "keyword"
          }
        }
      }
    }
  }
}

```

Only propertyC is recognised while adding an inference pipeline. The fields core.propertyA and core.propertyB are not recognised.

If I edit the pipeline processor definition with the field core.PropertyB, the following error is seen while linking to the index:

> This pipeline cannot be selected because the source field does not exist on the index.

How do we configure the inference pipeline with such inner fields?

---

<div class="post-metadata">

**Author:** ![warkolm](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/warkolm/32/39224_2.png) [@warkolm](https://discuss.elastic.co/u/warkolm)\
**Post date:** [March 23, 2023, 8:50am UTC](https://discuss.elastic.co/t/how-to-add-an-inner-field-as-a-source-field-in-ml-inference-pipeline/328323/2 "2023-03-23T08:50:28Z")

</div>

Welcome to our community! 😃

Are you actually using App Search, or Elasticsearch?

---

<div class="post-metadata">

**Author:** ![848bb15cf6e891bef7ba](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/848bb15cf6e891bef7ba/32/118656_2.png) [@848bb15cf6e891bef7ba](https://discuss.elastic.co/u/848bb15cf6e891bef7ba)\
**Post date:** [March 23, 2023, 11:21am UTC](https://discuss.elastic.co/t/how-to-add-an-inner-field-as-a-source-field-in-ml-inference-pipeline/328323/3 "2023-03-23T11:21:18Z")

</div>

Hi @warkolm,  
I am using Elasticsearch

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\
**Post date:** [April 20, 2023, 11:21am UTC](https://discuss.elastic.co/t/how-to-add-an-inner-field-as-a-source-field-in-ml-inference-pipeline/328323/4 "2023-04-20T11:21:58Z")

</div>

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.