---
title: "AI Model Building - Filter Junk Emails"
slug: "ai-model-building-filter-junk-emails"
updated: 2025-02-06T20:55:49Z
published: 2025-02-06T20:55:49Z
---

> ## Documentation Index
> Fetch the complete documentation index at: https://docs.revealdata.com/llms.txt
> Use this file to discover all available pages before exploring further.

# AI Model Building - Filter Junk Emails

## Introduction

“Junk” emails are typically emails sent from marketing companies, as such it contains no value for reviewers and ideally should be excluded from batching out to reviewers at the early stage of review.

Reveal AI provides various options to detect and exclude “Junk” emails after data has been ingested into AI.

## Process Flow

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/218%20-%2001%20-%20Email%20Cull%20Model%20Process%20Flow.png)

## Workflow Steps

1. **Process Data** Data first needs to be ingested into the Reveal platform. Check front-end to confirm data has been successfully ingested and document counts match expectations.

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/218%20-%2002%20-%20Doc%20Count%20to%20verify.png)
2. **Add Models** Before running models, follow the steps below to create an AI-enabled tag for each of the following models:
  1. Advertisements and Promotions
  2. Out of Office
  3. Personal & Family Events
  4. Sports News
    1. Log in to Reveal.
    2. Click on the **Project Admin**button in the Navigation Bar.
    3. Click on the **Tags**menu option.
    4. With the **Tags**tab selected in the left pane, click **Add Tag and Choices**.
      1. Create a *Multi-Select*
      2. **Add new choice** to create 4 choices matching the models above.
      3. Make sure ***Prediction Enabled*** is selected for all choices.

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/218%20-%2003%20-%20Create%20Multi-Select%20Classifier%20Tag.png)

Click **Add** and this will create 4 classifiers on the **Supervised Learning** page. Refer to Reveal’s Knowledgebase on how to add an AI-enabled tag: [***Build & Configure a Classifier***](/reveal/docs/build-configure-a-classifier).
3. **Run Models** Follow the steps below to add AI Models to the corresponding classifier. Repeat this for all 4 classifiers created in the step above.
  1. Open the classifier’s **Edit Classifier** page.
  2. Under **AI Library Models** section, search for the corresponding AI Model and add it. Notice the latest model comes with “V1.5” as part of the model name. For example, to add the model for *Advertisements and Promotions*, select ***Advertisements & Promotions V1.5***.

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/218%20-%2004%20-%20Added%20Advertising%20Promotion%20Model.png)
  3. Click **Run Full Process** button that appears below upon selection to apply the selected model to the classifier.
4. **Run Search** Follow the steps below to create saved searches based on the classifier results above.
  1. Confirm all four classifiers have finished running and the Status is ***Ready***:

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/218%20-%2005%20-%20Check%20New%20Classifier%20Status.png)
  2. Create a saved search to find all documents scored above certain threshold, for example, 85 or higher. Note that this might need to be adjusted based on actual data:

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/218%20-%2006%20-%20Classifier%20Score%20Search.png)
  3. From the search results above, QC a few docs to confirm the quality of the results. Adjust threshold if necessary.

## Related

- [AI Model Building Workflow](/ai-model-building-workflow.md)
- [Evaluate an AI Model](/evaluate-an-ai-model.md)
- [Apply AI Model from Library](/apply-ai-model-from-library.md)
- [Migration Export – Local Discovery](/migration-export-local-discovery.md)
- [Find Stories Using Entities](/find-stories-using-entities.md)