---
title: "Rebuild Families, Duplicates, and Analytics after Overlays"
slug: "rebuild-families-duplicates-and-analytics-after-overlays"
updated: 2026-07-15T15:44:16Z
published: 2026-07-15T15:44:16Z
canonical: "docs.revealdata.com/rebuild-families-duplicates-and-analytics-after-overlays"
---

> ## 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.

# Rebuild Families, Duplicates, and Analytics after Overlays

The fields described in this document are listed with their display names followed by their SQL table names in capitals and between brackets. Any time a listed field is overlayed after the initial data load and subsequent indexes, the described processes must be executed to ensure the database is updated with the new information.

At its bare minimum, Reveal only needs unique Begin Numbers (BEGDOCs) to create database entries. Once the Begin Numbers are created, they cannot be modified or overlayed (except under exceptional circumstances, i.e. hybrid workflow).

Item ID and Parent ID are database fields generated by the system that must be always set to auto generated and should never be overlayed. If during the initial data upload Begin Number Attach (BEGGATTACH) is not provided, Parent ID must be unmapped, otherwise the data load process will fail.

As an example, a basic data load with no additional metadata and only natives and text would look like the below:

- **Item ID (ITEMID)** – Always set to auto generated.
- **Parent ID (PARENT_ITEMID)** – Always set to auto generated. It should not be mapped if Begin Number Attach (BEGATTACH) is not provided during the initial data load.
- **Begin Number (BEGDOC)** – Populated with a unique ID or control number for every document in the database. Duplicate entries will be skipped and not loaded.
- **Begin Number Attach (BEGATTACH)** - Populated with the parent’s Begin Number (BEGDOC) for all records in the database, including parents and standalone documents.
- **Exported Native File (EXPORTED_NATIVE_FILE)** – Populated with the paths to the available native files.
- **OCR Path (OCR_PATH)** - Populated with the paths to the available text files.

Once the initial data is loaded, and subsequent information is added by the means of an overlay, the processes below must be followed to ensure that the database is updated with the new information.

## Family relationships

### Required fields

| **Reveal Display Name** | **Reveal Field Name** |
| --- | --- |
| Begin Number Attach | BEGATTACH |

### Rebuild process

Rebuild family relationships after the overlay via **Project Admin** > **Rebuild Document Data**.

## Duplicates in Document Viewer relational pane and search expansion

### Required fields

| **Reveal Display Name** | **Reveal Field Name** |
| --- | --- |
| Duplicate ID | MD5_HASH |

### Rebuild process

Rebuild duplicate records via **Project Admin** > **Rebuild Document Data**

> [!NOTE]
> **Note**
> 
> The processes of rebuilding families and duplicates can be triggered at the same time under **Project Admin** > **Rebuild Document Data**. An automated index job is triggered after the job completes and no further action is required unless there is an indication the job failed after looking at the logs.

![](https://cdn.us.document360.io/3e21d801-ca9f-4c51-93db-9cbd32741f3d/Images/Documentation/Screenshot 2026-03-16 at 3.33.49 PM.png)

## AI Sync fields

See [Reveal Analytics Requirements](/reveal/docs/reveal-analytics-requirements) for a full list of fields required for AI sync.

## Rebuild Process

After the overlay process is completed, re-index metadata, OCR/Loaded and/or Extracted for all records in Review Manager (select the All Documents option, not unindexed/errors only). You can set the system to index based on a list of Begin Numbers (BEGDOCS) or select the full load file.

> [!NOTE]
> **Note**
> 
> The re-index of text is necessary to re-trigger a new AI sync build after the one-hour window. The Extracted index is not mandatory, but it is recommended because the system will try to extract detectable text from available natives and will send it to the AI in case there is no text file available in OCR Path (OCR_PATH).