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

# Deduplication

> 4-layer duplicate detection to clean your paper library

When searching multiple databases, the same study often appears more than once. EviSyn uses a multi-layered approach that catches **95%+** of duplicates — including near-matches that traditional tools miss.

## Running Deduplication

Navigate to **Deduplication** from your project dashboard.

<img src="https://mintcdn.com/ravian/YwBZFdlBbtQ_-NDF/images/deduplication.png?fit=max&auto=format&n=YwBZFdlBbtQ_-NDF&q=85&s=220ddaac91f823693ebf6497f8685e7e" alt="Deduplication page with duplicate groups" width="1952" height="1760" data-path="images/deduplication.png" />

You have two scan options:

<CardGroup cols={2}>
  <Card title="Scan for Duplicates" icon="magnifying-glass">
    Fast scan using DOI matching, exact title matching, and fuzzy title matching (85%+ similarity). Takes seconds.
  </Card>

  <Card title="Run Semantic Analysis" icon="brain">
    Deep scan using AI-powered semantic similarity. Catches papers that describe the same study with completely different wording. Takes longer but finds more duplicates.
  </Card>
</CardGroup>

## The 4 Detection Layers

| Layer | Method          | What it catches                                                         |
| ----- | --------------- | ----------------------------------------------------------------------- |
| 1     | **Exact DOI**   | Identical papers with the same DOI across databases                     |
| 2     | **Exact title** | Same title with different formatting or capitalization                  |
| 3     | **Fuzzy title** | Titles with slight differences (abbreviations, typos) — 85%+ similarity |
| 4     | **Semantic**    | Papers describing the same study in completely different words          |

## Reviewing Duplicate Groups

Each group of potential duplicates shows:

* **Match type** badge (e.g., "Fuzzy Title Match")
* **Similarity score** for each paper (e.g., 100%, 88%, 85%)
* Full metadata: title, authors, journal, DOI

### Resolution Options

For each group, you can:

| Action                        | What it does                                                              |
| ----------------------------- | ------------------------------------------------------------------------- |
| **Keep This**                 | Designate this paper as the primary record                                |
| **Keep All (Not Duplicates)** | Mark all papers in the group as unique — they are not actually duplicates |

<Tip>
  Always check the DOIs and author lists before resolving. Fuzzy matches sometimes catch related but distinct papers (e.g., study protocol vs. results paper).
</Tip>

## What's Next?

Once all duplicate groups are resolved, proceed to [AI Screening](/user-guide/ai-screening) to screen your papers.