> For the complete documentation index, see [llms.txt](https://docs.zingg.ai/zingg/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.zingg.ai/zingg/running-zingg/link-across-datasets.md).

# Link across Datasets

Run the link phase to match records across two separate datasets using your trained Zingg model.

{% hint style="success" icon="right-long" %}
`Link` and `Match` are two EQUAL operations. Both use the same trained model. Neither is a subset of the other.

* `Link` - match records across two datasets. Each record from the first source is matched with all records from the remaining sources.
* `Match` - find duplicates within one dataset.
  {% endhint %}

The link phase is used when you have two datasets that are individually duplicate-free but need to be matched against each other. Common use cases include reference data mastering, data enrichment, and linking records from different source systems.

Link uses the same trained model as match. The output structure is identical to match output, with one additional column: `z_source`, which identifies which source dataset each record came from.

### Output Fields

Link output contains all input fields plus four Zingg-generated columns:

* `Z_CLUSTER` - unique identifier shared by all records resolved as the same entity across both datasets. In Enterprise this is the persistent Zingg ID.
* `Z_MINSCORE` - lowest similarity score within the cluster.
* `Z_MAXSCORE` - highest similarity score within the cluster.
* `Z_SOURCE` - the source dataset each record came from. Use this to trace each resolved record back to its origin system.

{% tabs %}
{% tab title="Community" %}

### Python

```python
options = ClientOptions([ClientOptions.PHASE, "link"])
zingg = Zingg(args, options)
zingg.initAndExecute()
```

### CLI

```bash
./scripts/zingg.sh --phase link --conf config.json
```

{% hint style="success" icon="right-long" %}
The link config file needs two data sources defined. Sample config for link available at:\
`github.com/zinggAI/zingg/blob/main/examples/febrl/configLink.json`
{% endhint %}

### Read and View Output

```python
output = spark.read.csv(
    "/tmp/febrlLinkOutput",
    header = True
)
display(output)
```

{% endtab %}

{% tab title="Enterprise" %}

### Python

```python
options = ClientOptions([ClientOptions.PHASE, "link"])
zingg = EZingg(args, options)
zingg.initAndExecute()
```

### CLI

```bash
./scripts/zingg.sh --phase link --conf config.json
```

### Read and View Output

```python
# Read link output
output = spark.read.csv(
    "/tmp/febrlLinkOutput",
    header=True
)
display(output)
```

{% hint style="info" icon="right-long" %}
Enterprise link output includes Zingg ID instead of `Z_CLUSTER`, plus the `Z_SOURCE` column identifying the source dataset of each record.
{% endhint %}
{% endtab %}

{% tab title="Enterprise Snowflake" %}

### CLI

```bash
./scripts/zingg.sh --phase link --conf config.json \
--properties-file <location to snowflake.properties>
```

{% hint style="info" icon="right-long" %}
Enterprise link output includes Zingg ID instead of `Z_CLUSTER`, plus the `Z_SOURCE` column identifying the source dataset of each record.
{% endhint %}
{% endtab %}
{% endtabs %}

<details>

<summary><strong>When should I use <code>link</code> instead of <code>match</code>?</strong></summary>

Use match when your goal is to find duplicates within a single dataset. For example, identifying that two customer records in the same CRM represent the same person.

Use link when you have two separate datasets that are individually duplicate-free but need to be matched against each other. For example:

* Matching a supplier list from one system against a vendor master from another
* Enriching a customer list with data from a reference dataset
* Linking records from two different source systems before consolidation

Both match and link use the same trained model. They are equal operations, not a hierarchy. You do not need separate training for link.

</details>


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation by asking a question.

Perform an HTTP GET request on the following URL with the `ask` and `goal` query parameters:

```
GET https://docs.zingg.ai/zingg/running-zingg/link-across-datasets.md?ask=<question>&goal=<user_goal>
```

`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is what the user is ultimately trying to achieve, the reason they need the answer. Sharing it helps GitBook give you a better, more relevant answer. A goal is most helpful when it describes the outcome the user wants rather than restating the question. For example, with `ask=how do I create an API token`, a goal like `automate deployments from our CI pipeline` lets GitBook tailor the answer to that use case.

The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.