How to use the More Like This feature

This page explains how to create MoreLikeThis queries. These are used to find documents similar to a given document.

A typical use case is a e-commerce or news web site suggesting similar products or content to visitors looking at an article.

First, create a new query in the Query tab. Choose the "More like this" type.

Creating new more like this query

This query's point is to return documents similar to another. This baseline document must be identified in the Document query field. Assuming a schema with a field named product_id, we could for instance have a query for product_id:345. This query must be written in the usual query pattern format.

The field Like text must be left empty.

Select the language of your documents in the Language list.

Now you must designate the field(s) that will be used to find similarities between documents. This is done in the Fields area.

In the Analyzer list, pick the same analyzer as used for this field in the index's schema.

Configuring more like this query`

The second tab allows for the configuration of some metrics. These default values are fine to start with.

Setting metrics

At last one returned field must be configured:

Configuring returned fields

Here is another example, where products are identified in a field named numkey_prod. Similarity is computed using the field nom_genre, which lists product categories.

More like this results

Of course this query template can be used in an API call, where "Document query" can be easily overiden. Have a look at the More Like This API section of our documentation.

View/edit on GitHub

comments powered by Disqus