When we launched DocParse storage, we included a search API for vector, keyword, and hybrid search to more easily explore your documents. Using the Aryn DocParse SDK, you can quckly add search to your apps and document workflows.

For a more interactive experience out of the gates, we've now shipped our search UI in the DocParse console! Let's take a look.

Running a search query

Navigate to the Search UI using the left nav, and select your DocSet to search using the dropdown to the left of the search bar. Next, select your Search Type, which includes Hybrid, Vector, Keyword, and Lexical. Your results will vary, so it's important to choose the Search Type that best matches your use case.

Type in your search query, and hit enter! We'll use Hybrid Search to look for "Amazon Web Services revenue" in our DocSet containing quarterly earnings reports (some of which are from Amazon). DocParse search will return the most relevant chunks from the documents in your DocSet.

‍

‍

If you click on a chunk, you will see the full document and other relevant chunks in that document in the Document Viewer.

‍

Using filters for finer-grained search

In my example above, let's say I only wanted to retrieve earnings calls from 2024. Assuming you had extracted the year from each document using Extract Properties, you can filter on this metadata in my search query. To do this, you can click on the Filter button and add the desired property and value.

‍

‍

You can easily extract additional properties from the documents in your DocSet, and use those as filters in your searches. If I wanted to add the company name to each earnings report, I can use the Extract feature in the DocSet Explorer (under Storage in the left nav) to do so.

‍

‍

Once I have my new property, I can add a filter for company_name so I can only retrieve Amazon's earnings calls in my search.

‍

Get searching today!

This is available for all Aryn DocParse users, and we'd love to hear about how you're leveraging these features for your document workloads. Hit us up on Slack!