Skip to content

Unstructured

Unstructured converts documents into labeled pieces of text and layout that an app can index.

Visit Unstructured

What can document parsing add before indexing?

Partitioning separates a document into pieces such as titles, paragraphs and tables. That gives your app material to check before it creates vectors. Preserve the filename and page information where available and compare the result with a sample of your actual layout.

When does it cost money and what should you keep?

The open-source library can be run by your team. The hosted pricing page counts processed pages and describes an initial allocation. Check any outside processing before using private files. Keep originals, processed pieces and the parsing configuration. A separate tool stores or searches the vectors.

Did the fictional table keep its meaning after parsing?

Draw a sample manual with a heading and two table rows.

  1. Compare the processed pieces
    Mark which heading applies to the table and where the source page is recorded.
  2. Choose what will be indexed
    Only send the intended public passage and its source label into the paper search plan.