Generate Natural Language Query From Files

For natural language searches, NLPatent can now draft a query from uploaded documents. This is especially helpful when the relevant information is scattered across several documents, or when an invention disclosure already exists.

At least one file containing relevant information is required. Most common formats are supported (CSV, DOCX, GIF, HTM, HTML, JPEG, JPG, JSON, MD, PDF, PNG, PPT, TEX, TEXT, TXT, WEBP, XML)

How to Create A Natural Language Query from Files

1. Start a search

Start a search and navigate to the Natural Language option. Then in the bottom left corner, click Attach Files

2. Attach the files

Several files can be uploaded at once, and formats can be mixed (a PDF and a DOCX, for example). All uploaded documents are treated equally when the query is generated, so the order doesn't matter. Once the files are attached, click Generate from Files.

3. Generate + Review Query

It will then take 10-20 seconds for the query to generate, all uploaded content will be considered to generate

Once ready, the query appears in the query field. Reviewing before running is recommended, since the AI can make mistakes and small wording tweaks often improve results. The query field can be edited directly.

5. Run the search

When the query is ready, click Start Search and proceed as normal in the search interface.

Once finished, you will see the query in the query field. Read through the query to see if it makes sense, and you can edit the wording directly if you'd like.

Troubleshooting

"Not enough context" message. The AI couldn't find enough to work with. Options: add another file, upload a longer document, or write the query manually.