Our tips for successful document uploads

Stick to these tips and your uploaded documents can be used as an optimal knowledge base for text generation!

Tip 1: Be specific!

To ensure that the knowledge from your uploaded document can be used accurately for text generation, the more specific and focused the content in the document is, the better. What does that mean exactly?

A good example: You upload a document with specific SEO guidelines that are explained in a focused way in the document. There is no other information or content in the document. You then select the document for text generation where the SEO guidelines are to be taken into account.

Bad example: You upload a document in which, among other things, SEO guidelines are listed. However, in addition to the SEO guidelines, the document contains a lot more information about the company strategy, the next marketing campaign, etc. The document consists of a total of 50 pages, of which only 3 pages focus on the SEO guidelines. You then select the document for text generation, where only the SEO guidelines are to be taken into account.

Our document upload system works best when the text query matches the content of the uploaded document. The more irrelevant content (for the text query) is included, the poorer the expected precision and quality of the result.

Tip 2: Only text content is taken into account!

Although your uploaded documents can contain images, graphics and tables, this content cannot be processed. For example, if you want to upload a .docx file that contains content in tables, reformat the content into simple, non-formatted text content.

Tip 3: The less formatting in the document, the better!

Please note that the "optimal document" is one that is completely free of formatting. The simpler and less formatted the document, the better it can be read by neuroflash and then processed as background knowledge for text generation.

So if you experience that certain content does not seem to be recognized during upload (e.g. because neuroflash is not able to incorporate this information into generated texts during text generation), simplify the document as much as possible and upload it again.

Tip 4: Do not use scanned .pdf documents

Scanned documents in .pdf format are not recognized. So make sure that your .pdf uploads are native .pdf documents.

Tip 5: Uploaded document vs. manually created knowledge

In addition to uploading .doc, .docx and .pdf documents, you can of course still create knowledge manually by inserting text. When is what best?

The document upload is suitable if:

• there is a lot of content (long texts) that you want to store as a knowledge base (note the maximum file size of 5 MB)
• all content from the selected document is relevant for your text queries
• you want to interact with the content from the document and you want to find out, for example, what answers the document has to your questions via your text queries
• if you want to work even faster and make your life even easier 😊

Adding knowledge manually is useful if:

• the knowledge is processed in documents that do not comply with the upload guidelines (e.g. incorrect format, file size too large)
• the knowledge you need is only a small part of a much larger document (see tip 1)