Azure QnA Maker Need Advice on Data Source

Question

We are trying to create an FAQ chat bot using the Azure QnA maker service. I have read the documentation which says that the allowed file formats for creating a knowledge base are PDF, word, xls, doc, tsv and txt. Our published product documentation is available in PDF form but this has too much information and we are not inclined on using this file to build the knowledge base in the QnA maker. Instead we were hoping to use our raw documentation files which are in DITA format.

Could you please advice us on what we should be doing in this case? Any suggestions or recommendations are welcome.

Thanks.

Answer 1

As you mentioned, you can import the following file types: PDF, word, xls, doc, tsv and txt

I would highly suggest that you transform your DITA documentation into tsv file, which is a simple question TAB answer (see a sample here). In that way, you will avoid potential errors that can be made when the PDF / Word / XLS / Doc files are imported, due to the specific syntax of each type.

With the TSV, you are the master of the right split of your questions / answers.