Document(Word, PDF) or Web URL based search


(Karthik Nagulapally) #1

Hi Team,

Do we have feature to upload the documents to Kore.ai or providing Web URLs to search for the response according to the user utterance.

Following are the platforms providing the above service,

  1. DialogFlow Knowledge beta
  2. IBM Watson Discovery Services
  3. AI Q&A System using TensorFlow and NLP

Example: [http://deeplearninganalyticsdemos-564769043.us-east-2.elb.amazonaws.com](http://Q&A Demo)dialogflow%20knowledge%20beta


(Madhu) #2

Hello @knagulapally

Kore.ai bot platform is working on a service to extract Knowledge from PDFs and crawling Web URLs. The feature is currently in development and will be in beta in Q4’2018. The Knowledge extraction service will be part of the product early Q1’2019 and subsequently other file formats will be added to the Extraction service.


(Karthik Nagulapally) #3

Hello @Madhu,

Thank you for the response.


(Vandana Agarwal) #4

Hello,

i was trying Kore.ai service to extract knowledge from PDFs but every time it gives me response as ‘Failure’.

My PDF Link : https://github.com/vandana992/Docs/blob/master/Sample_PDF.pdf

Please suggest me what format of PDF should i use so that it can be able to exract.


(Yoga Ramya Mendu) #5

Hi @vandana.agrawal,

The knowledge extraction feature expects the PDF file to have “bookmarks” for its Questions/headings to get processed successfully.

From further releases, the PDF will get processed even without Bookmarks.

As of now, we request you to add bookmarks for the content in your PDF.

Please find a sample PDF that gets imported into Knowledge extraction.
http://www.jointaccreditation.org/sites/default/files/PDF%20Bookmarking%20revised.pdf

Let us know if you need any clarification on the above.

Regards,
Yoga Ramya.


(Vandana Agarwal) #6

Hi @yogaramya.mendu,

We will try the suggested feedback and get back to you if catch any problem.
Thanks for the response.