Motivation: While finding relevant literature is a challenge, deep reviews of technical articles can present another bottleneck for researchers. It is estimated that scientific trainees in degree programs read over 200 papers per year, devoting a whopping 23.5% of their time to this task. Moreover, while senior scientists can more easily digest technical vocabulary in their own domain, venturing into other specialties can cause cognitive overload. Thus, conversational AI stands to greatly speed and enhance the review process for junior and senior scientists alike.

Enter, DocChat! DocChat ingests one or more PDFs of choice, and allows researchers to narrow the scope of their work by asking specific questions about the content of the article(s).

Follow the steps below to try DocChat as an assist in your research process.

1. Navigate to the DocChat application, and determine whether PDFs of interest will be uploaded from a local or remote location. A URL may also be used for web-based PDFs. We will use a URL from the following review, “Scientific Literature, information overload.”

2. Simple select ‘Webpage’, then add the chosen URL. Click ‘Add’.

   

3. Once the document has been uploaded, navigate to the right hand dashboard and query the document.

   

4. DocChat will return both a semantic answer, and a citation for the document pages referenced in its answer. Review the original literature and re-prompt as necessary.

   DocChat continually improves. You can vote an answer up or down to help.