DinoSARS

• 

  Add new Entities to narrow down your search

• 

  Look at the top Entities in the results

• 

  Choose your Data Collection

• 

  Look at the top Concepts in the results

• 

  Enter your question and some of the concepts you are interested

Inspiration

Since January 2020, there has been more than 23,000 COVID-19 papers published, and the number of publications is doubling every 20 days. Innovative solutions are needed to address this information overload.

What it does

DinoSARS is an AI-driven search engine that presents semantic metadata such as high-level concepts, named entities, and keywords with search snippets. Keywords can be used for search refinement to make another search from the results. Keyword queries can be combined with concepts and entities to zoom in on more targeted content.

How I built it

We developed the orchestration layer and web application that interfaced with IBM Watson AI services. The web interface is a single page application developed with Bootstrap and jQuery.

Challenges I ran into

Researchers and medical professionals don't want to use search engines and tools they are not familiar with.

Accomplishments that I'm proud of

Building the AI-driven knowledge engine.

What I learned

IBM Watson is a good foundation for an AI solution. But customization and development is required to meet requirements of end users.

What's next for DinoSARS

We want to put more efforts on the information extraction, especially on numeric values. We are also considering other user interfaces such as chatbots and graph visualizations.

Built With

• figma
• ibm-watson