[ANN] ORKG ASK - Neuro-symbolic answering of research questions against 80M papers

Dear all,

 

We recently celebrated the 5th anniversary of the Open Research Knowledge
Graph (orkg.org) and launched a new exciting feature: ORKG ASK, which allows
to answer research questions against a corpus of 80 Million Open Access
publications:

 

https://ask.orkg.org

 

ORKG ASK is a neuro-symbolic question answering engine, which uses embedding
(Nomic) and a vector DB (Qdrant) for retrieving the most relevant
publications for a question and then uses an open-source LLM (Mistral 7b)
for synthetizing an answer to the question and extracting further
information. 

 

Users can use faceted-filtering, run custom extractions, build their own
library of papers and search queries, download the results as well as curate
and persistently store them in the ORKG.

 

More details and links to the code can be found here:
https://ask.orkg.org/pages/about 

 

Best,

 

Yaser, Allard and Sören on behalf of the ORKG development team