Showcase the Automatically Extracted Leaderboard Data from AI Domain Papers for ORKG Users
(The issue is based on preliminary discussion with @aoelen and @Kabongosalomon)
Context: The ORKG will have an automatic information extraction model to process Artificial Intelligence papers in bulk. For each paper, the current trained model will run on the full-text of the paper and will extract three information points: viz., Task(s) attempted in the paper, Dataset(s) used in experiments, and Metric(s) used to evaluate the results. This model will be extended to include other information points, but the first model has three stated ones.
An example use-case of the model: we run it offline in bulk on all arxiv AI papers and upload the extracted information to the ORKG using the mass upload Python function.
Feature: However, to make it interesting for users to interact with the data and with the ORKG, we discussed the following feature: To make a change in the Frontend for the AI domain such that the information is organized along the lines of the https://paperswithcode.com/sota or something similar.
Potentially then this data uploaded in the ORKG can be leveraged in crowdsourcing experiments either from: a) general crowd; b) email the authors
Nevertheless, the discussed feature mainly targets new user interaction for the AI domain in the way they see the information a bit more transparently.