Congratulations to our PHd student Amir David Cohen Nissan for being recognized for the best article award at the AKBC 2022 conference

The subject of the article:

Supervised Relation Classification as Two-Way Span-Prediction

Summary:

Extracting connections from text is a task where the input is a segment of natural language and 2 entities (for example, "the United States sent aid to Ukraine", "the United States", "Ukraine"), and the output is the connection between the entities. (in our case, "allies").

Extracting connections is a well-known task used by many systems in academia and industry.

In our research, a new approach to entity extraction was presented, based on predicting parts of the text, instead of by classification as is done today. This method achieves better results compared to the methods that exist today.