Bar-Ilan University | President’s Report 2022

52 Data science innovation since 2017 Most of the time, Hebrew’s succinctness is a perfect fit for the faced-paced, straightforward culture for which Israel is known. Why, after all, waste time on a full sentence when you can express the same thing in just one word? Yet as technology increasingly becomes a pervasive part of Israeli life, its root-based language has revealed a downside. Natural language processing (NLP), a subset of artificial intelligence (AI) and data science, enables computers to understand both text and voice data, and to respond with text and speech of their own. Used, for example, in search engines, autocorrect, and by Apple’s beloved Siri, NLP is the future of human-machine interactions. But as of now, that future doesn’t speak the ancient tongue. NLP’s infrastructure—its theory of howto understand language and the associated algorithms—has largely been developed for English, a predominantly linear language. In Hebrew, by contrast, every word includes multiple pieces of information, which for machines can pose a particular conundrum: To understand aword, theymust first understand the sentence— but to understand the sentence, they need first to understand theword. Unless computers succeed in doingwhat for Hebrew speakers comes naturally, tomorrow’s Israelis will need to speak English, or else risk getting left behind. In keeping with its mission as Zionist institution, and thanks to its standing as a national leader in AI and data science— established in 2017, the University’s interdisciplinary Data Science Institute includes nearly 70 researchers from 11 departments—Bar-Ilan has taken up the challenge of developing an infrastructure for Hebrew. Headed by Prof. Reut Tsarfaty, theworld’s top researcher inHebrewNLP, theNatural Prof. Sarit Kraus The Department of Computer Science