Save "Sefaria's New Enhanced Topic Pages"
Sefaria's New Enhanced Topic Pages
Sefaria’s team, which has been conducting text analysis using machine learning and natural language processing tools for several years, spent the last year using new AI tools to create more resources to help users find and explore more texts by topic rather than searching the library by book title or author.
The basis for this work was developed by Sefaria’s learning team, who are all seasoned scholars, editors, and translators, in collaboration with a team of engineers dedicated to this specific project.
First, the learning team curated 150 topic pages manually. They explained their process in detail to the engineering team, who then painstakingly translated the frameworks and criteria with which they were working into a series of prompts for the AI algorithm. In addition to learning the texts in the library and taking in hundreds of examples of excellent topic pages, the algorithm was also introduced to other important elements for discernment, such as historical context and interconnections between different bodies of work.
After months of development, the teams curated 1,000 topic pages. Of the 15,500 introductory paragraphs for sources, 14,000 were written by AI and edited by Sefaria team members. Currently, the set of Enhanced Topic Pages includes 10% written by humans and 90% generated by AI.

Anatomy of an Enhanced Topic Page

We use cookies to give you the best experience possible on our site. Click OK to continue using Sefaria. Learn More.OKאנחנו משתמשים ב"עוגיות" כדי לתת למשתמשים את חוויית השימוש הטובה ביותר.קראו עוד בנושאלחצו כאן לאישור