We are incredibly proud of the winning teams who produced cutting-edge digital tools that will open access to the National Library's treasures in new and creative ways!
On November 23-24, some 150 programmers, project managers, web developers, and others huddled together at the Library for 24 hours with one purpose:
To make the Library’s treasures as accessible as possible.
The participants of the Hackathon had two technological challenges with which they had to contend. The first was the Creative Challenge, in which teams had to come up with, creative ways to make the Library’s treasures as accessible as possible. The second was the Data Challenge, where teams had to develop tools to improve the metadata of items kept in the Library’s collections.
The winning independent team was Team Pi! They created an interactive educational tool that enables children and pupils to do associative research through video and audio files, bringing to life historical items from various collections in the Library, making the items and subjects being researched tangible and accessible.
The tool itself is based on several AI and Machine Learning Technologies. These introduce new types of information to the library, improve access to existing information, and create innovative product opportunities. These technologies will develop new points of access for new audiences and improve the research capabilities of the library’s materials. The technologies are based on voice recognition from visual and audio files, pattern recognition in other kinds of files as well as ranking them by priority. Over time the tool will be able to find connections between items through the AI and Machine Learning patterns, as well as user based contexts placing them together.
Team Pi’s presentation is available here.
The demo for the tool is here.
The winning team from a company or organization was Wikimedia Israel. They created a data tool that will identify people in historical pictures even when the picture itself is lacking a description of the figures in it. This tool will help search and find famous and historical people in the Library’s collections.
This tool will enable Wiki editors to search and locate images of historical value from a wide variety of databases. In addition, a tool was developed for Wikitext that can splice scanned manuscripts into sections made up of several words or single lines, specifically for objects that don’t cooperate with OCR software. Then with the help of a chat-bot, editors can upload these sections to Wikitext via Telegram or Facebook. This tool will enable researchers and readers to search for manuscripts that have been scanned but haven’t been fully digitized.
Team Wikimedia’s presentation is available here.