Voting is closed. 16 liked.
Title: LODStories: Learning About Art by Building Multimedia Stories
LODStories is a web-based application that enables users to explore linked data about art and discover interesting connections between artists and artwork. After exploration, the application will generate summaries about the artists or artworks users have selected, and they can construct a narrative story by choosing related images or videos.
Team: This project was developed by a team of undergraduate students at the University of Southern California. The students were Bonnie Jia, Dipanwita Maulik, Jianliang Chen, Linda Xu, Yinyi Chen, and Yuting Liu.
Advisors: Craig Knoblock, Miel Vander Sande, Pedro Szekely
The Internet today contains massive amount of data, and is often used for education. However, whereas we learn by exploring how topics are connected, it’s often hard for us to sort through the Internet to find the most relevant and interesting information about a topic, and the way information is presented is often not engaging for learning. To address these problems, we have developed an openly available web application called LODStories that allows people to explore artists and their collections based on the concept of semantic web in a more human-friendly way. Through this approach, we believe LODStories can advance the field of art and cultural education and benefit both the teaching and the learning community.
Open Data Sets
We focus specifically on the data in the Linked Open Data cloud, using open data sets from DBpedia and data from the Smithsonian American Art Museum, which has been mapped to a similar ontology. The application allows people to learn about art, artists, and their connections while also constructing interactive multimedia stories. LODStories uses text from DBpedia and museum databases, images from the Google Search API, and videos from the YouTube API to allow users to create an educational video that can then be saved and shared. A machine learning algorithm was also applied to rank the most interesting connected topics and natural language processing techniques are used to ensure the data collected about the topics are in a human-readable format. LODStories helps turn the great deal of information regarding art and cultural education into an easily accessible, interactive and multimedia format to engage and educate the users.
Distribution and Intellectual Property
LODStories is a live application that is hosted on the web for public use and can be accessed via this link: http://lodstories.isi.edu.
The software is available on GitHub at: https://github.com/InformationIntegrationGroup/LinkedDataEduApp, published using an open source Apache 2 license.
In terms of next step on LODStories, we are working with the American Art Collaborative, which is a group of 14 museums and archives in the US and we plan to create a version of LODStories that will work with the data from all of these museums. Our next target datasets are the art collection data from Crystal Bridges, Aman Carter, Dallas Museum of Art, The Gilcrease Museum and the Smithsonian Portrait Gallery. We are also in the process of expanding the machine-learning model with user feedback. This will help optimize our predictions for the most interesting paths, and help users to explore more engaging topics through the paths. Incorporating the data from the museum will also expand users access to more images and videos and also increase the breadth and depth of the topics we cover.”