Saallen
commented
9 years ago We appreciated this submission but unfortunately this session does not fit within the narrative of our Space for 2015- Hope you will submit another session for 2016Closed mmmavis closed 9 years ago
Saallen
commented
9 years ago We appreciated this submission but unfortunately this session does not fit within the narrative of our Space for 2015- Hope you will submit another session for 2016
[ Google Spreadsheet Row Number ] 314 [ Facilitator ] Anil Thanki
Description
Leveraging the power of the “collective” web to drive the building of a simple ontology for describing common concepts in science. Scientific datasets are often very badly described, if at all, and even small amounts of metadata in a consistent format, i.e. an ontology, can greatly improve the power of those datasets. Specifically, our objective is to develop an open platform whereby individuals can quickly contribute to the semantic description of objects (captured as images) in a domain of scientific knowledge, be it animals, plants, geo-locations, physical characteristics, or anything else! Describing the images with common and consistent tags, we will process the tags into an ontology, and attendees will then be able to answer and visualise simple queries of interests about the target objects and their relationship to other related terms in the knowledge space, showing the power of web-enabled linked data to produce and discover connected datasets.
Agenda
Our idea relies on collaborative working of a group of people, in an open and fluid workshop format, to facilitate knowledge-building. Our session will be supported both by soft and hard infrastructure: a live service (powered by Python and Django) to enable participants to upload images and associated metadata or tags (e.g. geo-location mapping with OpenStreetMap and Leaflet). This information will be held in a document-based database (e.g. MongoDB). An inference engine will transform metadata tags into an ontology. A visualisation service will provide real-time views of the ontology being built (using D3.js). A web-based service will allow users to query the system. The session mentors have prior experience in these technologies, and can teach attendees about them during the workshop. Participants can contribute both to the content (images and metadata) and/or to the development of the system, or chat about the system and technologies used.
Participants
The session will be structured in a way that engages attendees of varying numbers and technological knowledge levels. With fewer participants (e.g <15), participants can both have the opportunity to contribute to the content, as well as help in extending the enabling technologies of the system. With more participants (e.g. >=15), more images and metadata could be collected in order to improve the semantic enrichment of the ontology-building process and the precision of answers provided to queries. Even more attendees would be great for pair or group programming around the technologies used, producing documentation, and testing the platform as it is being developed.
Outcome
We hope that the proposed idea will enable the emergence of services that exploit the power of a community of individuals (via the web) to develop ontologies (or linked information), which would be useful both in research and commercial contexts. We will teach attendees about each of the technologies used in varying degrees of detail depending on their experience and interests. We will make entry level documentation available for those that wish to join the session and get working quickly, as well as for those that wish to continue working on the platform after MozFest. We will keep track of achievements, comments and feedback in an Etherpad, and all source code will be available in real time via Github.