This week we have our web interface for earth science owl search set up:
While I’m still debugging on some minor indexing issues which affect the search results. We discussed a lot regarding how to improve the performance, which includes further data pre-processing and new calculating methods for entity entropy. Also, I worked on enabling the “individuals” facet on the right side of our current annotator, which allows the user to click on a class in order to see instances of the class.
For next week, I would work on integrating the search bar into our annotator main page with Katie and also start on owl recommendation design. We consider 3 measures recommending an rdf-owl description:
1. identify by data type (time, date, measurement, integer)
2. identify by user history (favorite owl, favorite class.. )
3. identify by csv column title/content (semantic meaning)
Giving certain weights to all 3 and return a final ranked recommendation result based on the column the user selected. This requires collaboration with the user management system team and taking use of user history data. Expecting further brainstorm with mentors.