Page tree

Versions Compared


  • This line was added.
  • This line was removed.
  • Formatting was changed.



Michael Bobak do we want to create additional proposals to capture the following as well? It is easier to have a single page per proposal.

Presently our users have something like a focused google-dataset-search, (via aggregated metadata), that given a query returns potential hits with: Title (w/link to the page w/the data), Abstract, & a metadata ('details') link.

1: Notes:

The metadata-tab is a great replacement for the 'details'; the only thing that is needed would be to have that tab selected when you get to the page. Especially given that the Clowder dataset is presently a stub to hold the metadata; it has no files. So files is a bad default tab.2:   Since the 'Title' now links to the interim Clowder stub page, I've added the link to the data page at the end of the description. So it will still show up in the search results.  It often but not always ends up as a clickable link.   If it was more consistent it would be a more useable search.

3: It has been suggested that we have a per space config to select some choice metadata to put in the upper right of the dataset page, e.g. the dataset-landing page.

4: Also, to get a google-dataset-search experience, it has to be done w/o a login, it turns out that doesn't presently work right now:

5:   I had also considered just using the Clowder API, to hook into our present web page at one point.  One thought there was having something like a gDoc read-only link, or in this case a key, or there could be a search-user w/read-only access to all the metadata of interest.6:   It would also be nice to have the preview image be able to display the space's logo-image, especially as there are not files to preview, &that is what our users have come to expect.

7: I've been experimenting w/carrot2 clustering of search results, which I hear there is a plugin for.

8: We can presently search by any combination of repository/spaces, via checkbox, which might be possible, but I'm not sure how it is done offhand. There is the hourglass for (just one) space, but then we get also into having to pick a facet, and none of them are from the uploaded jsonld.

9: I really like the metadata-search page, but most of our repos (&other aggregators) also allow for selecting a geo-bounded box from a map.  As almost all our datasets have lat/lon coordinates in their jsonLD metadata.  It will be a feature this geoscience based search will need.

We continue to make use of & DCAT2 which has a mapping.  This along w/CSVW also makes the data discoverable via google dataset search.  Though our more focused faceted search should be of even more use to the EarthCube search users.