...
- https://github.com/LD4P/discovery/projects/2 for issues etc.
- Research: how to go from knowledge graph to an index
- Research decision points, Use cases
- Jamboard link
- Expect to include Works. Need to do something beyond what we already have live from the OCLC concordance data.
- References/bibliography list (beginning)
- BANG! (Bibliographic Aspects Newly GUI'd)
- 10/27: Lessons learned and documentation still outstanding
- BAM! WOW! (Browsing Across Music With Obtainable Wikidata)
- 2022-10-13
- Kevin (Stanford) returned spreadsheet of properties present in Wikidata that are not commonly found in library metadata for musical works – to help guide which data would be an enhancement; need to monitor for additional content. Steven wrote query connecting musical work to what it was created for (e.g.: song created for a soundtrack) AND finding if LCCN exists... to then query whether we have that in the catalog.
- Huda started a GH branch - plan is to pick use cases and set something up that represents that (e.g.: musical work in catalog has info button to Work that'll include K-number and other properties). pulled latest from Blacklight-Cornell and having issues getting Dev to run – D&A has been in multi-week updating to Ruby3; what is in Dev right now will not run in without significant updates to machine and code. Greg'll need to make tasks to get work done; not insignificant effort.
- Hoping to go live with the new D&A code today or early next week; can start work for LD4P to run this code right away – will take a new work days but has good notes since has run this twice for D&A. Not urgent need.
- 2022-10-20
- Worked on setting up info box on item page. Info box retrieves information in this way: get values from "authortitle_facet" field, parses to remove "|" to generate a form that can be used to query the LOC suggest service for name title authorities, uses the localname to query against Wikidata for properties of interest. Currently displaying heading, Wikidata entity, date of first performance, location of first performance, and music created for. Next steps are to add remaining properties of interest from Kevin Kishomoto's group's spreadsheet.
- Options for display: Repurposing author title browse page similar to how we updated the author and subject browse pages to include more information, or incorporating information into the item page similar to how we incorporated Discogs information
- Greg getting some errors from Solr backup scrip that need investigating
- Worked on setting up info box on item page. Info box retrieves information in this way: get values from "authortitle_facet" field, parses to remove "|" to generate a form that can be used to query the LOC suggest service for name title authorities, uses the localname to query against Wikidata for properties of interest. Currently displaying heading, Wikidata entity, date of first performance, location of first performance, and music created for. Next steps are to add remaining properties of interest from Kevin Kishomoto's group's spreadsheet.
- 2022-10-27
- Using properties provided by MLA librarians, experimenting with bringing in information to the Info pop-up
- One SPARQL query for entity; updated code to look for multiple values for any property and then removes duplicates. Allows for grouping in the results/UI
- Some properties require special handling - catalog codes (2 pieces of information that go together: catalog + code. Another is "music for", which has two pieces of data)
- Example without results: Puccini's Turandot: our heading is different from the authorized heading. BIBID: 4983368. The author-title is constructed in this case for the raw data. Vendor record for NetLibrary, Inc. - is there a pattern we can use to better build the SOLR index (e.g.: using the first 600 as seen in 4983368)?
- Question for Tracey - how do we increase coverage for the feature Huda is building? E.g. netlibrary 600s with correct Name/Title (not reflected in the author_title_facet)
- Question for later: hide bad data through queries OR expose that and deal with it later?
- Question: where should button to go to connect to existing author-title browse from the item-view?
- Question: When to link back to the catalog, and when might want to link back to Wikidata for context about related thing (e.g.: work is dedicated to (person). (person) may have an LC heading, which can be used to pull up related resources... but if they do not, do we then link back to Wikidata?
- Mid-November: get interviews with music students with what we have at that time and assess which pieces of information interest them; gather feedback.
- 2022-10-13
- DAG Calls
- 2022-10-27: Nat'l Lib of Sweden presented on 10/25 - two points of note: card/chip model where ask what are the properties of interest for a particular entity and what is the graph for that entity? Similar to what Phil has in a linked data summit question: record. These are the data points needed to make sense of the entity. Card = data displayed; Chip = search snippet. Implementation specification. Internally used custom ontology as a mapping/hub so they have equivalences and subclassed off other ontologies (schema, BF, DC, others) to allow for system needs alongside other needs. JSON-LD blob that represents graph and ElasticSearch index runs what users are seeing (catalogers AND discovery). Triplestore exists but sits separately to populate postgres. Lots of questions. Great session.
...