IMPLEMENTATION MODEL

The Linking Lost Jazz Shrines workflows fall into five general categories: oral history transcript preparation, identification and/or creation of items in Wikibase, preparation of transcripts in Sélavy, Omeka S metadata updates, and the creation of linked data triples. These workflows should generally be followed in this order, with the caveat that once Wikibase URIs are identified and/or created, transcripts can be prepared in Sélavy at the same time that Omeka S item metadata is updated.

Note: All workflows are predicated upon the existence of an agreed-upon linked data model or "ontology". Information about the data model used for the Linking Lost Jazz Shrines Project can be found here.

Oral History Transcript Preparation

• 1. Conduct and Transcribe Oral Histories

  2. Format Transcripts (Sections I.A–I.D of Transcript Processing Instructions)

  3. Identify Relevant Entities (Section I.E of Transcript Processing Instructions)

  4. Add TXT Version of Transcript to GitHub (Section 1.F of Transcript Processing Instructions)

     • GitHub folder: https://github.com/weeksvilleheritagecenter/linking-lost-jazz-shrines/tree/master/data/formatted_transcripts

Identification and/or Creation of Wikibase Items

1. Identify Existing Entity Items in Wikibase

   • Entities can be manually searched in the Wikibase

   • For bulk data work, all of the person, music group, and music venue items from the Wikibase can be downloaded for comparison to the transcript entities.

   • Future potential enhancement: Write python script compare transcript entities to existing Wikibase entities

2. Create New Entity Items in Wikibase

   • Items can be added individually or in batch to Wikibase, with the assistance of the Semantic Lab at Pratt

   • New Wikibase items must include "instance of" (person, music group, or music venue) and "part of project" (Linking Lost Jazz Shrines) statements

   • Optional statements include Wikidata QID and/or Library of Congress authority ID

   • Future potential enhancement: Write python script to create new Wikibase items (current script in code section is for adding statements to already existing items)

Omeka S Updates

1. Add Missing Linking Lost Jazz Shrines Entities as Subjects to Omeka S Item (Omeka S Items Documentation)

2. Add Wikibase URIs for Linking Lost Jazz Shrines Entities in Subject List (Omeka S Items Documentation)

3. Add Library of Congress URIs for non-Linking Lost Jazz Shrines Subjects (Omeka S Items Documentation)

4. Add Wikibase or Library of Congress URIs to Other Applicable Metadata Fields (Omeka S Items Documentation)

5. Add JSON of Omeka S Item Metadata to GitHub

   • Process: From Omeka S Item page, (1) right click and select "View page source"; (2) copy all content between "<script type="application/ld+json">" and "/script>"; (3) paste into "pretty print" formatter (optional); (4) download JSON file; and (5) upload to GitHub

   • GitHub folder: https://github.com/weeksvilleheritagecenter/linking-lost-jazz-shrines/tree/master/data/omeka_s_metadata

   • Future potential enhancement: Write python script to extract linked data metadata through web scraping

6. Add CSV of Omeka S Item's Subjects to GitHub

   • Process: manually "copy/paste" from Omeka S site

   • GitHub folder: https://github.com/weeksvilleheritagecenter/linking-lost-jazz-shrines/tree/master/data/omeka_s_subject_lists

   • Future potential enhancement: Write python script to parse Omeka S JSON Metadata (from either web or from JSON document) and output a CSV subjects

Prepare Transcript in Sélavy

1. Upload Transcript (Section II.A of Transcript Processing Instructions)

2. Clean and Block Transcript (Sections II.B and II.C of Transcript Processing Instructions)

3. Sent Transcript Text Through NER Service (Sections II.D and II.E of Transcript Processing Instructions)

4. Create rdf:types (documentation pending completion)

5. Create Identities (documentation pending completion)

6. Using Omeka S Subject Lists as Reference, Assign URIs and rdf:types to Identities (Sections II.F.4 and II.F.5 of Transcript Processing Instructions)

Creation of Linked Data

1. Generate Linked Data Triples

   • Temporary Preparatory Process: Google Spreadsheets (documentation pending completion)

   • Intended Long-Term Process: Sélavy (Future Work)

2. Export Linked Data Triples from Sélavy (Future Work)

3. Load Linked Data Triples into Triple Store (Future Work)

4. Integrate Linked Data into Visualization (Future Work)

Omeka S (https://omeka.org/s/)

The Omeka S platform is used by the Linking Lost Jazz Shrines project for access and discovery of the project data, especially for the Weeksville Heritage Center community, and for those who may not have familiarity querying a Wikibase instance or accessing data via GitHub. Whereas Omeka Classic only provided text strings field types for metadata values, Omeka S allows for the inclusion of Uniform Resource Identifiers (URIs), the building blocks of linked data. Support for the WHC's Omeka S instance will continue beyond the Linking Lost Jazz Shrines project, as WHC uses Omeka S for numerous other initiatives and projects.

For the Linking Lost Jazz Shrines Omeka S site, WHC plans to install two modules: First the "Sharing" module (https://omeka.org/s/modules/Sharing), to allow for the sharing of Omeka S content via social media and the embedding of Omeka S content in other sites, and secondly the "Mapping" module (https://omeka.org/s/modules/Sharing; https://omeka.org/s/docs/user-manual/modules/mapping/), to enable the incorporation of geospatial data into the Omeka S site.

Wikibase (https://wikiba.se/)

Wikibase is a knowledge base software for linked data, and is most commonly known as the technological infrastructure that supports Wikidata. The Linking Lost Jazz Shrines project data is held in the Semantic Lab at Pratt's instance of Wikibase (base.semlab.io), which holds and will continue to hold all of the data for the various Semantic Lab at Pratt projects.

Wikibase Resources:

• Wikibase Basic Install Tutorial: a video tutorial and written documentation for installing your own instance of Wikibase

• Wikibase for Research Infrastructure: discussion of the Semantic Lab at Pratt's approach to using the Wikibase platform for the research infrastructure, starting with the Linked Jazz project data

DADAlytics: Sélavy (http://159.89.242.202:3000/)

The Semantic Lab at Pratt is developing DADAlytics, a tool comprised of a Named Entity Recognition (NER) Service and Sélavy, an application to create linked data triples from a textual document. The intention of the Linking Lost Jazz Shrines project was to use Sélavy to create linked data triples, and export data from Sélavy to be loaded into the Semantic Lab's Wikibase.

The project's testing the alpha version of Sélavy revealed that, while the tool is effective in generating batch triples where a subject is related to all the instances of an rdf type within a specific block of text, the tool does not yet have the capacity to create triples where a subject is related to specific object instances, a core need of the Linking Lost Jazz Shrines project. To this end, Linking Lost Jazz Shrines has utilized Sélavy to format transcripts, identify entities, and assign associated Wikibase URIs to those entities, but the project has not yet been able to use Sélavy to generate linked data triples. The continued development of Sélavy by the Semantic Lab at Pratt will enable the eventual creation of linked data triples for the Linking Lost Jazz Shrines project.

Sélavy Resources:

• DADAlytics: Sélavy How To: a video demonstration of the early alpha version of Sélavy tool

• Linking Lost Jazz Shrines Transcript Processing Instructions: includes screenshots and descriptions of Sélavy usage, as it pertains to the Linking Lost Jazz Shrines project

GitHub (https://github.com/)

GitHub has been utilized to store Linking Lost Jazz Shrines project data. Data in the project repository will continue to evolve as the project continues beyond its initial granting period, inclusive of the eventual linked data triples that will be created with the Sélavy tool, and any code that may be written to automate future data creation. As of September 2020, the Linking Lost Jazz Shrines GitHub repository contains the following datasets for each of the ten pilot transcripts:

• Formatted transcripts (txt)

• Sélavy entities, classes, and URIs (csv)

• Omeka S subjects (csv)

• Full Omeka S metadata (json)

Google Suite

While not novel or unique technology, the Linking Lost Jazz Shrines project would be remiss not to mention the utilization of Google suite tools for documenting the project, organizing the data, and preparing for the creation of linked data triples. Google Sites was used to create this documentation site, Google Drawings were used to visualize and communicate the enriched Linked Jazz ontology, Google Docs was used to keep running notes of project and to create the Transcript Processing Instructions document, and—perhaps most importantly—Google Sheets was used to organize project data, track updates for the various transcript processes, and preliminarily create linked data triples in preparation for the completion of Sélavy.

The Linking Lost Jazz Shrines Project itself has not yet generated any code related to its implementation model. The project did, however, use the below sets of code, written by Semantic Lab at Pratt co-director Matt Miller. Once Sélavy is upgraded so that specific linked data triples can be created within the application, code for (1) exporting linked data triples from Sélavy and (2) loading those triples into Wikibase will be written and tested by the Semantic Lab at Pratt.

Export Entities from Sélavy

• Description: given a Sélavy document ID, export entity labels, rdf:types, and URIs from Sélavy

  • README: https://github.com/SemanticLab/wikibase-load-scripts/blob/master/selavy/README.md

  • Code: https://github.com/SemanticLab/wikibase-load-scripts/blob/master/selavy/export_entities.py

Load Data into Wikibase

• • • Description: given a csv of Wikibase items (QIDs), properties (PIDs), and values, load data into Wikibase

    • README: pending

    • Code: https://github.com/SemanticLab/wikibase-load-scripts/blob/master/load_data.py