Home > Project > Project Reflection 2

Project Reflection 2

I. Terms added to Web Protégé

 

1. Linocut
2. Indonesia-Present_Day
3. Lombok-Present_Day
4. African
5. Squirrel
6. John_Muafangejo

 

II. Terms linked from Getty’s AAT in Protégé Desktop

 

1. opus vermiculatum (mosaic process) à added as child of Mosaic (TT of branch = Technique)
2. Classical à added as child of Greek (TT of branch = Culture/Civilization)
3. Mohenjo-daro à added as child of Indus_Valley (TT of branch = Culture/Civilization)
4. Postmodern à added as child of Movement
5. commemoratives à added as child of Function

 

NB: In the Culture/Civilizations branch, I moved Harappan to under Indus_Valley (see its skos:editorialNote for details)

 

III. Triples Added

 

After creating the Object Property of genreOfAgent, taken from a Proton extension (http://www.ontotext.com/proton/protonext.html#genreOfAgent) and defining its Domain as Artist and Range as Movement, I created the following triples between artists and their genres:

 

Subject	Predicate	Object
Degas	genreOfAgent	Impressionism
Picasso	genreOfAgent	Neo-classicism
Picasso	genreOfAgent	Cubism

 

I also added the following triples to place Artists in their Time_Period after adding the GND ontology’s periodOfActivity to the Object properties:

 

Subject	Predicate	Object
da_Vinci	periodOfActivity	Renaissance
John_Muagangejo	periodOfActivity	Modern_18th-21st_C

 

To try some mapping to the Getty ULAN, I also added some triples using skos:exactMatch. I placed these in the class annotation pane after adding skos:exactMatch to the Annotation properties:

 

Subject	Predicate	Object
da_Vinci	skos:exactMatch	http://vocab.getty.edu/ulan/500010879
Degas	skos:exactMatch	http://vocab.getty.edu/ulan/500115194
Picasso	skos:exactMatch	http://vocab.getty.edu/ulan/500009666

 

IV. Component 2 Reflection

 

            The three exercises of adding to the Taxonomy of Jewelry and Beads vocabulary, creating linked classes with Getty’s AAT, and enhancing the Taxonomy with triples proceeded with some challenges, critical points of assessment, and moments of learning. First let me offer a descriptive and reflective workflow outline that illustrates the decision-making process I used to add terms to the vocabulary. I have split out the process for terms sourced from BAC and AAT, since slightly different thinking guided my work in each.

Step 1: Identify Term

• BAC: Identify topics represented in the BAC records not in the controlled vocabulary (CV). The aim here was to tailor the CV to the content of BAC, despite the CV’s focus on Jewelry and Beads.
• AAT: Identify obvious gaps in CV that are represented in AAT. The aim here was to increase the breadth of topics that fit naturally within the CV, with a view toward future tagging.

Step 2: Add term to CV

• BAC: Adding a term sourced from concepts in the BAC records proved challenging. The main reason for this is that the CV seems to have been created for http://amandapirone.omeka.net/ and BAC’s collection is more about contemporary theater, flyers, and visual art. Nevertheless, I was able to place some terms appropriately (e.g., linocut in Techniques). This highlights the risks of extending a vocabulary beyond its scope without a holistic plan for all intended content domains.
• AAT: The decision of which term to add from AAT was driven less by lack of coverage in BAC and more by what topics were blaringly absent, given the CV’s scope. My main aim here was more about building the breadth of the CV and less about coverage of BAC. Again, the main challenge here was identifying AAT links to cover BAC topics, the scope of which the CV for jewelry did not quite accommodate (e.g., I added Mohenjo-daro under Indus_Valley without knowing whether the collection included this).

Step 3: Critically consider placement

• BAC: I considered whether the CV had a natural home where the branch and level of depth accommodate the new term. (e.g., Natural subjects easily accommodates Squirrels). I also had additional qualifications, discussed in step 4.
• AAT: Consider whether the hierarchical placement of the term within AAT aligns with corresponding placement in the CV.  For example, opus vermiculatum (mosaic processes) sits in a process branch directly under mosaic in AAT. This context is aligned with the Techniques branch as a child to Mosaic.

Step 4: Determine placement qualifications

• BAC and AAT: The qualification for placing a term in the CV was based on review of the branches’ top terms to determine contextual meaning of the term being placed. A good example was the placement of African in the Culture/Civilizations branch because I wanted to underscore the cultural aspect of the term’s scope. It is African Culture described in the BAC resource http://www.bacarchive.org.uk/items/show/4208, not African anything else. Thus, the branch’s top term drove the context of descendent terms. This is why Africa-Historical under Regions is completely different than African under Civilizations and why I settled on the placement of the latter. Note that in my initial tagging in component 1 I did settle on Africa-Present_Day. This part of the exercise gave me an opportunity to properly refine this by adding a more suitable term.

    An ontology can be an extremely useful tool for giving users access to unique relations that exist between topics and items as they appear within a collection. It can help answer questions about content, visualize relationships to explore, and offer links to related resources available elsewhere. For example, the Object properties I added to link Artists with Movements or Time_Periods could be used to find related material based on the artist, the movement (genre), or the artistic period of time, or any combination of these items. One can discover knowledge without even really knowing what specifically to look for. This is the value of ontologies over other, less complex controlled vocabularies.  Although crowdsourcing is a useful tool for enhancing the degree to which an ontology is relevant to users, given its potential complexity, control should be centralized. Otherwise, people can make different decisions, for example, about what needs to be a class vs. an instance, and that will lead to inconsistencies. That said, incorporating machine learning to source terms would lead to better integration from more dispersed sources and meet the criteria of the predefined logic outlined by the ontology’s complexity.