2nd August: Discussing Fluidity

Summary of the Fourth Round Table. 02.08.2022:

The fourth session of the Round Table Meetings was held under the topic of Discussing Fluidity. It was used for discussions and building on the vocabulary from the previous session (19.07.2022). To kick-start the discussion, three questions were asked by Mirco Schönfeld:

1. How can fluid ontologies make my life easier?
2. Which kind of content is presented by fluid ontologies?
3. How will I come into contact with fluid ontologies?

Fluid ontologies can lead to an overall enrichment of metadata by adding more information to an existing database. Cyrus Samimi criticized this point however in the way that one might not necessarily call it fluid ontology. Some information/data is simply static (like a date of birth) and thus not fluid. For other data, like the information about one’s living situation, fluidity might apply, however. Although this would then be a very simple understanding of the term.

Anke Schürer-Ries added to this by bringing in network theory. Fluid ontologies might thus enhance the possibility of finding networks, meaning the historical networks and economies that have been created through data already entered. And to have networks be expanded by fluid ontologies. With the term expansion she means that fluid ontologies might create possibilities to link e.g., visual data (like a photograph that has been used in different research before) to metadata. This could then lead to an enrichment in one’s own research in that more networks of an object or data could be found. We would be looking for connections through information we do not have, so that people can link concepts with each other and are able to differentiate the semantics of the same word in different concepts and constellations.

This discussion then led to talking about different perceptions of data and how local & academic perceptions differentiate. This point was raised by Cyrus Samimi, who’s expectations regarding fluid ontologies were similar to the ones above. The local perception and description of phenomena like rain for example, thus differs to that of a researcher. This then brings in a new more fluid and non-western perspective of the phenomena. Perceptions can also vary greatly between different generations. One could then see how semantics shift over time or see how multiple perceptions co-exist at the same time. The enrichment of data with this local context through fluid ontologies also helps in decolonising data further since local perceptions tend to be hard to find in data.

A worry Sulayman Sowe expressed was about the amount of context information needed before fluidity actually “begins”. Meaning thus has to be given to something, like a photograph, before it can even have fluidity. The worry here is that fluidity is limited by the original description an object has. For it to be fluid many people are needed to add context and meaning. The fluidity might become constant if only one description is used. With minimal description, the DRE would then have less fluidity. A solution to this, as Anke Schürer-Ries proposed, would be to add information and input over time, since actual research time is limited.

One the biggest issues about tagging all the data and adding meaning is therefore the time and workforce needed, especially when one wants to engage with the data on a more critical level. Onca data is in the research data management system however, we other people could work on it years after the original projects finished, thus solving the issue of time.

The two additional approaches to this, as proposed by Mirco Schönfeld are: 1. A technical solution, where we come up with algorithms that build on artifacts of the data and try to derive some ontological distinction from our data. And 2. One can ask for funding for a few months to put research data you collected into the system. This extra time can be used to tag, clean, and publish the data.

Apart from not having enough time, a problem Britta Frede is facing with having European & African partners, is trying to get researchers to describe the data. The main issue here is digital literacy. Handbooks in French Arabic and English and a taxology/ontology to enter different ideas to have a way of tagging have been developed to bring different perspectives together, but this proves to be difficult. In the African context people are not sure if this digital way is useful or not. Engaging the researchers in the ACCs from the beginning to better understand the problems the African partners have might help with this. In the end accessibility to data is key to ensure ongoing fluidity in data bases, no access means static data base.

Another solution for data tags could also be technical solutions like AI, Jae-Sook Cheong suggested. AI thus creates the context of the data. This would be an automated solution to make the data descriptions as well as enhancing data. This method would however need validation again because AI is not that developed yet, Anke Schürer-Ries criticised. It would thus remain a time-consuming process. It might also be biased when creating context information, which is especially important to consider in the context of decolonizing African studies. In the end we don’t need to “reinvent the wheel” but adapt it to the African context.

The suggestion at the end was to invite as many people (also with nontechnical backgrounds) as possible to the meetings. This would include PIs and Cluster members and not only members of the DRE team, since people from other disciplines and positions can broaden and enrich these discussions.