Session Summaries by Marques Océane

DH Theory, 06/11/2024

The course began with a debate on the US presidential elections, and more specifically on our opinion of the election of Trump as the 47th President of the United States. We then learned a little more about DH theory and David Irving’s publishing career, as well as his Holocaust trial. In fact, in several of his works, he deliberately quoted false information and for this he was strongly criticized by Deborah Lipstadt. After making us understand that to prove events like the Holocaust, a critical analysis of (archive) sources is very necessary, we moved on to the practical part of the course. We had to document our own network analysis (via Palladio) that we had made out a few sessions earlier. To begin with, M. Düring gave us an introduction to each of the steps: selecting the data, modelling it, normalizing it, linking it and classifying it.

Chiara and I tried to apply these steps to our ‘Barbie Birthday Party’ example. We encountered several complications with this exercise, as we had filled in the boxes in Excel randomly (‘Relationship’ and ‘Sympathy’ categories). In other words, it was almost impossible to give any origins or reasons for these links. The only sources that can be used for this data are the films featuring the various ‘Barbie’ characters. However, even if this didn’t really work for our example, I think that this process avoids the manipulation of data to justify erroneous interpretations or distorted theories, particularly in historical studies.

Networks & PalladioEU Parliament Archives, 30/10/2024

The Archives Unit Dashboard is an invaluable tool for archival research, with a comprehensive set of features to facilitate access to parliamentary documents. Thanks to its accessible and intuitive structure, it enables researchers and the public to easily explore a variety of archives, while offering innovative features such as time-lapse or oral sources. This makes it particularly suitable for academic research or large-scale projects requiring extensive consultation of parliamentary documents. The Dashboard’s interface and ease of use are its strong points. The tool lets you filter documents by language, type and year, making searches more precise and adapted to users’ needs. However, despite this accessibility, I noted certain limitations that reduce its effectiveness. For example, keyword searches (“Top Words”) seem to be limited to French, which may pose problems for users who wish to consult multilingual documents or who prefer to search in another language. This may have something to do with the settings, or it may be that French is the only official language of the documents (I haven’t looked into this yet), but it remains unclear. Also, the associated keywords in the Eurovoc word cloud sometimes lack relevance. When I searched for the term “logement”, some of the associated words seemed inconsistent or kind of inappropriate, which can limit the accuracy of results for specific searches. During the conference, Ludovic Delepine and Marco Amabilino, from the Archives of the European Parliament, discussed several technical aspects of the tool. One of the major functionalities is the extraction of text from scanned images using OCR (optical character recognition) tools, enabling document images to be transformed into searchable text. Although this technology is now well advanced, it still has certain limitations in terms of character legibility, particularly on lower-quality images. At the same time, the tool offers a document similarity search, making it easier to identify relevant or related documents. They have also made it clear that this is a platform open to the general public, but which only processes documents up to 1994. Although I found the first part of the conference enriching and appropriate, I have to admit that the second part was more difficult to follow. The presentation contained many technical terms, and I wasn’t always able to put them into a clear context. The questions asked by the participants were also quite specialized, which reinforced the feeling of a gap between my level of understanding and that of the experts.

Here are a few questions I had noted before the conference:

• To what extent can AI tools respect archival concepts such as provenance and authenticity?
• Are there any risks of loss of meaning or context due to automation?
• In your tests of AI tools for digitizing archives, have you encountered any technical or ethical limitations that seem difficult to overcome?
• Will you soon be making it possible to search for words in English (or other languages)?
• Have you found it easier or harder to access certain information?

Networks & Palladio, 23/10/2024

In this lesson on social networks, the concept is introduced using the example of a wedding party to illustrate several fundamental notions. Terms like “reciprocity” are highlighted, meaning that a relationship in a network must be mutual for it to work both ways. Guests each have an “ego” network, their own group of immediate relations. Networks are also delimited by boundaries. For example, wedding guests are divided according to attributes such as age, interests or relationship status. The way these people interact within their groups, creates specific dynamics. At the old people’s table, conversations often come from a single person, forming a particular network pattern that contrasts with that of singles, where there is a greater density of connections. There’s also the term “broker”, which refers to a person who circulates from one network to another. In other words, network structures are therefore dynamic and can change over time, with information and relationships evolving in line with interactions. There’s also the difference between “unipartite” and “bipartite”. Social networks are also made up of affiliations and interactions, and the concept of social network is founded upon four essential dimensions: the self-perception of historical actors, its function as a metaphor for groups of related actors, its role as a constructed object of scholarly research and its use as a theory/method driven approach to research. To conclude this theoretical part, we had to create a network visualization ourselves, using the “Palladio” tool. A particularly useful tool for exploring relationships between different elements, and therefor ideal for analyses in research projects.

Maps, 16/10/2024

This 5th course taught us more about maps, and particularly the “StoryMaps” tool. After trying our best to represent the Belval campus on a map, we learned more about Geographic Information Systems (GIS), which are analysis and visualization tools that go beyond the simple creation of maps. 2 leading software packages exist: ArcGIS and QGIS. These programs enable to superimpose different layers of information. This course reminded me that maps are an essential tool for researchers and historians because of their considerable informative impact. In addition to star maps or location maps, there are also topographic maps, also known as reference maps and thematic maps. What I also learned was the difference between location and place. Place, unlike location, which acts as a landmark, is a kind of social construct. For example, we call our university campus in Belval “campus”, but people in the financial sector might refer to it as “Fond Belval”. It all depends on people’s contexts and perspectives. My group then focused on a project entitled “History of the Olympic Marathon, 1896-2020”, created using the interactive tool “StoryMaps”. In addition to the runners’ personal details, the different durations of each winner’s 42-kilometer race could be compared. This project, which highlighted the evolution of runners’ performances, and the participation of women, did, in my opinion, raise questions about scale and physical representation. However, as a historian wishing to teach, I find that “StoryMaps” can make learning history livelier and more interactive for students, that way, enriching their educational experience.

Online Newspaper Archives, 09/10/2024

The course got off to an exciting start with the introduction of a new platform, “Impresso”, which highlights the importance of newspapers as witnesses to past knowledge, where valuable ideas are preserved. I was a little skeptical about its use and real utility. However, thanks to the group work, I was able to see that this tool could be complementary, especially when combined with resources such as eluxemburgensia. My colleague and I compared the results obtained by searching for our villages, “Saeul” and “Bascharage” respectively. Not understanding why the term “Saeul” had (many) more mentions in newspapers than “Bascharage” - even in non-Luxembourg newspapers - we soon realized that the searches were confusing “Saeul” with the German term “Säule”. During the course, I also discovered a new concept: the scientific term “OCR”, which I’d never heard of before. On the other hand, I found it a pity that the part devoted to the “Github” tool was dispatched in just 15 minutes. That was a lot of information to assimilate in a short space of time. Fortunately, we received a PDF file to guide us through the rest, which made it easier to understand the tool. Personally, not being very comfortable with technology, I sometimes find it difficult to master these tools. However, I realize that it’s essential to take the time to explore them in greater depth, as they offer interesting opportunities for research, analysis and comparison.

Web archives, 02/10/2024

This lesson made us aware of the importance of web archives and the evolution and conservation of data. From more personal information-such as life details or works of individuals- to more scientific and journalistic elements, the different groups showed us how these archives are becoming increasingly popular. Our group analyzed in detail the evolution of the “Luxembourg.lu” web page. We examined how the categories and subcategories of the site have changed between 1996 and 2024, highlighting what aspects were given more pertinence over time, as well as the little change in the URL, but the new site also offers the possibility of changing languages. A major tool for this analysis was the Internet Archive’s Wayback Machine. It gave us the possibility to observe how many times the website was captured, using this as a measure of how the internet itself has evolved over the years, but also providing concrete data on its changes. This lesson aroused my curiosity, and this theme also raised some questions about the future of this type of work and tool, for example: how can the accessibility of archived data be improved to enable more in-depth analysis by historians or researchers? To conclude, I have learned that digital preservation tools can offer wise analyses of societal trends and changes in terms of the presentation of information and in the growth of importance of web-based data over time, demonstrated by the example of the portal of the Grand Duchy of Luxembourg.

From Sources to Data, 25/09/2024

In recent years, the word “data” has become more common, often referring to various types of sources. Historically, the term originated from mathematicians, but after the 18th century, it has come to apply to any field of intellectual activity. “Data” refers to small elements like observations or measurements. For example, a dataset in an Excel sheet may include rows of tweets, with each row representing one tweet. The process of exporting the tweet into the sheet involves converting it into data, likely including details like the tweet’s text, time, user, etc. Therefore, the distinction between data and metadata is important: in this case data refers to the raw information, while metadata provides details like when the tweet was posted, by whom, location and other contextual elements. In other words, metadata helps organize and contextualize sources. This additional information helps historians, or simple users, to understand the context and provenance of the source, making it safer and easier to reuse. Finally, I also learned that tools like “NGram” can help analyze data trends but require specific conditions to use effectively. In some way, this lesson was useful for my historical studies because it helps me understand how data and metadata are used to organize, contextualize and analyze historical sources. Even if it was difficult at first, it gave me a better overview of the digital tools that can enrich research, especially int the way information is structures and preserved. It also sparked my curiosity by giving me a different perspective on the evolution of archival practices and how they influence historical research today.