Update summaries-mahe.md

_objects/summaries-mahe.md

@@ -39,13 +39,13 @@ Curiosity then led me to Dominique Santana’s work “A Colônia Luxemburguesa
 
 The course on networks, particularly the “Palladio” application developed by Stanford University, allowed me to understand the usefulness of these tools for historical research. I greatly appreciated the articles read in preparation for the session, which clarified the ex cathedra lecture; without them, I would have been completely lost. Once I "understood" the concepts of networks, nodes, relationships, and attributes, the presentation was clear. The marriage example reminded me of the complex organization of my wedding, and having experienced it, it facilitated my understanding. During the presentation, I often thought about the usefulness of this method for resistance networks during World War II, and I was pleasantly surprised to see that I was not mistaken, as it was used as an example during the presentation. Regarding the practical exercises carried out during the session, I was very cautious, aware of my digital shortcomings, by reproducing a wedding party. It is not original, of course, but in my case, it is sometimes better to play it safe. I am still not very comfortable with the interface, but I could see that it worked nonetheless. A diagram allows for a global overview, enriching a historical narrative without cluttering it. The difficulty lies in the choices to be made, but that is the nature of any narrative. The course being an introduction to digital history, it is logical that we cover the topics with an overview. However, it is unfortunate that we cannot, due to lack of time, deepen our knowledge of these tools to develop our skills in the field to sharpen our tools for writing the master’s thesis. That said, it is a matter of choice, as we did not choose the master’s in digital history, sometimes with regret, but we must remain realistic; I would have been completely lost.
 
-[7th Session], [30.10.2024]
+## [7th Session], [30.10.2024]
 
 During the presentation on October 30th, Ludovic Délépine and Marco Amabilino explored the integration of artificial intelligence with the European Parliament Archives. They explained how to access various documents via the website, for example by year or by topic. They mentioned Karen Spärck Jones and the idea of introducing a tool to count the number of times a word appears in a document, which helps assess the importance of a particular word or document. Other AI pioneers were mentioned, but as the conference was aimed at a more exact sciences-oriented audience than humanities, the historical approach to computing was unfortunately only briefly outlined.
 They also explained the archiving rules and why only documents from 1952 to 1995 are available. AI is presented as a valuable research tool, but users must remain critical of the information generated by AI. The presentation was informative, although some technical terms may have been confusing for those less familiar with computer language.
 In conclusion, AI is seen as a tool for understanding history, but it is essential to continue analyzing and questioning the information it produces. The overview provided by technologies in general offers a new vision of history.
 
-Summary of Dashboard Usage, <https://archidash.europarl.europa.eu/ep-archives- anonymous-dashboard>
+## Summary of Dashboard Usage, <https://archidash.europarl.europa.eu/ep-archives- anonymous-dashboard>
 
 The European Parliament Archives Dashboard is an intuitive and visually appealing tool, designed to provide a relatively clear and detailed overview of the documents available online. It comprises five main charts, each offering a unique perspective on the archives.
 The first chart displays the total number of documents, giving an idea of the archive’s scope. Next, a pie chart categorizes the documents by language, illustrating the contributions of various EU member states. The predominance of French, German, and English is explained by the history of European integration, while documents from the most recent enlargement countries are not yet included as their archives are not accessible. This chart is particularly useful for understanding the linguistic diversity of the archives.
@@ -54,23 +54,33 @@ The final chart shows the distribution of documents by start date, with a notabl
 The records are categorized into seven types: fonds, series, file, title, long title, PE number, and reference code. The current naming conventions, such as EU.HAEU/AC, can be confusing for non-specialized users, but once understood, they allow for a much more comfortable search compared to the paper era. An explanatory legend or table of contents would be beneficial to clarify the content of each fonds and make the dashboard more accessible to the general public.
 The dashboard also allows users to filter documents by type, language, and year, and offers an “Ask the EP Archives?” feature that enables users to ask questions and receive immediate textual responses. This feature aims to ensure that responses are based on documents from the European Parliament Archives, thus limiting the risk of AI “hallucinations.”
 
-Questions
+## Questions
 
 1. Does working with private, non-European development programs pose an ethical and internal security issue?
 2. How does AI handle multilingual documents and ensure translation accuracy? Do you still collaborate with the translation service?
 3. Does the thirty-year legal delay pose a democratic problem? As an emanation of democracy, shouldn’t part of the archives be available to citizens within a more “reasonable” timeframe?
 4. How is AI formatted to avoid biases and ensure the objectivity of search results?
 5. How are documents technically digitized? I imagine this varies depending on the era, but are scanners still used, for example?
 
-[8th Session], [06.11.2024]
+## [8th Session], [06.11.2024]
 
 The session on the reconstruction of sources and the application of historical criticism to digital history was an exemplary introduction. Although I am not aiming for a career in teaching, the way the debate was introduced following a burning current issue kept the entire class engaged.
 The core of the course itself was particularly accessible, especially compared to the article “Data Scopes for Digital History Research” by Rik Hoekstra and Marijn Koolen, which I found complicated.
 Historical criticism faces new challenges. I was struck by the difficulty of reconstructing a very simple model on Palladio. With a more complex system, it must be very complicated, hence the importance, regardless of the field or historical research method used, of employing a rational and strict methodology. The importance of appendices and referencing sources and archival collections, along with the “FAIR” method, acts as a mimicry regardless of how historical narratives are created. Since digital history is relatively recent, it is not surprising to see these ethical issues arise, which inevitably lag behind.
 On a more personal note, the “Findable” and “Accessible” aspects made me question the “feasibility” of my master’s thesis. Protecting my sources is crucial, given the subject matter, and there will inevitably be witnesses I cannot reveal. Legal and methodological questions will quickly come to mind regarding this work.
 
-[9th Session], [13.11.2024]
+## [9th Session], [13.11.2024]
 
 The session on the use of GPT and AI was very instructive. It is sometimes difficult to grasp computer science, but the initial explanation was clear. Regarding the exercise, we started by submitting the Strasbourg Declaration of April 22, 2024, available on the website of the Grand Orient of Luxembourg. We then asked the same questions, first open-ended, then more specific, to Chat GPT for Kenan and to Copilot for Pierre. We noticed that when the questions were specific, both AIs did not answer directly but tended to beat around the bush and repeat themselves. The chosen topic, Freemasonry, was deliberately discreet and evocative of certain public fantasies to challenge the AI.
 Overall, the experience did not provide us with new knowledge, even though no answers were strictly incorrect. We observed that both AIs tended to generate text that was not always relevant, similar to a student in an oral exam who does not answer the question but tries to steer the professor towards their own terrain.
 The use of these tools is nevertheless valuable for approaching an unknown subject, similar to how one might use Wikipedia, although extreme caution is advised. For a simple search, these GPTs can be recommended, but from an academic standpoint, they remain weak. In conclusion, this exercise allowed us to follow the use of two AIs generating factual but imprecise answers. That said, using these tools can be practical for getting an overview of a subject, offering a considerable time-saving compared to traditional methods.
+
+## [10th Session], [20.11.2024]
+
+missing
+
+## ## [11th Session], [27.10.2024]
+
+
+
+