- Julia Neidhardt
- Gerlinde Schneider
- Mete Sertkan
- Martina Trognitz, Lukas Gehrig: xmlTEIontheMap: first place
- Marta Palandri, Raphael Mitsch: The Quick War?: second place
We received two submissions, each using a different dataset and having a different objective. The submission The Quick War? by Marta Palandri and Raphael Mitsch aims to investigate how the sentiments of German newspaper articles that report on World War I and that are provided by the European Newspapers Project change in the course of the war. The submission xmlTEIontheMap by Martina Trognitz and Lukas Gehrig aims to visualize locations that occur in the “Karl Baedeker: Das Mittelmeer. Handbuch für Reisende: Digitale Ausgabe” on a map. Since the objectives of both submissions as well as the data the build on are completely different, the submissions are incomparable to each other. In general, we liked both of them. However, taking into account the creativity, reproducibility, reusability and elegance of each of the submissions (see below), we apply for granting xmlTEIontheMap the first place and The Quick War? the second place.
The Quick War? is an exciting project, which aims to apply computer science methods to historical data in order to gain new insights. Thus, it clearly can be located in the area of digital humanities. The idea of using dynamic topic modelling is rather innovative. However, in the end this was not realized; only regular topical models were used. Furthermore, the approach applied for sentiment analysis is rather naïve. This might explain that no difference in sentiments could be detected over time. The submission xmlTEIontheMap is an interesting project as well, and it fits perfectly into the field of digital humanities. However, the idea is rather straightforward. The web prototype could be improved by changing the visualization dynamically with the text rather than having a static visualization. However, overall the introduced way of processing and displaying the data could lead to new insights.
Unfortunately, the reproducibility of The Quick War? is not given. Although some issues could be resolved by installing packages manually and changing hard-coded paths in the python code, some required files were missing. For the submission xmlTEIontheMap reproducibility is given for the data extraction and enrichment part. The python code works and and is properly documented - even limitations of the implemented approach and possible solutions are discussed. The visualization, on the other hand, is not reproducible due to “cross origin request” issues when working with Chrome (which is suggested) and due to a DOMException thrown by the library “CETEI.js”.
The submission The Quick War? is not reusable. Due to the lack of a proper documentation it is hard to find the errors and make it runnable. It’s positive that there is a blog post, where the main idea and the proposed approach of the project are described and some results are listed. However, more details should be provided, e.g., in the diagram the labels are missing. For the submission xmlTEIontheMap, some parts are reusable. In particular, the enrichment of the location entities with geolocation information is a valuable contribution, where future work can build upon. The reusability is also improved by the comprehensive documentation. The visualization cannot be reproduced but at least the prototype is provided on a web page.
Although some parts of the code of The Quick War? are well readable, there is all in all a lack of clarity. It’s not easy to modify and reuse. The python code of the submission xmlTEIontheMap is readable, easy to modify and can be reused. However, we would suggest to separate the data from the actual code, e.g., with respect to the city names mapping. The HTML code is also readable and well structured. However, it would make sense to put the JavaScript into a separate file, for even more clarity.