Assembling fragments: a two-layer knowledge management tool to explore algorithms and their social functions #IST23 @asist_ec

My second liveblog from the Information Science Trends conference taking place in Uppsala, Sweden and online. It is Assembling fragments: a two-layer knowledge management tool to explore algorithms and their social functions, from Rayya Roumanos and Olivier Le Deuff. The abstract is in this archive https://zenodo.org/communities/information_science_trends/?page=1&size=20 and also here Roumanos started by explaining the aims and scope of the ALGO-J research project - which arises from collaboration between journalist and other organisations with the researchers. The aim is to to provide journalists with the necessary resources to understand and critique algorithms.
Journalists are now dependent on algorithms for various aspects of their craft, including gathering and disseminating news - as well as reporting on stories to do with algorithms. However, journalists are not necessarily algorithm-literate. The fact that the code for algorithms is hidden does not help. However there is also a certain lack of knowledge, skill and curiosity about algorithms amongst journalists. Roumanos talked about the need to opt for taking a socio-technical (rather than just a technical) perspective - there are issues of power, bias etc. in algorithms.
In order to investigate algorithms, theere is a need to make the algorithms not just visible, but hypervisible. LeDeuff talked about the Cosma tool, which enables document graph visualisation and demonstrated https://cosma.graphlab.fr/ (see photo) - leDeuff mentioned influences from the documentalist Paul Otlet and also Ted Nelson The data within the system (the "index cards" are produced by the researcher (with definitions etc. based on the literature)  and also including articles about journalism and algorithms).
Algorithms are used to create the graph, and people will be able to play with it to see the impact of the algorithm and the effect of changes to the algorithm (via sliders that are on the tool's screen). Journalists will have input to how it is developed and how it can be used, which should help them be able to understand and investigate algorithms.
The project website is at https://algoj.hypotheses.org/