Brussels / 3 & 4 February 2024


Node Reduction through Artificial Intelligence Inferences using Graphology and SigmaJS: A Case Study on Hypertextual Conversations in Freight Train Graffiti in the North American Region.

Embark on a journey into the intricacies of social sciences research, exploring the fusion of different platforms for a single project. Drawing from my hands-on experience in my Communication studies master's thesis, "Visualization of a hypertextual interaction field in the form of a network using computational processes. Case study: Graffiti on freight trains in North America", this presentation shares the challenges faced by social science students in developing robust programming projects without prior tech expertise.

It underscores the pivotal role of open source tools, and community, in overcoming these challenges, shedding light on the logical links facilitated by adaptable programming projects. This allows the integration between platforms and the possibility of creating logical connections between data and theoretical-methodological aspirations.

The synergy of spaCy, TensorFlow, Instagrapi, Graphology, and SigmaJS exemplifies the potential of open-source software. The presentation explores intricate adjustments made to spaCy, such as hashtag split, term indentification and graffiti entity identification (as writers and crews) without Named Entity Recognition (NER). It also share the application of custom object detection model in TensorFlow. The talk emphasizes the decision-making flexibility within Graphology/SigmaJS for network construction and visualization, with a focus on "reduction by inference" as a clear outcome resulting from the amalgamation of theoretical perspectives and different programming libraries or platforms.

In conclusion, the interplay between TensorFlow or spaCy, model inference in Python, the role of SQL as a link between platforms, and network formation with PHP to Graphology syntax, culminating in visualization with SigmaJS in JavaScript, underscores the significance of open-source tools. "Reduction by inference" emerges as a crucial aspect for identifying symbolic elements in social phenomena, exemplified by graffiti on freight trains. This achievement is made possible only through open-source tools and their potential for linkage through "data".


Photo of Angel R. Abundis Angel R. Abundis