Regional Special Section of Communications of the ACM Recognizes the Work of IMFD Researchers
July, 2024.- Three IMFD works are featured in the recent edition of Communications of the ACM: Latin American Regional Special Section. The special section, coordinated by Fabio Kon (University of Sao Paulo, Brazil), Sebastian Uchitel (University of Buenos Aires), and IMFD researcher and DCC UCH academic, Barbara Poblete, seeks to present the region’s most interesting and impactful advances in computer science.
“Latin America is a very heterogeneous continent, with great cultural, geographic, demographic, ethnic, linguistic, scientific, and technological diversity. When it comes to computer science, Latin American researchers have made significant contributions in multiple areas, such as software engineering, databases, networks and distributed systems, artificial intelligence, theory of computation, and computer science education. In this Regional Special Section, we present only a small part of the work that Latin American researchers are currently carrying out,” the committee highlights in the special section’s presentation video.
Better understanding the constitutional process
In Telar and TelarKG: Data-Driven Insights into Chile’s Constitutional Process, the multidisciplinary team from the Millennium Institute for Foundational Research on Data (IMFD) presents the work carried out to analyze and better understand Chile’s first constitutional process, which aimed to weave together diverse data from disparate sources related to the constitutional process, in order to obtain insights and communicate the results to Chilean public opinion. “The analyses carried out were diverse, and included observation of the use of bots on social media, the most discussed topics on social media during the process, the categorization and cohesion of emerging ideologies among constitutional convention members based on their voting patterns, etc. To disseminate the results, we had a weekly program on CNN Chile dedicated to the Telar Project to present and discuss our findings, and share them directly with Chilean society,” says Aidan Hogan, DCC U. Chile academic and IMFD researcher.
“Telar and TelarKG: Data-Driven Insights into Chile’s Constitutional Process” was worked on by Renzo Angles (U. de Talca), Naim Bro (U. Adolfo Ibáñez), Ivania Donoso-Guzmán (Universidad Católica de Chile), Juan Pablo Luna (Universidad Católica de Chile), Aidan Hogan (Universidad de Chile), Juan Reutter (Universidad Católica de Chile), Henry Rosales-Méndez (Universidad de Chile), and Sergio Toro (Universidad Mayor).
Differential privacy
Another published article is “Gradual Differentially Private Programming” by IMFD researchers and Universidad de Chile academics Matías Toro, Federico Olmedo, and Éric Tanter, who explains that this research tackles the problem of how to develop software systems that use user data without violating their privacy, through a technique known as differential privacy. “Specifically, what we propose there is to make the type systems designed to help programmers respect differential privacy more flexible, through the use of a gradual approach, that is, one that mixes static verification with dynamic verification. This approach will facilitate the progressive and selective adoption of privacy annotations in programs that exploit sensitive individual data, thus providing early feedback to developers so they avoid accidentally violating data privacy,” he comments.
Efficient implementation of graph databases
Finally, Communications of the ACM includes in its edition on Latin American research the article “Tackling Challenges in Implementing Large-Scale Graph Databases”, by researchers Diego Arroyuelo (Universidad Católica de Chile), Aidan Hogan (Universidad de Chile), Gonzalo Navarro (Universidad de Chile), Juan Reutter (Universidad Católica) and Domagoj Vrgoč (Universidad Católica). This work reports on the research carried out at IMFD regarding the efficient implementation of graph databases. “These databases make it easy to represent objects and relationships between them, and are emerging strongly as an alternative to traditional relational databases in cases where information is more heterogeneous and unstructured. All major internet companies are developing their own graph database engines. An example of a public graph database is Wikidata, which is used to deliver information on Wikipedia pages,” explains Gonzalo Navarro.
The article reports on the main results obtained and what they are currently working on. Regarding this, Navarro states: “One challenge with these databases is that solving queries over them is computationally expensive. For example, a more efficient alternative to BlazeGraph is being sought to replace it, and IMFD is proposing one of our prototypes, MillenniumDB, to be the backend that maintains Wikidata.” The researcher also mentions that they describe Ring, a development carried out at IMFD, which allows solving queries for a very significant subset of SPARQL using far fewer storage resources. He also notes that this article “describes the challenges they currently face, which have to do with providing access to multimodal information, that is, not only about the relationships represented in the graph, but also about other information that objects may contain, depending on the type of data they represent.”
—