Images recognition, democracy, communications and databases: Six Fondecyt 2023 for IMFD researchers

January, 2023.- Six works by researchers from the Millenium Institute Foundational Reseach on Data received the Fondecyt funds for 2023. These belong to the programs of the National Agency for Research and Development, ANID, to promote scientific-technological-based research in the various areas of knowledge, through the financing of research projects of excellence, oriented to the production of knowledge.

The IMFD researchers who received the funds are Benjamín Bustos, academic from the Department of Computer Science at the Universidad de U. Chile; Juan Pablo Luna, academic from the Institute of Political Science and the School of Government of the P. Universidad Católica; Denis Parra, academic from the Department of Computer Science of the P. Universidad Católica; Gonzalo Navarro, academic Department of Computer Science of the University of U. Chile; Cristian Riveros, academic from the Department of Computer Science of the P. Universidad Católica; and Sebastián Valenzuela, academic from the Faculty of Communications of the P. Universidad Católica.

Image processing

The project of Benjamín Bustos, an academic at DCC U. Chile, works in the area of digital archaeology, performing symbol analysis on ceramic objects, which are scanned in 3D. In this research, the design, development and evaluation of algorithms that allow to automatically label the patterns that appear in ancient ceramics are proposed, based on the work of Kunisch in his book “Ornamente Geometrischer Vasen: Ein Kompendium”, where he defines a taxonomy of labels that may appear on ancient pottery. For Benjamín Bustos, this award will allow progress in the project: “We hope to investigate algorithms that allow the automatic or semi-automatic identification of the patterns that appear drawn on ancient ceramics or also in other contexts such as rock art”, and highlights the important collaborative work developed with researchers from Anthropology and Archeology with whom they carried out a collaborative workshop, which makes “this project now more interesting given the possibilities of being able to test the methods and algorithms that we will develop with new data and in local contexts”.

Also in the area of image processing with algorithms, the interdisciplinary research of Denis Parra, an academic at DCC UC, uses deep learning techniques for the recognition of medical images, which makes it possible to optimize diagnostic times in radiology, creating powerful tools useful for the world of health. For the researcher, this award represents “great satisfaction, but mainly a boost that we expected to continue the research that we have been carrying out for a couple of years on the generation of radiological reports from medical images. This project is interdisciplinary because it involves the development of AI algorithms under the scrutiny of doctors, whose role in this project is taken by the co-director Cecilia Besa of UC Radiology. ”As a goal, they hope to improve the ability to generate reports in Spanish, since much of the current technology for these models is only available in English.

Studying democracies and communications with data

With the research “Analysis of the informative repertoires of users of social platforms and their relationship with current news knowledge”, the IMFD researcher and academic at the Faculty of Communications of the P. Universidad Católica, Sebastián Valenzuela, seeks to analyze the informative repertoires of the users of social platforms in Chile and to evaluate the incidence of these informative repertoires in three dimensions of news knowledge: retention, comprehension and self-confidence. “A fundamental premise of democratic theory is that the media must provide accurate information about current affairs and that citizens must have a sufficient understanding of public affairs to be able to make informed decisions and participate effectively: Today they are the main media Social platforms such as Facebook, Instagram, WhatsApp, YouTube and Twitter are used by the population in Chile to find out about current events. But it is not clear that these media provide relevant and truthful content, as indicated by the phenomenon of misinformation”.

Meanwhile, the research by Juan Pablo Luna, an academic at the School of Government and the PUC Institute of Political Science, investigates the crises of democracies in Western societies. The study focuses on the cases of Bolivia, Chile, Ecuador, Paraguay, and Uruguay, and uses the cases of France, Germany, and the United States as controls. “It is an opportunity to understand in greater depth the type of crisis that contemporary democracies are experiencing today, emphasizing demand factors: what society expects from politics; over supply factors (such as political leadership, institutional factors) that are usually emphasized by recent literature”, explains Luna. The objective is to be able to consolidate an approach that recovers traditional insights from economics and political sociology to understand the crisis of democracy and eventually its potential remedies.

Robust data structures and algorithms for large amounts of data

In “Compact Data Structures for Graph Databases”, Gonzalo Navarro, an academic at DCC U. Chile and who was recently recognized as a Fellow of the ACM, delves into the work on graph databases. “This type of database -among which is, for example, Wikidata- are becoming popular as more flexible mechanisms than the traditional relational model to represent information”, explains the researcher. This flexibility comes with more powerful query languages, and where answering those more complex queries is more demanding in terms of resources. “My recent work shows that compact data structures, a branch that has been consolidating for a few decades and that combines information theory with data structures, can be a good answer to this problem. In the project I propose to continue with this line of work, to go from concept proofs to complete solutions that can completely replace the current systems that manage graph databases. This involves using the functionality provided by compact frameworks to further improve the performance of current solutions, integrate them into real systems, extend solutions to queries and more general query plans, enable dynamism, support graph analytics, and investigate new compact structures that can offer even better performance than the ones I have considered so far.”

The challenge represented by the enormous amount of data that is currently generated, for example in Internet traffic, social networks, weather forecasters or sensors, is the focus of the research by Cristian Riveros, an academic at DCC UC, “Constant-delay algorithms for Complex Event Recognition (CER)”. CER systems emerge as a response for the rapid processing of large data flows, giving timely responses to queries made to the systems. In this project, it is proposed to study the algorithms used in these systems, their feasibility and limits, in order to define when they can be used and when not.

The diversity of projects and topics that these investigations address is a reflection of the universality of the work with data that is carried out at the institute, as well as the interdisciplinarity that is actively promoted in the community of researchers. The Fondecyts are especially focused on projects that lead to new knowledge or applications, so these awards also recognize the academic and social value of the selected research.

More news
View : All
1st semester
2nd semester
No news in this category
Show more
Nothing to show