Social computing is an area of Computer Science that is concerned with the intersection of social behavior and computational systems. Loosely speaking, it is based on creating/recreating social conventions and social contexts through the use of software and technology. As such, it is often associated with the design of online social networks and social media applications as well as the analysis and modeling of user (individual and social) behavior when interacting with such platforms. The latter raises a multitude of challenges related to the inference, representation and modeling of user behavior patterns, notably due to the great diversity of signals, large volumes of (often noisy) data, as well as great heterogeneity and dynamicity of user behavior. Such challenges require a diverse set of tools, often relying on state-of-the-art network modeling, as well as approaches to deal with large volumes of (often textual) data, with techniques from areas such as machine learning (ML), natural language processing (NLP) and information retrieval (IR). In this course, we intend to discuss state-of-the-art problems and techniques in the convergence of social computing, machine learning, natural language processing and information retrieval.

.

In this course, we intend to discuss state-of-the-art problems and techniques in the convergence of social computing, machine learning, natural language processing and information retrieval. The course will be composed of three parts: 1. Introduction to social computing (8 hours) This module will focus on concepts related to network modeling, including concepts from Complex Networks as well as state-of-the-art techniques to deal with noisy networks. If there is the need for leveling and background, a brief portion of this module (2-3 hours) will be dedicated to present state-of-the-art (ML, NLP, IR) techniques such as Attention and Generative Models (such as Transformers and GPT) in the last part of this module. Otherwise, if the students have the necessary background, examples of state-of-the-art solutions using these techniques to solve Social Computing problems will be presented based on research developed by the professors ´ research groups. 2. Brief overview of quantitative evaluation in computer science (2 hours) This module will review fundamental concepts related to the scientific methodological procedure required for proper evaluation of the methods discussed in the other modules. We intend to discuss issues related to training/test data splitting, statistical tests, confidence intervals, etc. 3. Guided discussions on fundamental and recent papers (10 hours) This module will consist of guided discussions led by the two instructors about key and recent papers on the convergence of the fields. We will focus on open problems as well as future directions. In the discussions we will pay special attention to issues concerning experimental design correctness, hypothesis formulation, statistical validity and confirmation of results, presentation issues, reproducibility among other concerns. The papers will be presented by the course ´s students in the form of seminars. Students are supposed to study the assigned papers (from a previous list compiled by the professors, prepare the presentations and present them in class. The other students are also expected to read the papers beforehand and prepare questions for the presenter to motivate and drive discussions during the class

In presenza

Presentazione report scritto

P.D.2-2 - Marzo

Grading will be based on: 10% for attendance and participation in the discussions 30% for the seminar 60% for a short research project For the research project, there will be a list of potential themes that the instructors will explain in the beginning of the course. The research projects can be developed by groups of up to 2-3 students. Each group will have to submit a short paper on the project by the end of the course. Since this is supposed to be a research project, we understand that the results may not meet the original expectations. Thus, grading will be based on the quality of the work developed in terms of the clarity of presentation (clearly defined hypothesis), adherence to the general topic of the course and robustness of the applied evaluation methodology.