On automatically generating source code

Staff - Faculty of Informatics

Date: 31 March 2022 / 16:30 - 17:30

join here

Speaker: Matteo Ciniselli

Abstract: Recommender systems for software developers have been defined as software applications that can provide information able to support developers in different tasks. Among those, code recommenders, including code completion tools, are considered one of the killer features of modern Integrated Development Environments (IDEs). 
In this talk I'm going to present our research in this area. I'll mostly focus on two works. In the first we experimented with different data-driven techniques to investigate their behavior in challenging code completion scenarios, like predicting an entire block of code. In the second, I present our preliminary study that investigates the limitations of state-of-the-art models, and the characteristics of code recommender systems desired by developers.

Biography: Matteo Ciniselli is a Ph.D. student at the Software Institute (USI) in Lugano, Switzerland. He received his MSc. in Mathematical Engineering from Politecnico di Milano, Italy, in April 2015. His research interests include the study of deep-learning models to support code-related tasks. More information available at: https://www.inf.usi.ch/phd/cinism.

Chair: Jesper Findahl

*************************

In February 2019, the Software Institute started its SI Seminar Series. Every Thursday afternoon, a researcher of the Institute will publicly give a short talk on a software engineering argument of their choice. Examples include, but are not limited to novel interesting papers, seminal papers, personal research overview, discussion of preliminary research ideas, tutorials, and small experiments.

On our YouTube playlist you can watch some of the past seminars. More details on the next seminar, the upcoming seminars, and an archive of the past speakers are available here.