Supervised Activities

Some of the completed thesis, project works, and other activities supervised by our research group

Master Theses

  • Small transformers for Bioinformatics tasks

    Student: Luca Salvatore Lorello
    Course: Master Degree in Artificial Intelligence
    Session: July 2021
    This thesis focuses on the training of small neural architectures for bioinformatics tasks concerning two genetic languages: ìthe human genome and the eukaryotic mitochondrial genome.

  • Advanced techniques for cross-language annotation projection in legal texts

    Student: Francesco Antici
    Course: Master Degree in Artificial Intelligence
    Session: July 2021
    This work investigates the cross-language annotation projection technique based on sentence embedding and similarity metrics to find matches between sentences. Several combinations of methods and algorithms are compared, among which there are monolingual and multilingual embedding neural models. The experiments are conducted projecting from English to Italian, German, and Polish.

  • Automatic extraction of scientific articles based on user queries expressed in natural language

    Realizzazione di un sistema per l'estrazione di articoli scientifici in base a query utente in linguaggio naturale (Italian Title)
    Student: Mauro Rondina
    Course: Master Degree in Computer Engineering
    Session: February 2021
    The development of a pipeline for automatic extraction of scientific articles from online resources and retrieval via user queries. Users can formulate short search queries in natural language (alike a web browser) and a ranking of most relevant scientific articles is produced as a result.

Bachelor Theses

  • Development of an argumentative chatbot in python

    Sviluppo di un chatbot in linguaggio python con l'integrazione di Argumentation Mining.
    Student: Federico Spurio
    Course: Bachelor Degree in Computer Engineering
    Period: July 2020
    A simple chatbot with the capability of retrieving top-K arguments given a user claim or statement. Arguments are automatically retrieved from a fixed Argumentation Mining (AM) dataset. Both fixed and dynamic (based on RL) user interfacing strategies are evaluated. In the latter scenario, the chatbot is able to learn from user feedback in order to adapt to the current interlocutor.

Project Works