Teacher
Giovanni Da San Martino
Dipartimento di Matematica, Università degli Studi di Padova,
giovanni.dasanmartino[at]unipd.it
INF/01

Aim
Dealing with texts in natural language poses several challenges, for example, to machine learning algorithms: texts are discrete signals and the meaning of a word depends on its context; moreover, the addition of a single word to a sentence can drastically change its meaning. We will review how words, sentences and entire documents are represented.
In the second part of the course, we will review some natural language processing tasks that require ad hoc modeling strategies, not commonly discussed in standard machine learning courses. Such tasks include information extraction and text span identification (used in question answering, machine summarisation, etc...).

Syllabus
1. Natural Language Processing pipelines versus end-to-end models
2. How to represent words and sentences in a (machine-learning) algorithm
3. Natural Language Processing tasks requiring ad hoc modeling: an overview

Introductory reading
None

Course requirements
Although the course will try not to assume any previous knowledge, having attended any basic machine learning course would help.

Examination modality
To be defined

Course material, enrollment and last-minute notifications
Made available by the teacher at this Moodle address

Schedule
18 January 2024, 10:00-12:00 (Room 2AB40)
25 January 2024, 10:00-12:00 (Room 2AB40)
30 January 2024, 10:00-12:00 (Room 1BC50)
12 February 2024, 10:00-12:00 (Room 2AB45)

Location
Rooms written in the schedule above and located at the Dept. of Mathematics, via Trieste 63 Padova

<< Courses in 2023-2024