Teaching code

4S008105

Teacher

Flavio Massimiliano Cecchini

Coordinator

Flavio Massimiliano Cecchini

Credits

6

Language

Italian

Scientific Disciplinary Sector (SSD)

L-LIN/01 - HISTORICAL AND GENERAL LINGUISTICS

Period

II semestre (Lingue e letterature straniere) dal Feb 15, 2021 al May 29, 2021.

Lessons timetable Moodle Seminars 0

Learning outcomes

The course illustrates concepts of linguistics combined with those of computer science. It will introduce the students to the basics of the science of language, of the different dimensions of natural languages and of the CA (Computer-Aided) tools relevant to the representation and to the treatment of data. At the end of the course, the student: - will acquire basic knowledge about natural language, CA methods and tools for the linguistic analysis; - will apply the knowledge using some tools during class for the representation and the treatment of linguistic data; - will start developing critical thinking about theories, methods and tools; - will be able to communicate and interact more effectively with technicians and experts in multimedia and IT processes; - will be able to find out more autonomously further bibliography, data and tools.

Program

The course is split into two main sections, roughly equivalent in size, each with the aim of presenting Computational Linguistics starting from one of its two extremes: the first section will focus on text, its representation and the formalisation of linguistical data, while the second section will introduce mathematical/statistical concepts and methods for the analysis of language (mainly intended as written text).

After a general introduction about the objectives of Computational Linguistics and a sketch of the discipline's history, the first part will regard the following topics:
- linguistic resources: corpora
- text encoding and standardisation: Unicode & UTF8
- linguistic units and annotational layers
- from sentences to trees: Universal Dependencies as an example of morphosyntactic annotation

The second part will deal with the following topics:
- statistical concepts related to the linguistic phenomenon: Zipf's law and word graphs
- automated analysis and instruments: Markov chains, entropy...
- mentions to different possibilities of automatic analysis: morphological analysis, Named Entity Recognition, Sentiment Analysis...
- non-projectivity as a particular case study for parsing (i.e. automated syntactic analysis)
- evaluation cycles and methods

For what concerns the second part in particular, the course has among its goals to back some of the theory with practical sessions involving the use of methods and programs from the own terminal. To this end, further directions will be given at the start of the course.

The course might be subjected to slight variations, based on previous knowledge by part of the students, on the possible need to tackle some topics more in-depth, and on the management of technical details.

Examination Methods

The exam will consist of two phases: a written composition of the student's choice but agreed upon by the professor, and an oral interrogation, both equivalent with regard to the final grade.

The form of the elaborate is not fixed, but it will need to focus on or take inspiration from the topics dealt with during the course and should be i nthe format of a 4-10-page article. It will be posssible to touch upon more theoretical or more practical aspects. Some ideas:
- create a small corpus and comment how automated linguistic analysis works for it
- use Python to write a script for tokenisation, parser evaluation, or similar
- develop and discuss an annotation strategy following the Universal Dependencies formalism for a language at your choosing (preferably NOT English nor Italian), and possibly confront it with already existing annotation styles

Among the main evaluation criteria, formal correctness will come before originality of the topic.

The oral interrogation is expected to last around half an hour (duration might vary according to the number of students taking the exam) and, starting from the composition, it will touch upon any of the course's topics, possibly comprehending reasoning questions.