(in Polish) Przetwarzanie języka naturalnego i sztuczna inteligencja 2100-CB-M-D3PJSI
Course program:
Basic information about storing text in digital form.
ASCII character set and encoding of characters outside this set.
UTF encodings, with special emphasis on UTF8.
Issues related to languages with extensive inflection (e.g. Polish), suffixes, prefixes, stems, circumflexes, lemmatization.
Basics of regular expressions
Refining texts for further analysis.
TF-IDF and Zipf's law
AI basics - embedding texts in vector space.
Language models.
Topic detection as an application of cluster search and dimension reduction methods.
Analysis of emotional overtones as an example of application of language models.
Mode
Prerequisites (description)
Course coordinators
Main fields of studies for MISMaP
Type of course
Requirements
Learning outcomes
Topic and keyword detection and analysis of emotional overtones of texts as a method of detecting unfriendly information and threats in the infosphere (K_W05)
Application of language models and AI methods to analyze texts for false, unfriendly information or topics indicating possible threats (K_W10)
Analysis of texts from selected sources using AI tools as a method of early detection of threats - at the stage of their planning (K_U03)
Demonstrate early threat detection from textual sources and raise awareness of the importance of the infosphere as a space where potential threat sources and broad information attack vectors can be found (K_K01).
Assessment criteria
Execution of an individual credit project on the basis of selected information sources.
Practical placement
n/a
Bibliography
Sowmya Vajjala, Bodhisattwa Majumder, Anuj Gupta, Harshit Surana:
Przetwarzanie języka naturalnego w praktyce. Przewodnik po budowie rzeczywistych systemów NLP, Helion
Aleksander Molak: Wnioskowanie i związki przyczynowe w Pythonie. Nowoczesne uczenie maszynowe z wykorzystaniem bibliotek DoWhy, EconML, PyTorch i nie tylko, Helion
https://huggingface.co
https://spacy.io
Additional information
Additional information (registration calendar, class conductors, localization and schedules of classes), might be available in the USOSweb system: