Natural Language Processing - AA 2020/2021
Prof. Roberto Tedesco
Today's lecture is suspended.
The slide deck about affective computing has been modified a bit: just reordering of a few slides.
On slide deck about summarization and QA, slides 11,12, and 18 have been fixed.
Lecture of Wednesday October 7th is suspended.
Following a good question I received about vocabularies in NLP, I put a new slide 80 into the deck "NLP3-WORDS-RT.pdf".
Today's lecture will be held on-line.
I'm so sorry for the issues during the today's lab about NLTK. So, no video. Just attend the lecture next Wednesday, where I'm going to repeat NTLK intro and examples (hoping no more issues... fingers crossed!)
For the Lab 1 session, come with a computer and install NLTK (including corpora; see slide deck, about NLTK data). JupiterLab is also useful, although not compulsory.
Due to graduation exams, the NLP lecture of Thursday 24th is suspended.
Derivation of the alpha value, for the Katz backoff, added to file NLP3-derivations.zip
Slide deck "Error correction, prediction, N-grams": fixed slide 20. Slide 67, instead, was correct (because it was (1+1)*(1/3))
Slide deck "Introduction to NLP" fixed: group1 ODD numbers, group2 EVEN numbers
Web site is on-line
- Slides shown during lectures.
- Most of the course slides are based on the book: D. Jurafsky and J. H. Martin, Speech and Language Processing, 2nd edition.
- Another useful book: C. D. Manning and H. Schuetze, Foundations of Statistical Natural Language Processing.
- Introduction to the NLTK toolkit (version 3): S. Bird, E. Klein, and E. Loper: Natural Language Processing with Python (look at the HTML version)
- Papers (listed at the end of each slide package)
- Tools: NLTK toolkit (see above)
- POS tagging, chunking, and parsing
- Summarization and Question Answering
- Conversational Agents
- Paralinguistics and sentiment analysis
- Voice, ASR, TTS
- Machine Translation
- Deep Learning and NLP
- How to take exams:
- Exam is written and lasts 2h30min.
- Exam is composed of 3 topics, each composed of 3 questions (9 questions in total).
- Among such questions, you could find one or two simple exercises (drawing simple parse trees, calculate simple formulas about smoothing, etc.); you do not need any calculator.
- Whenever you want to attend an exam, or need to register your final mark, you must enroll. PLEASE, DO NOT FORGET TO ENROLL TO THE CALLS. NO EXCEPTIONS WILL BE CONSIDERED!
Download course materials
2020-09-16 (classroom - for both group 1 and group 2): Introduction and Morphology
2020-09-17 (on-line): Words (morphology, error correction, prediction, n-grams)
2020-09-18 (on-line): POS tagging
2020-09-18 (classroom, group 2) / 2020-09-23 (classroom, group 1): chunking, formal grammars
2020-09-25 (on-line): full parsing
2020-10-8 (on-line) and 2020-10-9 (on-line): Semantics
2020-10-15 (on line): Summarization and question answering
2020-10-16 (on line): Pragmatics (discourse) / 2020-10-22 (on line): Pragmatics (dialogue and CA)
2020-10-16 (classroom, group 2): Sentiment analysis (2020-10-21 for classroom, group 1, was not hold)
2020-10-22 (on line) / 2020-10-22 (classroom, group 2) and 2020-10-28 (classroom, group 1): Voice
2020-10-29 (on line) / 2020-10-30 (on line) / 2020-10-30 (classroom, group 2) / 2020-11-4 (held online, group 1): Neural Networks and NLP
- Lab 1 - 2020-09-25 (classroom, group 2) / 2020-09-30 (classroom, group 1; actually, hold on-line)
- Lab 2 - 2020-10-1 (on line)
- Lab 3 - 2020-10-9 (classroom, group 2) / 2020-10-14 (classroom, group 1)
At the ARCSLab laboratory, students and researchers study models ad technologies related to the development of adaptive and relational systems. Research topics include:
- Multi-channel access to web services and information
- Multimodal communication and storytelling
- Virtual contexts for rehabilitation and education
- NLP, including speech analysis and voice processing
- Particular attention is given to applications for fragility, disease and disability
Theses are available on such topics! Have a look!