Hope the topic is not too broad for this forum. Last year I posted a twitter asking for references to prepare a course on introduction to NLP:
@ebender was very helpful suggesting her books that are great! But I am curious to repeat the question here. What would be a great course to introduce NLP. Materials? Suggestions of content and practical activities? How to avoid the ML-solve-all-my-problems approach and bring people for the linguistic-motivated side of NLP?!
Here’s some materials I put together with my seminar participants earlier this year:
Some of it may be relevant, I think.
In case it’s useful - I believe the following can be accessed by anyone - let me know if not:
The notes are mine I believe (it’s a course I developed and taught for years), slides are partly mine and partly others but I guess they all have attributions on them. Weiwei Sun revised the course and taught it last year but you’d need to ask her for access to that material.
Not sure why the link wasn’t there
OK - this time I put the link in quotes and it still disappeared. Will just email it separately if anyone wants it.
@AnnC, did you want to link to one of these pages?
Thanks - the relevant link is the Course Materials tab on the second of the pages - directly linking to the tabbed page is what doesn’t work. The top link doesn’t give public access to the notes.
Thank you @AnnC , @olzama, and @guyemerson !! I will check the materials. These links will give me a good idea; I have one month to prepare for the course; I saw the primary reference is Speech and Language Processing, the same I was considering.
So you didn’t even mention HPSG?
The assignments are interesting; I will need to check to see if they are feasible for the students we have here and if I will have time to make them myself before asking them!
Jurafsky and Martin used to mention HPSG (you will find a chapter on contraint-based formalisms in the second edition) but they dropped it in the 3rd edition
The Cambridge course also used to include feature structures, e.g. see Computer Laboratory – Course pages 2014–15: Natural Language Processing
Sorry - sent this previously but just as a direct reply
HPSG is mentioned a couple of times in 5.5 in the notes - it’s a background section admittedly. I used to teach a simple unification-based approach, but gave up. Happy to discuss why over a virtual coffee at some point but basically I wanted to start with CFG (because they are compscis so know about that from programming languages) and then teaching them something about dependencies seemed more important.
I would be very interesting to hear more about the history of the course and why pure unification-based approach was abandoned. My course start this Monday, two classes per week until December. First week will be mostly general introduction. Still deciding for the best topics to cover given that I didn’t have much time for prepare material. But I will surely use some of the slides and course notes from Department of Computer Science and Technology – Course pages 2019–20: Natural Language Processing!
Happy to send you latex for the notes and slides if you want.
You can treat my material as CC-BY but in the cases where I used other people’s stuff (I think mainly Aurelie Herbelot’s - it should be attributed where relevant) you’d need to check that was OK with them.
Oh, that would be great! Sure I will give the attributes. I can check with her! The latex will save me a lot of time!!!
By the way, how long is a lecture in Cambridge? I was not able to present Finite-state techniques in only one lecture… my bad?!