Beskjeder
It appears that reading list includes the Early Parser (a specific, top-down instantiation of our generalized chart parsing framework), but we did discuss it specifically in the lectures. I do not have access to an up-to-date copy of Jurafsky & Martin (2008) right now, but Section 13.4.2 should be the one on the Early Algorithm, which we do not require you to prepare for the exam, though do recommend as useful background to our notion of generalized chart parsing.
Last week, we posted the results from our third quiz in devilry; we have heard no complaints, so assume everyone agreed to our scoring. Jakob is busy grading submissions to Problem Set (3b), and we will record who has qualified for the exam in the course of today (probably 35 of the 39 who submitted for the final problem set; in comparison to 48 who submitted for the first problem set, this seems like a decent success rate to us).
There is a detailed model solution for (3b), including comments to the theoretical questions, in SVN. In case you have any specific questions, please email us at the inf4820-help address. in the lecture tomorrow, we will give a shorter high-level summary, to then use the bulk of the session to prepare for the final exam. Please work through our...
We just updated our model solution in SVN (and added a link to the course schedule), to address an interesting corner case: the default value for array cells is not defined by the Common Lisp standard, unless the :initial-element keyword argument is given to make-array(). Because our code depends on ‘empty’ cells (corresponding to unseen events) to contain nil (the default element in ACL, but apparently not in SBCL), this is an important correction in terms of standard compliance. At the same time, we have filled in missing comments and generally tried to improve readability of the code; please see for yourselves.
Jakob is working hard to send feedback as fast as possible, and by now has completed reviewing well over half of the submissions we received.
We have just posted the final screen cast from last week and our next Problem Set (3a), with its submission deadline of Wednesday, October 29. A model solution for Problem Set (2b) will be presented in the laboratory session this coming Monday, and there will be time to start theory and coding for (3a).
Jakob is already making good progress posting feedback to submissions for (2b), and we believe we have recorded all point awarded for our Bonus Problem Set (2c) in Devilry; if you feel you should have been rewarded for submitting to (2c) but no points show in Devilry, please make contact with us.