Lecturers: Robin Cooper, Grégoire Détrez
Resit 2
Resit 2 is now available.
Final resit
The problems for the final resit for Computational Semantics (Spring 2011) will be made available 16th January, 2012. Deadline (non-negotiable) for submission of solutions is 3rd February, 2012. During the period of the resit we will answer questions concerning the exercises and it will be possible for you to make submissions and obtain feedback so that you can improve your submission if necessary in time for the 3rd February deadline. However, we cannot guarantee feedback in time for resubmission on work that is submitted after 27th January.
Resit problem sets now available
The resit problem sets are now available. Deadline for submission is 9th September. You have received an individual email indicating which parts of the resit you need to complete if you have not already passed the course. Good luck!
New requirements and arrangements for resit in August
On looking over your final submissions for the course I recognize that many of you have had serious difficulties in completing the exercises. I have therefore decided to change the requirements. The new requirements are:
In order to pass the course (Swedish grade G) you need to obtain a passing grade on exercise sets 1-3.
In order to pass the course with distinction (Swedish grade VG) you need, in addition to passing the course, to obtain a passing grade on exercise set 4.
There will be a resit exam which will consist of similar exercise sets 1-4 and the requirements for that will be the same as above. However, those of you who have already obtained a passing grade on some of the exercise sets will not need to redo those exercise sets in the resit. The resit exercises will be made available on the course website on 19th Aug and the final deadline for submissions for the resit is 9th Sept. This deadline is not negotiable. During the period of the resit we will answer questions concerning the exercises and it will be possible for you to make submissions and obtain feedback so that you can improve your submission if necessary in time for the 9th September deadline. However, we cannot guarantee feedback in time for resubmission on work that is submitted after 6th Sept.
I will email each of you individually indicating the results you have obtained. If you have not yet passed the course I will indicate which parts of the resit you need to obtain a passing grade on in order to pass the course.
Deadline extension
We have decided to extend the final deadline of the course to Friday 10th June. We will accept both problem sets 1-4 and the VG problem set up until 11:59 pm on that date. No exercises (including revisions of earlier versions) will be accepted after the deadline. We will announce a resit possibility (if necessary) after that date which will take place in August and may involve different problem sets.
The course gives a basic introduction to model theoretical semantics for natural language (as developed for example in Montague semantics and Discourse Representation Theory) and its implementation in Python. It also introduces theorem proving and its application to reasoning in natural language applications.
We will start by drawing a parallel between the kind of models used in model theoretic semantics and databases and show how this can be exploited for natural language interfaces to databases. We will look at the nature of ambiguity in natural language, the kind of problems this creates for computational semantics and proposals that have been made to give underspecified meaning representations that cover several meanings in a single computational representation. We will consider the fact that semantics provides us with a tool to explain inferences that we draw when we are presented with information in natural language. We will consider how natural language semantics can be related to theorem provers and discuss the extent to which the inferences we need to account for in natural language are logical inferences. We will discuss how logical semantics is related to dynamic linguistic processing in real time and how this can give us a logic based view of dialogue management, that is, determining what a implemented dialogue system should say at a given point in a dialogue. Finally, we will look towards a future for computational semantics where machines learn new meanings as they interact with users, using their previous linguistic experience as a resource.
If you are taking this course please join the Google group http://groups.google.com/group/compsemvt11 so that you will receive mail concerning the course. All mail that has been sent to this group is archived at this url.