CmpE 561 - Natural Language Processing

NEWS:

Deadline for the final project is June 14. Your task is to write a program which takes a Turkish sentence as input and which produces all questions that can be produced from this sentence (having this sentence as the answer) as output. You are free to use any approach, although I really don’t think that you can create ALL possible questions (without also creating lots of unsyntactic ones) if you use no knowledge of syntax.

(E.g. “Köpeğin kuyruğu yeşildi” -> “Ne yeşildi?”)

You are free to use the infrastructure existing in the files to be provided by the instructor, or, if you wish, write everything from scratch.

The new deadline for the syntax project is the final exam date; all demos will be done on that day. In this project, you start by requesting the syntax-related PROLOG files from the instructor by email. Your task is to select a text of at least 20 sentences from a Turkish newspaper website (use a site which has an archive), parse these sentences, determine which syntax rules are necessary for this job, and write the necessary DCG rules which would successfully analyze these sentences (and all other sentences with similar structures) and produce representations of their parse trees, as shown in Chapter 3 of the textbook. Obviously, if the words appearing in the sentences are not present in your lexicon, you have to add the appropriate entries there. You are free to use the infrastructure existing in the files to be provided by the instructor, or, if you wish, write everything from scratch.

Deadline for late submissions for the ELIZA and poetry projects is May 15.

Midterm results are HERE

Coordinates:	FFF 678, ETAZ09
Textbook:	“Natural Language Processing for PROLOG Programmers” Michael A. Covington, 1994.
References:	“Natural Language Processing in PROLOG: An Introduction to Computational Linguistics” Gerald Gazdar, Chris Mellish, 1989. “Prolog for Natural Language Processing” Annie Gal, Guy Lapalme, Patrick Saint-Dizier, Harold Somers. John Wiley & Sons, 1991.
Instructor:	Cem Say
Goals:	To acquaint the student with the basics of NLP using the PROLOG language

Prerequisite: Knowledge of Turkish

You do NOT need to have taken a PROLOG course earlier, we will teach a sufficient amount of PROLOG to people who don’t have previous experience with it in an early stage in this course.

Projects

Three PROLOG programming projects involving Turkish NLP

Plan: (Subject to change at any moment!)

1. Natural Language Processing: An Introduction

2. Overview of PROLOG

3. Templates and Keywords

4. Finite-State Techniques

5. Morphology

6. Turkish Morphology

7. Introduction to Syntax

8. Syntax with DCG’s

9. Turkish Syntax/Introduction to Semantics

10. Semantics, Logic, and Model Theory

11. Further Topics in Semantics, Generating Paraphrases

1 Midterm, 1 Final exam