SGN-9206 Signal processing graduate seminar II, Fall 2007
Semantic analysis of text and speech

Semantic analysis tries to figure out the meaning of linguistic input. In other words, language is processed so as to produce common-sense knowledge about the world. Applications include information retrieval, human-computer interaction, machine translation, and many others, some of which are covered by the course. Semantic analysis is closely related to artificial intelligence but is a narrower and better defined field.

Several methods for semantic analysis exist and there are active projects on this topic at the leading universities of the world. The aim of the course is to cover the mainstream approaches and to review case studies representing the state-of-the-art.

The problem is clearly not yet solved and one aim of the course is to analyze the strengths and weaknesses of various approaches and encourage creative thinking.

Course information

Lectured Fall 2007, Periods I - II

5 credits

Suitable for postgraduate studies

Instructor: Anssi Klapuri

Literature: Speech and language processing (book by Jurafsky and Martin) and selected articles. The book can be borrowed from Anssi Klapuri (TF311) or Jouni Paulus (TC318).

Lecture time and place

Periods I-II: Tuesday 14 - 16, TB223

Scheduled presentations and slides of past presentations (forgot your password? e-mail anssi.klapuri at

Seminar attendance.


Exam will be on Wed 28.11.2007 at 9.00 o'clock (room not yet known). Please remember to register to the exam normally.

Practical implementation

The first 2 x 45 minute lectures are given by the instructor. The rest of the seminar consists of student presentations.

Each individual student prepares a 45-minute presentation on a topic chosen by the instructor. The students are allowed to tell their preferences in choosing the topic.

The students act as "opponents" to each other, giving feedback and asking questions before each presentation (based on draft slides prepared by the presented) and during the presentation. Each student acts as an opponent once.

Course plan:

Requirements for passing the course

This seminar in previous years

Last modified: Tue Sep 25 10:46:21 EEST 2007 - Anssi Klapuri, klap @ cs tut fi