Semantic Role Labeling
Semantic Role Labeling is a multidisciplinary topic between Artifical Intelligence (Natural Language Processing and Understanding) and Linguistics Semantics. Semantic Role Labelling deals with the semantic analysis of sentences, often, for practical applications, such as Information Extraction, Question Answering, Information Retrieval, Text Summarization and so on. In this analysis, words and phrases in a sentence are labeled with Conceptual Roles such as Agent, Theme, Patient, Cause, etc. hoping that they can provide for a minimum level of language understanding (particularly for machines) so that questions such as who, whom, why, what, when, where, how, etc. can be answered in a systematic way.
This is an introductory course in which we will address three major topics:
- - Defining key concepts in semantic role labeling and argument-structure analysis
- - Introducing lexical knowledge bases that are developed for this purpose
- - Introducing computational methods, mostly supervised machine learning techniques, for developing semantic role labeler
The content will be offered in a way that it can be helpful to the audience of different background (Linguistics, Information Science, Computational Linguistics).
- - Download slides for Session 1 here
- - Download slides for Session 2 here
- - Download slides for Session 3 here
- - Download slides for Session 4 here
- - Download slides for Session 5 here Processable Levin Patterns / Verb Classes
- - Download slides for Session 6 here Get FrameNet From Here Browse in HHU Server Short frame def. in one page Help us! Browse it on a concordancer: ex. 1ex. 2
- - Download slides for Session 7 (Propbank) and session 8 (VerbNet etc.) (See also please NomBank and Nomlex)
- - Download slides for Session 9 and 10 -- An introduction to introduction of classification and machine learning.
- - Download slides for Session 11 -- Feature Extraction.
- - Download slides for Session 12 and 13 -- Semantic Role Classification.