a semantic analyzer for aiding emotion recognition in chinese

14
A Semantic Analyzer for Aiding Emotion Recognition in Chinese ICIC 2006 Jiajun Yan, David B. Bracewell, Fuji Ren, and Shingo Ku roiwa. Department of Information Science and Intelligent Syste ms Faculty of Engineering, The University of Tokushima.

Upload: trory

Post on 19-Jan-2016

35 views

Category:

Documents


2 download

DESCRIPTION

A Semantic Analyzer for Aiding Emotion Recognition in Chinese. ICIC 2006. Jiajun Yan, David B. Bracewell, Fuji Ren, and Shingo Kuroiwa. Department of Information Science and Intelligent Systems Faculty of Engineering, The University of Tokushima. Introduction. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

A Semantic Analyzer for Aiding Emotion Recognition in Chinese

ICIC 2006

Jiajun Yan, David B. Bracewell, Fuji Ren, and Shingo Kuroiwa.Department of Information Science and Intelligent SystemsFaculty of Engineering, The University of Tokushima.

Page 2: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Introduction

• Semantic analysis helps to understand the roles and relations between objects, humans, etc. in the sentence.

• In this paper, we propose a system for understanding emotion in Chinese verbs.

• The emotion “felt toward” and “felt by” can be known.

Page 3: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

The SEEN System

Page 4: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Syntactic Analysis

• Morphological Analysis– Zhang, H., Yu, H., Xiong, D., Liu, Q.: Hhmm-based Chinese lexi

cal analyzer ictclas.In: the Second SIGHAN workshop affiliated with 41st ACL. (2003)

– Based on Hidden Markov Model

• Chinese Parsing– Zhou, Q.: A statistics-based Chinese parser. In: Proceedings of

the Fifth Workshop on Very Large Corpora. (1997)

– Because the parser was used on the Penn Chinese Treebank, and it is freely available

Page 5: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Headword Assignment

Page 6: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Structures

Page 7: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Structures

• Fig. 3. Another representation of a semantic dependency tree

Page 8: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Functional Tags

Page 9: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Semantic Dependency Assignment

• Decision Tree Classifier– 4 Features

• Phrase Type• Headword & Dependent• Headword & Dependent

Part-of-Speech• Context

– Accuracy 84%

Page 10: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Chinese Emotion Predicates

Page 11: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Chinese Emotion Predicates

Page 12: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Experimentation

• 80 sentences (10 sentences per predicate) were collected and examined.

• Negated emotions in English are not looked at. The semantic dependency was manually given.

• The accuracy was 100%.

Page 13: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

Experimentation

• Example of a Currently Unclassifiable Sentence

Page 14: A Semantic Analyzer for Aiding Emotion Recognition in Chinese

SEEN System (After)

• DT Classifier -> Probabilistic Classification

• Add Rule-Based Correction

• Accuracy increase to 85.1%