Incorporate domain knowledge into support vector machine to classify price impacts of unexpected news

Ting Yu, Tony Jan, John Debenham, Simeon Simoff

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Citations (Scopus)

Abstract

We present a novel approach for providing approximate answers to classifying news events into simple three categories. The approach is based on the authors' previous research: incorporating domain knowledge into machine learning [1], and initially explore the results of its implementation for this particular field. In this paper, the process of constructing training datasets is emphasized, and domain knowledge is utilized to pre-process the dataset. The piecewise linear fitting etc. is used to label the outputs of the training datasets, which is fed into a classifier built by support vector machine, in order to learn the interrelationship between news events and volatility of the given stock price.

Original languageEnglish
Title of host publicationAusDM 2005 Proc. - 4th Australasian Data Mining Conf. - Collocated with the 18th Australian Joint Conf. on Artificial Intelligence, AI 2005 and the 2nd Australian Conf. on Artificial Life, ACAL 2005
Pages1-11
Number of pages11
Publication statusPublished - 2005
Externally publishedYes
Event4th Australasian Data Mining Conference, AusDM 2005 - Collocated with the 18th Australian Joint Conference on Artificial Intelligence, AI 2005 and the 2nd Australian Conference on Artificial Life, ACAL 2005 - Sydney, NSW, Australia
Duration: 5 Dec 20056 Dec 2005

Publication series

NameAusDM 2005 Proc. - 4th Australasian Data Mining Conf. - Collocated with the 18th Australian Joint Conf. on Artificial Intelligence, AI 2005 and the 2nd Australian Conf. on Artifical Life, ACAL 2005

Conference

Conference4th Australasian Data Mining Conference, AusDM 2005 - Collocated with the 18th Australian Joint Conference on Artificial Intelligence, AI 2005 and the 2nd Australian Conference on Artificial Life, ACAL 2005
Country/TerritoryAustralia
CitySydney, NSW
Period5/12/056/12/05

Fingerprint

Dive into the research topics of 'Incorporate domain knowledge into support vector machine to classify price impacts of unexpected news'. Together they form a unique fingerprint.

Cite this