Commonsense knowledge-aided news analysis for stock market surveillance

Xin Li; Kun Chen; Terrance Fung; Sherry X. Sun; Huaiqing Wang

Commonsense knowledge-aided news analysis for stock market surveillance

Xin Li
, Kun Chen
, Terrance Fung
, Sherry X. Sun
, Huaiqing Wang

Research output: Unpublished conference presentation (presented paper, abstract, poster) › Conference presentation (not published in journal/proceeding/book) › Academic research › peer-review

Abstract

Stock market surveillance is critical to maintain market fairness and uphold investors' confidence. This research takes a text mining approach to inspect news to help surveillance specialists investigate suspicious stock transactions. Noticing the important role of prior knowledge in humans' news comprehension, we propose to incorporate commonsense knowledge into this task through a graph model. Experiments on a dataset collected from the Hong Kong stock market show that commonsense knowledge, especially features extracted from inter-news commonsense relations, can significantly improve market surveillance performance.

Original language	English
Publication status	Published - 2010
Externally published	Yes
Event	20th Annual Workshop on Information Technologies and Systems, WITS 2010 - St. Louis, MO, United States Duration: 11 Dec 2010 → 12 Dec 2010

Conference

Conference	20th Annual Workshop on Information Technologies and Systems, WITS 2010
Country/Territory	United States
City	St. Louis, MO
Period	11/12/10 → 12/12/10

ASJC Scopus subject areas

Information Systems

Cite this

@conference{9759a9bbe4d0430882babd1abe351cee,

title = "Commonsense knowledge-aided news analysis for stock market surveillance",

abstract = "Stock market surveillance is critical to maintain market fairness and uphold investors' confidence. This research takes a text mining approach to inspect news to help surveillance specialists investigate suspicious stock transactions. Noticing the important role of prior knowledge in humans' news comprehension, we propose to incorporate commonsense knowledge into this task through a graph model. Experiments on a dataset collected from the Hong Kong stock market show that commonsense knowledge, especially features extracted from inter-news commonsense relations, can significantly improve market surveillance performance.",

author = "Xin Li and Kun Chen and Terrance Fung and Sun, \{Sherry X.\} and Huaiqing Wang",

year = "2010",

language = "English",

note = "20th Annual Workshop on Information Technologies and Systems, WITS 2010 ; Conference date: 11-12-2010 Through 12-12-2010",

}

Commonsense knowledge-aided news analysis for stock market surveillance. / Li, Xin; Chen, Kun; Fung, Terrance et al.
2010. Paper presented at 20th Annual Workshop on Information Technologies and Systems, WITS 2010, St. Louis, MO, United States.

Research output: Unpublished conference presentation (presented paper, abstract, poster) › Conference presentation (not published in journal/proceeding/book) › Academic research › peer-review

TY - CONF

T1 - Commonsense knowledge-aided news analysis for stock market surveillance

AU - Li, Xin

AU - Chen, Kun

AU - Fung, Terrance

AU - Sun, Sherry X.

AU - Wang, Huaiqing

PY - 2010

Y1 - 2010

N2 - Stock market surveillance is critical to maintain market fairness and uphold investors' confidence. This research takes a text mining approach to inspect news to help surveillance specialists investigate suspicious stock transactions. Noticing the important role of prior knowledge in humans' news comprehension, we propose to incorporate commonsense knowledge into this task through a graph model. Experiments on a dataset collected from the Hong Kong stock market show that commonsense knowledge, especially features extracted from inter-news commonsense relations, can significantly improve market surveillance performance.

AB - Stock market surveillance is critical to maintain market fairness and uphold investors' confidence. This research takes a text mining approach to inspect news to help surveillance specialists investigate suspicious stock transactions. Noticing the important role of prior knowledge in humans' news comprehension, we propose to incorporate commonsense knowledge into this task through a graph model. Experiments on a dataset collected from the Hong Kong stock market show that commonsense knowledge, especially features extracted from inter-news commonsense relations, can significantly improve market surveillance performance.

UR - https://www.scopus.com/pages/publications/84900462364

M3 - Conference presentation (not published in journal/proceeding/book)

AN - SCOPUS:84900462364

T2 - 20th Annual Workshop on Information Technologies and Systems, WITS 2010

Y2 - 11 December 2010 through 12 December 2010

ER -

Commonsense knowledge-aided news analysis for stock market surveillance

Abstract

Conference

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this