Knowledge-driven stock trend prediction and explanation via temporal convolutional network

Shumin Deng, Ningyu Zhang, Wen Zhang, Jiaoyan Chen, Jeff Z. Pan, Huajun Chen

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Downloads (Pure)

Abstract

Deep neural networks have achieved promising results in stock trend prediction. However, most of these models have two common drawbacks, including (i) current methods are not sensitive enough to abrupt changes of stock trend, and (ii) forecasting results are not interpretable for humans. To address these two problems, we propose a novel Knowledge-Driven Temporal Convolutional Network (KDTCN) for stock trend prediction and explanation. Firstly, we extract structured events from financial news, and utilize external knowledge from knowledge graph to obtain event embeddings. Then, we combine event embeddings and price values together to forecast stock trend. We evaluate the prediction accuracy to show how knowledge-driven events work on abrupt changes. We also visualize the effect of events and linkage among events based on knowledge graph, to explain why knowledge-driven events are common sources of abrupt changes. Experiments demonstrate that KDTCN can (i) react to abrupt changes much faster and outperform state-of-the-art methods on stock datasets, as well as (ii) facilitate the explanation of prediction particularly with abrupt changes.

Original languageEnglish
Title of host publicationCompanion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion)
EditorsLing Liu, Ryen White
Place of PublicationNew York
PublisherAssociation for Computing Machinery, Inc
Pages678-685
Number of pages8
ISBN (Electronic)9781450366755
DOIs
Publication statusPublished - 13 May 2019
Event2019 World Wide Web Conference, WWW 2019 - San Francisco, United States
Duration: 13 May 201917 May 2019

Conference

Conference2019 World Wide Web Conference, WWW 2019
CountryUnited States
CitySan Francisco
Period13/05/1917/05/19

Fingerprint

Experiments
Deep neural networks

Keywords

  • Event extraction
  • Explanation
  • Knowledge-driven
  • Predictive analytics
  • Stock trend prediction
  • Structured
  • Unstructured
  • REPRESENTATIONS
  • predictive analytics
  • stock trend prediction
  • explanation
  • event extraction
  • unstructured
  • structured

ASJC Scopus subject areas

  • Software
  • Computer Networks and Communications

Cite this

Deng, S., Zhang, N., Zhang, W., Chen, J., Pan, J. Z., & Chen, H. (2019). Knowledge-driven stock trend prediction and explanation via temporal convolutional network. In L. Liu, & R. White (Eds.), Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion) (pp. 678-685). New York: Association for Computing Machinery, Inc. https://doi.org/10.1145/3308560.3317701

Knowledge-driven stock trend prediction and explanation via temporal convolutional network. / Deng, Shumin; Zhang, Ningyu; Zhang, Wen; Chen, Jiaoyan; Pan, Jeff Z.; Chen, Huajun.

Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion). ed. / Ling Liu; Ryen White. New York : Association for Computing Machinery, Inc, 2019. p. 678-685.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Deng, S, Zhang, N, Zhang, W, Chen, J, Pan, JZ & Chen, H 2019, Knowledge-driven stock trend prediction and explanation via temporal convolutional network. in L Liu & R White (eds), Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion). Association for Computing Machinery, Inc, New York, pp. 678-685, 2019 World Wide Web Conference, WWW 2019, San Francisco, United States, 13/05/19. https://doi.org/10.1145/3308560.3317701
Deng S, Zhang N, Zhang W, Chen J, Pan JZ, Chen H. Knowledge-driven stock trend prediction and explanation via temporal convolutional network. In Liu L, White R, editors, Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion). New York: Association for Computing Machinery, Inc. 2019. p. 678-685 https://doi.org/10.1145/3308560.3317701
Deng, Shumin ; Zhang, Ningyu ; Zhang, Wen ; Chen, Jiaoyan ; Pan, Jeff Z. ; Chen, Huajun. / Knowledge-driven stock trend prediction and explanation via temporal convolutional network. Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion). editor / Ling Liu ; Ryen White. New York : Association for Computing Machinery, Inc, 2019. pp. 678-685
@inproceedings{1f94f8f747484c6dbe80b0cdaadb6cd3,
title = "Knowledge-driven stock trend prediction and explanation via temporal convolutional network",
abstract = "Deep neural networks have achieved promising results in stock trend prediction. However, most of these models have two common drawbacks, including (i) current methods are not sensitive enough to abrupt changes of stock trend, and (ii) forecasting results are not interpretable for humans. To address these two problems, we propose a novel Knowledge-Driven Temporal Convolutional Network (KDTCN) for stock trend prediction and explanation. Firstly, we extract structured events from financial news, and utilize external knowledge from knowledge graph to obtain event embeddings. Then, we combine event embeddings and price values together to forecast stock trend. We evaluate the prediction accuracy to show how knowledge-driven events work on abrupt changes. We also visualize the effect of events and linkage among events based on knowledge graph, to explain why knowledge-driven events are common sources of abrupt changes. Experiments demonstrate that KDTCN can (i) react to abrupt changes much faster and outperform state-of-the-art methods on stock datasets, as well as (ii) facilitate the explanation of prediction particularly with abrupt changes.",
keywords = "Event extraction, Explanation, Knowledge-driven, Predictive analytics, Stock trend prediction, Structured, Unstructured, REPRESENTATIONS, predictive analytics, stock trend prediction, explanation, event extraction, unstructured, structured",
author = "Shumin Deng and Ningyu Zhang and Wen Zhang and Jiaoyan Chen and Pan, {Jeff Z.} and Huajun Chen",
note = "The authors would like to acknowledge that this work is funded by NSFC 61473260/91846204, national key research program YS2018YFB140004 as well as Natural Science Foundation of Zhejiang Province of China (LQ19F030001), and supported by Alibaba-Zhejiang University Joint Institute of Frontier Technologies.",
year = "2019",
month = "5",
day = "13",
doi = "10.1145/3308560.3317701",
language = "English",
pages = "678--685",
editor = "Ling Liu and Ryen White",
booktitle = "Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion)",
publisher = "Association for Computing Machinery, Inc",

}

TY - GEN

T1 - Knowledge-driven stock trend prediction and explanation via temporal convolutional network

AU - Deng, Shumin

AU - Zhang, Ningyu

AU - Zhang, Wen

AU - Chen, Jiaoyan

AU - Pan, Jeff Z.

AU - Chen, Huajun

N1 - The authors would like to acknowledge that this work is funded by NSFC 61473260/91846204, national key research program YS2018YFB140004 as well as Natural Science Foundation of Zhejiang Province of China (LQ19F030001), and supported by Alibaba-Zhejiang University Joint Institute of Frontier Technologies.

PY - 2019/5/13

Y1 - 2019/5/13

N2 - Deep neural networks have achieved promising results in stock trend prediction. However, most of these models have two common drawbacks, including (i) current methods are not sensitive enough to abrupt changes of stock trend, and (ii) forecasting results are not interpretable for humans. To address these two problems, we propose a novel Knowledge-Driven Temporal Convolutional Network (KDTCN) for stock trend prediction and explanation. Firstly, we extract structured events from financial news, and utilize external knowledge from knowledge graph to obtain event embeddings. Then, we combine event embeddings and price values together to forecast stock trend. We evaluate the prediction accuracy to show how knowledge-driven events work on abrupt changes. We also visualize the effect of events and linkage among events based on knowledge graph, to explain why knowledge-driven events are common sources of abrupt changes. Experiments demonstrate that KDTCN can (i) react to abrupt changes much faster and outperform state-of-the-art methods on stock datasets, as well as (ii) facilitate the explanation of prediction particularly with abrupt changes.

AB - Deep neural networks have achieved promising results in stock trend prediction. However, most of these models have two common drawbacks, including (i) current methods are not sensitive enough to abrupt changes of stock trend, and (ii) forecasting results are not interpretable for humans. To address these two problems, we propose a novel Knowledge-Driven Temporal Convolutional Network (KDTCN) for stock trend prediction and explanation. Firstly, we extract structured events from financial news, and utilize external knowledge from knowledge graph to obtain event embeddings. Then, we combine event embeddings and price values together to forecast stock trend. We evaluate the prediction accuracy to show how knowledge-driven events work on abrupt changes. We also visualize the effect of events and linkage among events based on knowledge graph, to explain why knowledge-driven events are common sources of abrupt changes. Experiments demonstrate that KDTCN can (i) react to abrupt changes much faster and outperform state-of-the-art methods on stock datasets, as well as (ii) facilitate the explanation of prediction particularly with abrupt changes.

KW - Event extraction

KW - Explanation

KW - Knowledge-driven

KW - Predictive analytics

KW - Stock trend prediction

KW - Structured

KW - Unstructured

KW - REPRESENTATIONS

KW - predictive analytics

KW - stock trend prediction

KW - explanation

KW - event extraction

KW - unstructured

KW - structured

UR - http://www.scopus.com/inward/record.url?scp=85066884024&partnerID=8YFLogxK

UR - http://dl.acm.org/citation.cfm?doid=3308560.3317701

UR - http://www.mendeley.com/research/knowledgedriven-stock-trend-prediction-explanation-via-temporal-convolutional-network

U2 - 10.1145/3308560.3317701

DO - 10.1145/3308560.3317701

M3 - Conference contribution

SP - 678

EP - 685

BT - Companion Proceedings of the 2019 World Wide Web Conference (WWW ’19 Companion)

A2 - Liu, Ling

A2 - White, Ryen

PB - Association for Computing Machinery, Inc

CY - New York

ER -