Deep neural networks understand investors better

Nader Mahmoudi, Paul Docherty, Pablo Moscato

Research output: Contribution to journalArticleResearchpeer-review

3 Citations (Scopus)

Abstract

Studies that seek to examine the impact of sentiment in financial markets have been affected by inaccurate sentiment measurement and the use of inappropriate data. This study applies state-of-the-art techniques from the domain-general sentiment analysis literature to construct a more accurate decision support system that generates demonstrable improvement in investor sentiment classification performance compared with previous studies. The inclusion of emojis is shown significantly improve sentiment classification in traditional algorithms. Moreover, deep neural networks with domain-specific word embeddings outperform the traditional approaches for the classification of investor sentiment. The approach to sentiment classification outlined in this paper can be applied in future empirical tests that examine the impact of investor sentiment on financial markets.

Original languageEnglish
Pages (from-to)23-34
Number of pages12
JournalDecision Support Systems
Volume112
DOIs
Publication statusPublished - 1 Aug 2018

Keywords

  • Deep neural network (DNN)
  • Domain-specific
  • Emojis
  • Investor sentiment
  • StockTwits
  • Word embeddings

Cite this

Mahmoudi, Nader ; Docherty, Paul ; Moscato, Pablo. / Deep neural networks understand investors better. In: Decision Support Systems. 2018 ; Vol. 112. pp. 23-34.
@article{67070e5be4b44100bb8ff509df2c81b0,
title = "Deep neural networks understand investors better",
abstract = "Studies that seek to examine the impact of sentiment in financial markets have been affected by inaccurate sentiment measurement and the use of inappropriate data. This study applies state-of-the-art techniques from the domain-general sentiment analysis literature to construct a more accurate decision support system that generates demonstrable improvement in investor sentiment classification performance compared with previous studies. The inclusion of emojis is shown significantly improve sentiment classification in traditional algorithms. Moreover, deep neural networks with domain-specific word embeddings outperform the traditional approaches for the classification of investor sentiment. The approach to sentiment classification outlined in this paper can be applied in future empirical tests that examine the impact of investor sentiment on financial markets.",
keywords = "Deep neural network (DNN), Domain-specific, Emojis, Investor sentiment, StockTwits, Word embeddings",
author = "Nader Mahmoudi and Paul Docherty and Pablo Moscato",
year = "2018",
month = "8",
day = "1",
doi = "10.1016/j.dss.2018.06.002",
language = "English",
volume = "112",
pages = "23--34",
journal = "Decision Support Systems",
issn = "0167-9236",
publisher = "Elsevier BV",

}

Deep neural networks understand investors better. / Mahmoudi, Nader; Docherty, Paul; Moscato, Pablo.

In: Decision Support Systems, Vol. 112, 01.08.2018, p. 23-34.

Research output: Contribution to journalArticleResearchpeer-review

TY - JOUR

T1 - Deep neural networks understand investors better

AU - Mahmoudi, Nader

AU - Docherty, Paul

AU - Moscato, Pablo

PY - 2018/8/1

Y1 - 2018/8/1

N2 - Studies that seek to examine the impact of sentiment in financial markets have been affected by inaccurate sentiment measurement and the use of inappropriate data. This study applies state-of-the-art techniques from the domain-general sentiment analysis literature to construct a more accurate decision support system that generates demonstrable improvement in investor sentiment classification performance compared with previous studies. The inclusion of emojis is shown significantly improve sentiment classification in traditional algorithms. Moreover, deep neural networks with domain-specific word embeddings outperform the traditional approaches for the classification of investor sentiment. The approach to sentiment classification outlined in this paper can be applied in future empirical tests that examine the impact of investor sentiment on financial markets.

AB - Studies that seek to examine the impact of sentiment in financial markets have been affected by inaccurate sentiment measurement and the use of inappropriate data. This study applies state-of-the-art techniques from the domain-general sentiment analysis literature to construct a more accurate decision support system that generates demonstrable improvement in investor sentiment classification performance compared with previous studies. The inclusion of emojis is shown significantly improve sentiment classification in traditional algorithms. Moreover, deep neural networks with domain-specific word embeddings outperform the traditional approaches for the classification of investor sentiment. The approach to sentiment classification outlined in this paper can be applied in future empirical tests that examine the impact of investor sentiment on financial markets.

KW - Deep neural network (DNN)

KW - Domain-specific

KW - Emojis

KW - Investor sentiment

KW - StockTwits

KW - Word embeddings

UR - http://www.scopus.com/inward/record.url?scp=85049519992&partnerID=8YFLogxK

U2 - 10.1016/j.dss.2018.06.002

DO - 10.1016/j.dss.2018.06.002

M3 - Article

VL - 112

SP - 23

EP - 34

JO - Decision Support Systems

JF - Decision Support Systems

SN - 0167-9236

ER -