Review Reports - From Text Representation to Financial Market Prediction: A Literature Review

Round 1

Reviewer 1 Report

The paper is a review of text representation methods, sentiment analysis, and information retrieval methods from various data sources applied to investors' behavior in financial markets. What the authors achieve is a presentation of existing research directions in the area of text mining and deep learning with reference to financial markets - equity, cryptos, and foreign exchange.

Overall, I find the paper well-written and attentively conceived, although I believe it is somehow too long, and in many places, the text can be shortened. The point here is that the reader may lose interest in the topic addressed, so I recommend a revision of the manuscript from the perspective of text length.

One of the issues that I find rather strange is the authors' focus on publications in computer science and artificial intelligence domain, thus leaving aside publications in finance. Their choice is not explained and I believe leads to bias in the review. Moreover, the period under scrutiny is 2006-2022, without any explanation of this choice.

The Conclusions are rather weak, as no research limitations and directions for future research are present. The authors briefly mention several implications of their research, but more needs to be done for a proper Conclusions section.

Author Response

Authors’ Response to the Review Comments

Journal	Information
Manuscript ID	Information-1890814
Title of Paper	From Text Representation to Financial Market Prediction, A Systematic Review
Authors	Saeede Anbaee Farimani, Majid Vafaei Jahan, Amin Milani Fard
Revision	Minor revise

We appreciate the editor and referees' time and efforts in reviewing our manuscript and providing valuable comments.

This is the revised submission of our manuscript following the 'Minor revise' decision. We have addressed all of the referees' comments, and we hope that the revised manuscript meets the requested changes and you find it of acceptable quality to the Information journal.

Response to Reviewers' Comments

Reviewer 1:

(1): English language and style are fine/minor spell check required

We checked grammar style and spelling with Grammarly and a native English person. Thank you for your feedback.

(2): Does the introduction provide sufficient background and include all relevant references? Can be improve

We improved the introduction section aim to provide sufficient background and relevant references. Thanks.

(3):Are all the cited references relevant to the research? Can be improve

We removed redundant references that are not relevant to the finance and the NLP domain.

(4): Is the research design appropriate? Can be improve

We remove some redundant text and figures. Also, we add new section 2.1 to introduce our review methodology. Thank you for your comment.

(5): Are the methods adequately described? Can be improve

We presented more details about some methods in the introduction and section 2, especially adding more details about our review methodology in a new section 2.1. Thanks.

(6): Are the results clearly presented? Can be improve

We rewrote the conclusion section aim to cover research limitations and future research directions. Resolved.

(7): Are the conclusions supported by the results? Can be improve

We rewrote the conclusion section aim to cover research limitations and future research directions. Thank you.

We remove the redundant figures, tables, text, and references and reduced the page numbers by 2 pages. Thank you for your comment.

Thank you for your valuable comment. Aim to justify our choice for selecting the domain of reviewed publications, we updated text as follow:

We updated abstract as follow:

News dissemination in social media causes fluctuations in financial markets. (Scope)Recent advanced methods in deep learning-based natural language processing achieved promising results in financial market analysis. However, understanding how to leverage large amounts of textual data alongside financial market information is important for investors’ behavior analysis. “In this study, we review over 150 publications in the field of behavioral finance that jointly investigated Natural Language Processing (NLP) approaches and market data analysis for financial decision support. . This work differs from other reviews by focusing on applied publications in computer science and artificial intelligence contributed to heterogeneous information fusion for investors’ behavior analysis.” (Goal) We study the various text representation methods, sentiment analysis, and information retrieval methods from heterogeneous data sources. (Findings) We present current and future research directions in text mining and deep learning for correlation analysis, forecasting, and recommendation systems in financial markets such as stocks, crypto currencies, and Forex (Foreign Exchange Market).

Aim to justify our choice for our review duration 2006-mid 2022, we updated the second, third and fourth paragraph of introduction section.
We add section 2.1 “Review Methodology” for implication of our choices for selecting journal and conference publications that cover a wide range of topics, from finance and management science to computer science. Also, we noticed the reasons behind selecting review duration between 2006-mid 2022.

(3) The Conclusions are rather weak, as no research limitations and directions for future research are present. The authors briefly mention several implications of their research, but more needs to be done for a proper Conclusions section.

We updated the last paragraph of conclusion section aim to cover future research directions and our review limitations. Thank you for your feedback.

Author Response File: Author Response.docx

Reviewer 2 Report

It is necessary to specify the criteria that was applied to choose the information sources, and in effect, include their affiliations and relationships with the stock markets. In this way, the results will be better understood and a comparison can be made of how the study behaves.

I think the authors can be specified how they searched and located the information sources, the media economic and they can be put and develop a description of them and what if the networking or not between them. It is necessary to establish a relationship between these sources of information and which markets they represent, that is, if they are journalistic sources without any commercial or advertiser relationship or if they are actually spokespersons for the markets.

Author Response

Authors’ Response to the Review Comments

Journal	Information
Manuscript ID	Information-1890814
Title of Paper	From Text Representation to Financial Market Prediction, A Systematic Review
Authors	Saeede Anbaee Farimani, Majid Vafaei Jahan, Amin Milani Fard
Revision	Minor revise

We appreciate the editor and referees' time and efforts in reviewing our manuscript and providing valuable comments.

Response to Reviewers' Comments

Reviewer 2

(1) Are the methods adequately described? Can be improve

We presented more details about some methods in the introduction and section 2, especially adding more details about our review methodology in a new section 2.1. Thanks.

(2) Are the results clearly presented? Can be improve

We updated the last paragraph of conclusion section aim to cover future research directions and our review limitations. Thank you for your feedback.

(1 ) It is necessary to specify the criteria that was applied to choose the information sources, and in effect, include their affiliations and relationships with the stock markets. In this way, the results will be better understood and a comparison can be made of how the study behaves.

Aiming to specify the criteria for the selection of information sources, we add a new section 2.1 review methodology. In this section, we mention the way of selecting review sources from the prestigious journals and top-ranked conferences and refer to most frequent sources in table 1 in terms of citations and publications number. Thank you for your feedback. In second paragraph of this section we noticed “Most journalistic sources are without any commercial or advertiser relationship, while some conferences are sponsored by financial companies and stakeholders”.

(2) I think the authors can be specified how they searched and located the information sources, the media economic and they can be put and develop a description of them and what if the networking or not between them. It is necessary to establish a relationship between these sources of information and which markets they represent, that is, if they are journalistic sources without any commercial or advertiser relationship or if they are actually spokespersons for the markets.

Thank you for your feedback. In the second paragraph of section 2.1, we described the method we used for selecting reviewed papers. Besides, section 2.2 introduced the market and media Information sources were analyzed in reviewed work. In this section, we classified reviewed works based on market that they analyzed. Also, we noticed various media in terms of publication papers. We plot figure 4 that shows how [12,16] used the information published with spokespersons for the markets in their works and facilitate the future studies for other researchers.

Author Response File: Author Response.docx