An ontology based framework for mining dependence relationships between news and financial instruments

doi:10.1016/j.eswa.2011.01.148

Expert Systems with Applications

Volume 38, Issue 10, 15 September 2011, Pages 12044-12050

https://doi.org/10.1016/j.eswa.2011.01.148 Get rights and content

Abstract

Before news is input into financial trading algorithms/models, it needs human judgements for exploring the market implications of news content, distinguishing significance extent of news, and finding out the impact of polar type of each kind of news on certain financial instrument trading activity. But Dawes and Faust (1989) reported that people usually rely on clinical judgements, especially it is hard for them to distinguish valid decision variables from invalid ones in decision making. Thus, in order to alleviate this problem and provide more objective decision making support about news in financial market, an ontology based framework is proposed, for investigating the actuarial dependence relationships between news and financial instruments trading activities as well as identifying more valid news for trading decision making. This framework is expected to help people in financial market how to consider weight for each kind of news when inputted in trading algorithms/models of certain financial instruments.

Research highlights

► The main contribution is to provide an ontology based data mining framework. Specifically, an ontology is first proposed for processing news, and financial data. Second, a reasoning system is designed to integrate the background knowledge. With this framework, dependence extent between news and financial trading is known.

Introduction

News has become one of the most important inputs for financial trading decisions making. Some equity trading companies and brokerages have begun to treat real-time news as input variables of their trading algorithms/models.¹ But Dawes and Faust (1989) reported that people usually rely on clinical judgements, it is especially hard for them to distinguish valid decision variables from invalid ones in decision making. Dawes and Faust distinguished two approaches in making judgements during decisions: actuarial and clinical. In the actuarial approach, a decision is made on the basis of empirical relations between decision variables and a decision outcome variable. In the clinical approach, decisions are based on mental processes of human judgement. The advantage of actuarial models against human judgements has been replicated in many domains. While making clinical judgements, humans always have difficulty distinguishing valid decision variables from invalid ones. Lack of sufficient and clear information about relationships between decision variables and outcomes always make people lapse into making relatively less rational decisions in financial market; this problem becomes worse when decision makers are confronted with large amounts of information (Zielonka, 2002). Although financial researchers have verified that some news variables, such as macroeconomic indicators news (Flannery & Protopapadakis, 2002) impact financial market outcomes, there is no rational explanation for choosing such news as better predicators. Zielonka (2002) discussed that financial markets give equal decision weights to different kinds of news variables, though some of them have no quantitatively measurable effect on stock prices. In order to make better trading decisions, Dow Jones based on news² filter news automatically, but its service just tells users the news they are interested in; it could not tell the user how to consider it with trading decision weights. For example, let us consider a scenario where in two pieces of news come in on the same day: “Central Bank of the Republic of China brings down short-term interest rate to 3.25%” and “CPCC (China Petroleum & Chemical Corporation) profit drops 28%, hit by high oil price.” The former will usually be considered as good news for CPCC stock price while the latter will have a negative impact in commonsense knowledge. So which news will dominate the change in stock price trend? If a third piece of news of the Chinese currency being revalued comes on the same day, how will it be treated? How to assign decision weight for each piece of news? Thus it is necessary not only to filter news but also to identify relative importance weight of each kind of news for certain financial instrument trading activity before news is inputted into trading algorithms/models.

In this paper, an ontology based data mining framework is proposed to discover actuarial relationships between various kinds of news and market movements in terms of price trends, volume changes, and so on, that is the dependence of financial market movements on news, as well as the dependence extent order. This framework consists of ontology, expert defined rules base, and Bayesian network algorithm. Ontology is used to organize news and stocks into groups/classes and to represent the relationships among these groups/classes. Based on the ontology, an expert-rules reasoning system is constructed to indicate news possibly significant for a given financial instrument trading activity. The reasoning output of the expert system is used to construct a Bayesian network that specifies the dependence relationships between the stocks and possibly significant news. With the support of the Bayesian network algorithm, people can easily provide/know the significance order of each kind of news on certain financial instrument trading activity. The scenario’ questions above can be answered.

The main contribution of this research is to provide an ontology based data mining framework. Specifically, it provides: 1. an ontology method for processing news data (into classes of events), and financial instrument trading data regarding trading; 2. an expert reasoning system is designed to integrate the background (domain) knowledge in the data mining process through building Bayesian network model consisting of multi news variables with certain financial instrument trading activity and suggesting the polar (“positive”, “neural”, “negative”) type of impact of each news variable, on trading activities; and 3. an evaluation of the framework to identify possible improvements of data mining results is conducted. Based on this framework, people in financial market have actual information about dependence relationships between news and financial instruments, and they are helped to make better judgement and decisions with regard to news.

Section snippets

Related research

The process of investigating the relationships between news and financial instruments is a knowledge discovery process which usually involves three primary steps: data preparation, data mining tasks, and explanation of data mining results (Fayyad, Piatetsky-Shapiro, & Smyth, 1996). Thus, tasks included in this framework contain data preparation, data mining tasks, and explanation of mining results.

Before investigating the quantitative details of relationships between news and financial

An ontology based framework for mining the dependence relationships between news and financial instruments

This section presents an ontology based data mining framework for identifying dependence relationships between news and financial instruments. The framework enables selection of possibly significant news variables for certain financial instrument trading algorithms/models by Bayesian network technique, and incorporates the domain knowledge, which is about collected news data and financial instruments trading data, into the mining process.

The ontology strategy in this research, for processing

Experiment evaluation

A laboratory experiment was carried out to evaluate the effectiveness of the framework. The main goal of this framework is that it can help people to judge the relative impact weight of each kind of news and recommend news’ impact polar on financial instrument trading activities. Compared with past work, there are some innovations in this framework, first is that ontology is adopted to process news into different kinds of events, second, the same kind of event of news’ appearance frequency is

Conclusions

An ontology based framework is proposed for mining the dependence relationships between news events and financial instruments in this paper. The framework is made of three parts. The first is an ontology adopted from Wang’s work (Wang et al., 2008), which classifies news and stocks into different groups, according to the nature of their respective businesses, and the news categories. A class hierarchy is also constructed under each superclass. The second part is an expert-rules reasoning system

Acknowledgements

This research is supported by research grant from Inner Mongolia University (No. 720090114) and research grant from Inner Mongolia Governments (No. 2010BS0901).

References (17)

T. Hong et al.
Knowledge-based data mining of news information on the Internet using cognitive maps and neural networks
Expert Systems with Applications
(2002)
A.P. Sinha et al.
Incorporating domain knowledge into data mining classifiers: An application in indirect lending
Decision Support Systems
(2008)
S. Wang et al.
An ontology for causal relationships between news and financial instruments
Expert Systems with Applications
(2008)
M.L. Wong et al.
Data mining of Bayesian network using cooperative coevolution
Decision Support Systems
(2004)
Y. Cai et al.
Attribute-oriented induction in relational databases
(1991)
R.M. Dawes et al.
Clinical versus actuarial judgement
Science
(1989)
V. Dhan et al.
Abstract driven pattern discovery in databases
IEEE Transactions on Knowledge Discovery and Data Engineering
(1993)
Fayyad, U., Piatetsky-shapiro, G., & Smyth, P. (1996). Knowledge discovery and data mining; toward a unifying...

There are more references available in the full text version of this article.

Cited by (16)

Neural network embeddings on corporate annual filings for portfolio selection
2021, Expert Systems with Applications
Citation Excerpt :
Classification techniques such as Naïve Bayes and SVM have also been exploited by Nizer and Nievola (2012) on news text to predict the volatility of financial assets. Market volatility related to news and exploited with neural networks (NNs) has also been studied extensively by Zopounidis et al. (2010) Wang et al. (2011), on the other hand, proposed an ontology based framework to mine dependence relationships between financial instruments and news. Finally, Lupiani-Ruiz et al. (2011) presents a semantic search engine for financial news using Semantic Web technologies customized on the Spanish stock market.
In recent years, there has been an increased interest from both academics and practitioners in automatically analyzing the textual part of companies’ financial reports to extract meaning rich in information for future outcomes. In particular, tracking textual changes among companies’ reports can have a large and significant impact on stock prices. This impact happens with a lag implying that investors only gradually realize the implications of the news hinted by document changes. However, the length of these documents as well as their complexity in terms of structure and language have been increasing dramatically making this process more and more difficult to perform. In this paper, we analyzed how to face this complexity by learning arbitrary dimensional vector representations for US corporate filings (10-Ks) from 1998 to 2018, exploiting and comparing different neural network embedding techniques which take into account words’ semantics through vectors proximity. We also compared their ability to capture changes associated with future risk-adjusted abnormal returns with other more commonly used approaches in literature. Finally, we propose a novel investment strategy named Semantic Similarity Portfolio (SSP) that exploits these neural network embeddings. We show that firms that do not change their 10-Ks in a semantically important way from the previous year tend to have large and statistically significant future risk-adjusted abnormal returns. We, also document an amplifying effect when we incorporate a momentum-related criterion, where the companies selected must also have had positive previous year returns. Specifically, a portfolio that buys “non-changers” based on this strategy earns up to 10% in yearly risk-adjusted abnormal returns (alpha).
An ontological artifact for classifying social media: Text mining analysis for financial data
2020, International Journal of Accounting Information Systems
Citation Excerpt :
Their method automates ontology creation and involves steps such as preprocessing, obtaining term frequency counts, using concept frequency counts, development of the ontology using the concepts with the highest discriminatory score, scoring multi-word phrases, and finally converting the top scoring terms and concepts to a vector of values to use in subsequent analyses such as prediction and regression. Wang et al. (2011) propose a framework utilizing the ontology from Wang et al. (2008), an expert rules-based system, Bayesian Network models, and Bayesian algorithms to find the relationship between news articles and reported financial liabilities. All of these studies apply ontologies to more formalized textual sources such as reports, news headlines and articles, and sections of financial reports.
In this paper we utilize a structured natural language processing implementation of the Financial Industry Business Ontology (FIBO) to extract financial information from the unstructured textual data of the social media platform Twitter regarding financial and budget information in the public sector, namely the two public-private agencies of the Port Authority of NY and NJ (PANYNJ), and the NY Metropolitan Transportation Agency (MTA). This research initiative uses the Design Science Research (DSR) perspective to develop an artifact to classify tweets as being either relevant to financial bonds or not. We apply a frame and slot approach from the artificial intelligence and natural language processing literature to operationalize this artifact. FIBO provides standards for defining the facts, terms, and relationships associated with financial concepts. We show that FIBO grammar can be used to mine semantic meaning from unstructured textual data and that it provides a nuanced representation of structured financial data. With this artifact, social media such as Twitter may be accessed for the knowledge that its text contains about financial concepts using the FIBO ontology. This process is anticipated to be of interest to bond issuers, regulators, analysts, investors, and academics. It may also be extended towards other financial domains such as securities, derivatives, commodities, and banking that relate to FIBO ontologies, as well as more generally to develop a structured knowledge representation of unstructured data through the application of an ontology.
Toward a semantic-based location tagging news feed system: Constructing a conceptual hierarchy on geographical hashtags
2019, Computers and Electrical Engineering
Citation Excerpt :
Geographic locations have also affected on the news systems with or without the use of geographical hashtags [17–20]. Semantic web-based approaches, such as ontologies, in combination with machine learning methods have been used to improve the efficiency of news systems [21–23]. Semantic relation and similarity [24,25], event detection [26], different tagging and geo-tagging methods [27], and several news personalization and recommendation systems [28,29] are other related topics of the subject of this paper.
Online news and social networking sites have been significantly used in recent years. There has been a lot of efforts to provide appropriate contents for the end users; however, they proved not to be effective. We believe the semantic web (as the third generation of the web) is mature enough to undertake the responsibility of generating more user-centric content. One way to exploit the semantic web's capabilities for such purpose is to construct an ontology that establishes the relationships between hashtags. In this paper, we present the construction process of a news feed system based on the hierarchical relationships of geographic hashtag. Our experiments demonstrated that our proposed semantic-based hierarchical location tagging news feed system increases the quality of the user experience as well as the publication rate of the news while boosting the content match rate to the target audiences. The proposed system of this paper can be considered as a real step towards the realization of semantic web.
A survey of the applications of text mining in financial domain
2016, Knowledge-Based Systems
Citation Excerpt :
They analyzed the results of Monte Carlo simulation. Wang et al. [131] proposed a framework for finding the correlation between news articles and financial liabilities based on Ontology. They preprocessed the news articles and financial liabilities using Ontology.
Text mining has found a variety of applications in diverse domains. Of late, prolific work is reported in using text mining techniques to solve problems in financial domain. The objective of this paper is to provide a state-of-the-art survey of various applications of Text mining to finance. These applications are categorized broadly into FOREX rate prediction, stock market prediction, customer relationship management (CRM) and cyber security. Since finance is a service industry, these problems are paramount in operational and customer growth aspects. We reviewed 89 research papers that appeared during the period 2000–2016, highlighted some of the issues, gaps, key challenges in this area and proposed some future research directions. Finally, this review can be extremely useful to budding researchers in this area, as many open problems are highlighted.
Why do banks fail? An investigation via text mining
2023, Cogent Economics and Finance
Evaluating the effectiveness of Farsi-English query production using ontology: a case of scientometric ontology
2021, Aslib Journal of Information Management

View all citing articles on Scopus

View full text

An ontology based framework for mining dependence relationships between news and financial instruments

Abstract

Research highlights

Introduction

Section snippets

Related research

An ontology based framework for mining the dependence relationships between news and financial instruments

Experiment evaluation

Conclusions

Acknowledgements

Expert Systems with Applications

Decision Support Systems

Expert Systems with Applications

Decision Support Systems

Attribute-oriented induction in relational databases

Clinical versus actuarial judgement

Science

Abstract driven pattern discovery in databases

IEEE Transactions on Knowledge Discovery and Data Engineering