https://orcid.org/0000-0002-8050-123X
ABSTRACT
Business analytics and big data have been at the center of interest for researchers and practitioners for almost a decade now. The methods and processes that comprise business analytics, combined with the rich information that can be extracted from big data have enabled organizations to generate rich insight which is critical to decision making. The scientific inquiry in this interdisciplinary domain has had a long and successful history at the European Conference on Information Systems (ECIS). We provide a synthesis of prominent themes that have appeared during the past decade within the “Business Analytics and Big Data” track of ECIS. Based on the synthesis, we provide a narrative of how the field has evolved, as well as where we see future research efforts being focused. Specifically, we identify three areas that are likely to attract considerable research interest in the years to come. Within each of these three areas, we describe several key challenges that need to be addressed. We conclude with an overview of the six articles included in this special issue, and a description of how they contribute to our understanding of this domain.
1. Past, Present, and Future of Business Analytics and Big Data Research Seen Through the Lens of the European Conference on Information Systems
Business analytics summarises all methods, processes, technologies, applications, skills, and organisational structures necessary to analyse past or current data to manage and plan business performance. While in the past, business intelligence was rather focused on data integration and reporting descriptive analytics, business analytics is inherently forward-looking and centres its analyses on diagnostics, prediction, and prescription tasks (Hindle & Vidgen, Citation2018; Wanner et al., Citation2022). Big data is a natural companion as only large and possibly rich datasets enable comprehensive decision models to assist these tasks. The rise of big data over the past few years has naturally created new opportunities as well as challenges to the domain of business analytics. Thus, while there are ample opportunities for analysing big data to generate important insight and gain a competitive edge, there is also a renewed interest in the field of how organisations and individuals should adapt to reap the most out of big data.
The “Business Analytics and Big Data” track as a melting pot for topics in information systems (IS) and neighbouring disciplines has a long and successful history at the European Conference on Information Systems (ECIS). From its initial year in 2012 to 2021, the track has received 512 submissions. While the track invited research about business intelligence as well as about knowledge management in early years, both topics were later addressed in two separate tracks. In addition, in some years there have been other topic-related tracks at ECIS, such as about big data or artificial intelligence (AI). Nevertheless, the track has constituted over time the primary and steady nucleus for research in the fields of business intelligence and analytics, big data, and data science in IS research. The transformation from “Business Intelligence and Knowledge Management” in 2012 to “Business Analytics and Big Data” in recent years indicates the shift in focus towards data-rich applications and intelligent systems.
This shift can also be seen from the submissions to the track. To analyse the papers submitted in this track over the years, we used our track database and analysed the keywords in all papers to the different track iterations that were available to us with text mining. Marjanovic and Dinter (Citation2017, Citation2018) have conducted a similar in-depth analysis of 25 years of research published at the Hawaii International Conference on System Sciences.
Looking at the data, we can observe more detailed trends that support and illustrate this longitudinal change in topics. summarises notable top keywords across the different years as well as summarised across all years. We extracted them, where keywords were available, and preprocessed them by text unifying, stemming, NLP stopwords, and lastly manual alignment of terms.
Table 1. Track Topics from 2012 to 2021
While the early years of the track indeed comprised more classic “business intelligence” and “knowledge management” research, the shift towards the keyword “business analytics” is not only a mere shift in terminology. While early topics of the track centred around data warehousing, customer relationship management, and handcrafted decision support systems, there was always room for data mining or data science research, for example, with respect to text mining. However, around 2015 the focus moved to advanced industrial applications of predictive analytics and smart objects, which led to a focus on the analysis of large amounts of data, so-called big data, for various applications from industrial maintenance and advertising to social network analysis and process mining. In addition, the topics of self-service business intelligence, cloud analytics, and big data management as well as open data emerged. This is also evident through the diversity of topics in those years, which made the isolation of focal topics difficult (see also footnote in Table above). This shift can be largely attributed to the broad use of Internet-of-Things (IoT) devices and sensors, an increase in processing capabilities, the prevalence of cloud-based services, as well as the explosion of data primarily through social media and smart devices (Mikalef et al., Citation2020).
In recent years, we see a shift from dealing with big data to performing intelligent analyses for various financial and industrial applications, which take big data for granted. These analyses often rely on machine learning algorithms that learn from data rather than being programmed to do so. This provides a rich and engaging future research avenue as the integration of intelligent decision support systems in human work routines generates new challenges for business analytics research. In other words, many tasks that were previously manual decision-making processes, are automated through novel approaches that build on data and automate decision-making.
A further Latent Dirichlet Allocation of the keywords confirms this picture and highlights that data warehousing, knowledge management, and business intelligence have become established knowledge rather than foci of research around 2015, when IS research shifted towards analysing big data with statistical methods for predictive analytics. We see the same shift happening now towards machine learning and AI, where these topics become more than a tool to sophisticatedly analyse data, but a research domain and problem class of its own standing. In turn, these new frontiers open up a whole new set of research questions which need to be addressed, including understanding how to embed algorithms with tacit knowledge, understanding the shifts that AI-based applications introduce to the work patterns and routines of employees, as well as examining the changes they introduce in how organisations compete.
While the topics in some track years may be biased by the addition or removal of suggested topics in the call for papers, changes in co-chairs, or alterations of the track title, they also reflect gradual changes in the community and, hence, still – or rather therefore – paint an accurate and illustrative picture of the last decade of business analytics research. In doing so, it generally confirms the observations of and Wanner et al. (Citation2022) and Marjanovic and Dinter (Citation2017, Citation2018) and extends the latter’s work with the most recent shift in topics towards AI in business analytics. Thus, the analysis confirms the evolution of the field towards themes that feature more advanced approaches in analysing large, and often, complex datasets. Such approaches are now featured as core components of how organisations position themselves in the competitive business arena.
2. Perspectives and opportunities for business analytics and big data research in information systems research
This constant change gives rise to a number of topics that we see as prevalent for the coming years. While this list of topics is by no means exhaustive, it does provide some insight into upcoming and interesting themes that are likely to engage IS researchers in the next few years. They are the natural extension of current trends that we observe:
Artificial Intelligence for Business Analytics: With the availability of ever more (big) data, the analytical models necessary to analyse the data and make decisions become more complex to engineer and more often rely on advanced machine and deep learning algorithms. On the upside, the models enable human-like decision-making, one the downside this entails that human, data scientist and end users alike, often do not have transparency in the models and, thus, do not understand the models in their entirety (Janiesch et al., Citation2021). In addition to this, the automation of many previously manual tasks, raises the question of what effects this phenomenon has on employees whose work is changed, as well as to society as a whole (Ågerfalk et al., Citation2022). This gives rise to a number of challenges that need to be addressed for business analytics applications.
External validity of performance gains: The application of machine learning algorithms has become a convenience that enables the generation of impressive results over established methods in almost any field if only selected metrics are reported and single datasets are being used. For any application of intelligent systems to be a meaningful contribution to business analytics research rather than isolated business practice, it is important to consider the generalisability of the results. Otherwise, the progress of the discipline is stalled by only focusing on local optima (Duin, Citation1994; Hutson, Citation2020).
Governance of intelligent systems: The operation and maintenance of systems with human-like capabilities for designated tasks needs to be diligently planned to avoid employee resistance and algorithm aversion (Jussupow et al., Citation2020). This also entails transfer learning and retraining strategies to adjust to changing circumstances (Janiesch et al., Citation2021). Furthermore, governance of AI based on responsible principles entails many different actors at different levels within organisations and throughout them. Hence, understanding the nuances of how responsible AI principles can be translated into implementable practices throughout the lifecycle of projects is an important future research opportunity (Mikalef et al., Citation2022).
Human dignity and intelligent systems: The way (big) data is generated, collected, stored, analysed, and used for decision-making can trigger not only claims but also affronts to human dignity (Leidner & Tona, Citation2021). Decisions based on business systems can affect one’s autonomy, freedom, and values among others. As such, an important future research avenue is how to design business analytics systems for human dignity as well as policies and intervention to prevent intentional or unintentional harm.
Human bias in intelligent system engineering: Machine learning is susceptible to biases and reinforces them rather than diminishing them. Hence, it is important to understand where humans touch the data as well as the implementation process to be aware of and avoid contamination where possible (Van Giffen et al., Citation2022). Furthermore, it is important to question how biases may be contingent upon a number of different contextual factors, therefore creating ambiguity as to what is perceived biased and what not. Hence, a prominent direction of study is to examine how culture conditions our understanding of bias when designing AI applications, and what measures are taken to counter the presence of it.
Explanations of decisions: While any machine learning model is inherently traceable, intelligent system predictions are perceived as black boxes as the underlying models are too complex to comprehend without assistance (Miller, Citation2019). The efficacy of explanation augmentation provided by explainable AI research for the purpose of business analytics needs to be better understood to enable meaningful decision-making and hybrid intelligence (Dellermann et al., Citation2019). Furthermore, apart from enabling an explanation of the inner workings of the “black box” of AI, it is important that studies examine how such explanations should be communicated, at what level of detail information should be provided based on the context, as well as in which form it should be represented and visualised (Enholm et al., Citation2021).
Sustainable business models: One of the core foundations of responsible AI is that any such applications are utilised in a way that does not cause harm to society, individuals, and the environment (European Commission, Citation2019). Hence, we expect that future studies will examine the link between the deployment of AI-based applications and the pursuit of sustainable business models, such as circular economy strategies, or business models that place at the centre the role of corporate and social responsibility (Zhao, Citation2018).
Process Mining for Business Analytics: Sequential data such as events data logs enable first-hand insights into the processes that organisations perform. With the recent rise and beginning consolidation of process mining tool suites in the market, research into processes mining has evolved from a computer-science-fuelled topic of algorithm and software engineering to a domain-science-induced opportunity to analyse the behaviour of individuals, teams, and organisations based on rich process data.
Hyperautomation: This catchword refers to the rapid, scalable, business-driven automation of as many processes as possible. Gartner has named hyperautomation as one of the top strategic technology trends for 2022 (Stoudt-Hansen et al., Citation2021) and, as of now, it is unclear how it can be sustainably orchestrated across operations and measured for business performance (Axmann, Harmoko, Herm & Janiesch, Citation2021).
Process mining methods: Until recently (Evermann et al., Citation2017), process mining has largely relied on statistical methods from the realm of data mining. The applications of common machine learning algorithms for different purposes, such as image or text analysis to event logs and process models in a non-trivial fashion to enable trace clustering or next event prediction are only the forefront of possible applications.
Augmented business process management systems: The amalgamation of business process management, process mining, and AI technology gives rise to augmented process-aware systems that can execute processes more autonomous, conversationally actionable, adaptive, self-improving, and explainable (Dumas et al., Citation2022). Their engineering and contextualisation are to be explored.
Mining user behaviour: Process mining operates on (standardised) event logs from process-aware enterprise systems. This data is being stored through automated tasks on the completion of user tasks. Mining the user behaviour while they perform user tasks or manual tasks can be a great asset to further optimise partly digitalised yet not automated processes, for example, with robotic process automation (Leno et al., Citation2020).
Governance of Open and Linked Data: Big Data does not only enable advanced analyses, but it also is a topic of its own. As data is more and more forming the baseline of decision processes in ubiquitous intelligent systems, it becomes ever more important to enable appropriate data management to allow for traceability and confirmation of results. This includes data to replicate findings but also data that was used to train analytical baseline models that serve as the basis for further advanced and specialised applications through transfer learning.
Open data for business analytics: In science, a trend towards open science data is clearly visible and fuels multiple open data institutions, such as Dryad, or national initiatives such as the German NFDI. Open data in business outside of data competitions is still scarce and incentives are missing despite the obvious cooperative benefits for self-learning systems. Furthermore, there is a large debate regarding data that is collected or generated through public funds, where there are growing voices that such data should be made available for public use.
Governance of data and data analytics: Data storage, wrangling, and examination has evolved greatly in recent years and requires novel structures, policies, and controls to coordinate activities and aligning interests to maximise the value of data analytics in terms of structure, process, and relations (Gröger, Citation2018; Yamada & Peran Citation2017). Further, analyses have to be conducted according to regulations, hence the development of privacy-preserving mechanisms to handle the data becomes pivotal (Mendes & Vilela, Citation2017). In addition, access rights, as well as the role of data as a strategic asset needs to be carefully examined in order to be able to develop comprehensive data governance practices within organisations (Tallon et al., Citation2014).
Entity linking and inference of data: Linking data enables the creation of knowledge graphs for various analytical applications (Abramowicz et al., Citation2016). Tying data to individuals enables powerful and mass-individualised analyses. These applications give rise to new forms of data storage besides relational and NoSQL databases, such as triplestores. The impact of evolving digital infrastructures and linked data on business analytics will likely result in novel opportunities and challenges.
Fittingly, several of these thematic clusters are apparent from the papers that have been accepted for this special issue.
3. Research Included in this special issue
For this special issue of the Journal of Business Analytics, we invited about a dozen papers from the track Business Analytics and Big Data at ECIS 2020. They comprised the best reviewed papers, the most suitable topics given the theme of the track and journal, as well as the most engaging discussion at the virtual conference. The submissions were considerably updated, enhanced, and extended revisions of the respective conference papers. They were placed in a thorough multi-round review process involving old and new reviewers and associate editors. As a result, we were able to accept six papers for this special issue: two focus on the interaction of humans and machine learning algorithms, two focus on issues of process analytics, and two focus on data ownership and governance issues substantiating the three topic areas outlined above.
Palmer et al. (Citation2022) develop a domain-specific sentiment dictionary and use it for financial analyst communication. Their evaluations show that their dictionary outperforms other finance-related dictionaries and classifiers. While it is not superior to a sophisticated, domain-specific machine learning model, their research highlights that sentiment dictionaries have the advantage of intuitive use and high degree of explainability and bring balance to machine learning performance and human intuition.
Wanner et al. (Citation2022) focus on the problem of practically untraceable decision-making by contemporary machine learning algorithms. In a social evaluation of the goodness of explainability, they derive and examine factors that influence end user perception. In a survey, they compare six common machine learning algorithms in four treatments and find that the problem type is a moderator and trustworthiness of the model is the most important factor sustained differently for white-box and black-box models.
Andrews et al. (Citation2022) investigate root-cause analysis of process-data quality problems in process mining. They introduce the Odigos Root Cause reference framework to facilitate an informed way of dealing with data quality issues in event logs. Their framework supports prognostic and diagnostic approaches and detects eleven event, case, and label patterns.
Weinzierl et al. (Citation2022) propose a deep-learning-based method for detecting temporal workarounds in business processes. In their open access design science study, the authors devise a method and implementation that can detect seven different information-system-oriented and process-oriented types of workarounds with high accuracy, precision, and recall. Their research bridges the boundaries between business analytics, organisational routines, and business processes management.
Baijens et al. (Citation2022) establish and theorise data analytics governance. Their open access paper provides a descriptive framework and a viable-system-model-based view as a theoretical lens to discuss how data governance can create (business) value from data. The authors conducted 21 interviews within three case studies with companies that invested in data analytics. Their descriptive data analytics governance framework comprises nine structural, process, and relational sub mechanisms to assist a deeper understanding of the possibilities to govern data analytics in practice.
Fadler and Legner (Citation2022) revisit the topic of data ownership and clarify data accountabilities for big data and analytics. Based on four case studies, they identify ownership principles and three distinct types using within- and cross-case analyses. As a result, they provide several propositions and distinguish data ownership, data platform ownership, and data product ownership with distinct responsibilities as well as point out implications of data repurposing. The paper is open access as well.
These six papers highlight the diversity of topics in the area of big data and business analytics in IS research and contribute novel methodological considerations in terms of governance, ownership and process mining as well as provide novel insights into the developing interaction of intelligent machines and humans.
Acknowledgments
We thank Lukas-Valentin Herm, University of Würzburg, and Alexander Mayr, TU Dortmund University, for implementing and performing the keyword analysis on the ECIS 2012–2021 data. Patrick Mikalef received support in conducting this work from the Slovenian Research Agency (research core funding No. P5-0410).
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
- Abramowicz, W., Auer, S., & Heath, T. (2016). Linked data in business. Business & Information Systems Engineering, 58(5), 323–326. https://doi.org/10.1007/s12599-016-0446-0
- Ågerfalk, P., Conboy, K., Crowston, K., Eriksson Lundström, J., Jarvenpaa, S., Mikalef, P., Ram, S. (2022). Artificial intelligence in information systems: State of the art and research roadmap. Communications of the Association for Information Systems, Forthcoming. http://uu.diva-portal.org/smash/get/diva2:1620171/FULLTEXT01.pdf
- Andrews, R., Emamjome, F., ter Hofstede, A. H. M., & Reijers, H. A. (2022). Root-cause analysis of process-data quality problems. Journal of Business Analytics, 5(1) 51–75. https://doi.org/10.1080/2573234X.2021.1947751
- Axmann, B., Harmoko, H., Herm, L.-V., & Janiesch, C. (2021). A framework of cost drivers for robotic process automation projects. Paper presented at the 19th international conference on business process management (BPM) RPA forum. lecture notes in business information processing, Rome, 428.Springer. https://doi.org/10.1007/978-3-030-85867-4_2.
- Baijens, J., Huygh, T., & Helms, R. (2022). Establishing and theorising data analytics governance: A descriptive framework and a VSM-based view. Journal of Business Analytics, 5(1), 102–123. https://doi.org/10.1080/2573234X.2021.1955021
- Dellermann, D., Ebel, P., Söllner, M., & Leimeister, J. M. (2019). Hybrid Intelligence. Business & Information Systems Engineering, 61(5), 637–643. https://doi.org/10.1007/s12599-019-00595-2
- Duin, R. P. W. (1994). Superlearning and neural network magic. Pattern Recognition Letters, 15(3), 215–217. https://doi.org/10.1016/0167-8655(94)90052-3
- Dumas, M., Fournier, F., Limonad, L., Marrella, A., Montali, M., Rehse, J.-R., Accorsi, R., Calvanese, D., De Giacomo, G., Fahland, D., Gal, A., La Rosa, M., Völzer, H., Weber, I. (2022). Augmented business process management systems: a research manifesto. 2201.12855. arXiv. https://arxiv.org/abs/2201.12855
- Enholm, I. M., Papagiannidis, E., Mikalef, P., & Krogstie, J. (2021). Artificial intelligence and business value: a literature review. Information Systems Frontiers, Forthcoming. https://doi.org/10.1007/s10796-021-10186-w
- European Commission. (2019). Ethics guidelines for trustworthy AI. https://digital-strategy.ec.europa.eu/en/library/ethics-guidelines-trustworthy-ai
- Evermann, J., Rehse, J.-R., & Fettke, P. (2017). Predicting process behaviour using deep learning. Decision Support Systems, 100, 129–140. https://doi.org/10.1016/j.dss.2017.04.003
- Fadler, M., & Legner, C. (2022). Data ownership revisited: Clarifying data accountabilities in times of big data and analytics. Journal of Business Analytics, 5(1), 124–140. https://doi.org/10.1080/2573234X.2021.1945961
- Gröger, C. (2018). Building an industry 4.0 analytics platform. Datenbank-Spektrum, 18(1), 5–14. https://doi.org/10.1007/s13222-018-0273-1
- Hindle, G. A., & Vidgen, R. (2018). Developing a business analytics methodology: A case study in the foodbank sector. European Journal of Operational Research, 268(3), 836–851. https://doi.org/10.1016/j.ejor.2017.06.031
- Hutson, M. (2020). Core progress in AI has stalled in some fields. Science, 368(6494), 927. https://doi.org/10.1126/science.368.6494.927
- Janiesch, C., Zschech, P., & Heinrich, K. (2021). Machine learning and deep learning. Electronic Markets, 31(3), 685–695. https://doi.org/10.1007/s12525-021-00475-2
- Jussupow, E., Benbasat, I., & Heinzl, A. (2020). Why are we averse towards algorithms? Acomprehensive literature review on algorithm aversion. Paper presented at the 28th European conference on information systems (ECIS), AIS virtual conference. AIS.
- Leidner, D. E., & Tona, O. (2021). The CARE theory of dignity amid personal data digitalization. Management Information Systems Quarterly, 45(1), 343–370. https://doi.org/10.25300/MISQ/2021/15941
- Leno, V., Polyvyanyy, A., Dumas, M., La Rosa, M., & Maggi, F. M. (2020). Robotic process mining: vision and challenges. Business & Information Systems Engineering, 63(3), 301–314. https://doi.org/10.1007/s12599-020-00641-4
- Marjanovic, O., & Dinter, B. (2017). 25+ years of business intelligence and analytics minitrack at HICSS: A text mining analysis. Paper presented at the 50th Hawaii international conference on system sciences (HICSS), HI, Big Island.
- Marjanovic, O., & Dinter, B. (2018). Learning from the history of business intelligence and analytics research at HICSS – A semantic text mining approach. Communications of the Association for Information Systems, 43, 775–791. https://doi.org/10.17705/1cais.04340
- Mendes, R., & Vilela, J. P. (2017). Privacy-preserving data mining: methods, metrics, and applications. IEEE Access, 5, 10562–10582. https://doi.org/10.1109/access.2017.2706947
- Mikalef, P., Conboy, K., Lundström, J. E., & Popovič, A. (2022). Thinking responsibly about responsible AI and ‘the dark side’ of AI. European Journal of Information Systems, forthcoming , 1–12. https://doi.org/10.1080/0960085x.2022.2026621
- Mikalef, P., Pappas, I. O., Krogstie, J., & Pavlou, P. A. (2020). Big data and business analytics: A research agenda for realizing business value. Information & Management, 57(1), 103237. https://doi.org/10.1016/j.im.2019.103237
- Miller, T. (2019). Explanation in artificial intelligence: Insights from the social sciences. Artificial Intelligence, 267, 1–38. https://doi.org/10.1016/j.artint.2018.07.007
- Palmer, M., Roeder, J., & Muntermann, J. (2022). Induction of a sentiment dictionary for financial analyst communication: A data-driven approach balancing machine learning and human intuition. Journal of Business Analytics, 5(1), 8–28. https://doi.org/10.1080/2573234X.2021.1955022
- Stoudt-Hansen, S., Karamouzis, F., & Guttridge, K. (2021). Top strategic technology trends for 2022: hyperautomation. Gartner Inc. https://www.gartner.com/en/documents/4006920/top-strategic-technology-trends-for-2022-hyperautomation
- Tallon, P. P., Ramirez, R. V., & Short, J. E. (2014). The information artifact in IT governance: Toward a theory of information governance. Journal of Management Information Systems, 30(3), 141–178. https://doi.org/10.2753/mis0742-1222300306
- van Giffen, B., Herhausen, D., & Fahse, T. (2022). Overcoming the pitfalls and perils of algorithms: A classification of machine learning biases and mitigation methods. Journal of Business Research, 144, 93–106. https://doi.org/10.1016/j.jbusres.2022.01.076
- Wanner, J., Herm, L.-V., Heinrich, K., & Janiesch, C. (2022). A social evaluation of the perceived goodness of explainability in machine learning. Journal of Business Analytics, 5(1), 29–50. https://doi.org/10.1080/2573234X.2021.1952913
- Wanner, J., Wissuchek, C., Welsch, G., & Janiesch, C. (2022). A taxonomy and archetypes of business analytics in smart manufacturing. ACM SIGMIS Database: The DATA BASE for Advances in Information Systems, forthcoming. https://arxiv.org/abs/2110.06124
- Weinzierl, S., Wolf, V., Pauli, T., Beverungen, D., & Matzner, M. (2022). Detecting temporal workarounds in business processes – A deeplearning-based method for analysing event log data. Journal of Business Analytics, 5(1), 76–101. https://doi.org/10.1080/2573234X.2021.1978337
- Yamada, A., & Peran, M. (2017). Governance framework for enterprise analytics and data. Paper presented at the 2017 IEEE International Conference on Big Data, Boston, MA. IEEE.
- Zhao, W.-W. (2018). How to improve corporate social responsibility in the era of artificial intelligence? Paper presented at the 2018 International conference of green buildings and environmental management (GBEM). IOP conference series: Earth and environmental science, Qingdao. IOP.