Content uploaded by Mahsa Fischer
Author content
All content in this area was uploaded by Mahsa Fischer on Jan 10, 2025
Content may be subject to copyright.
Machine Learning for forecasting entrepreneurial
opportunities – A literature review
Daniel Szafarski[0009−0008−8992−5152] and Mahsa Fischer[0009−0000−6004−6789]
Heilbronn University of Applied Sciences, Heilbronn, Germany
{daniel.szafarski,mahsa.fischer}@hs-heilbronn.de
Abstract. Business opportunity is one of the key pillars of entrepreneur-
ship research. So far, there is little research on the combination of ar-
tificial intelligence and entrepreneurship. This article makes a twofold
contribution to this through a systematic literature analysis. On the
one hand, the existing approaches for generating or identifying the busi-
ness opportunity are compiled from 25 publications and presented in a
concept matrix. On the other hand, promising approaches and current
research focuses, such as the analysis of patents with the help of text
mining methods and Latent Dirichlet Allocation (LDA), are extracted,
which provides further research incentives.
Keywords: Machine Learning ·Business Opportunities ·Innovation ·
Entrepreneurship ·Success Prediction ·Text Mining ·Literature Review.
1 Introduction
The current economic situation is characterized by rapid change, dynamic in-
novation and uncertainty. True innovation and long-term market success often
lies in the creation or identification of new means that haven’t been recognized
by market participants previously [24, 4, 10]. Therefore, the discovery and de-
velopment of opportunities has been a central component of entrepreneurship
research for many years but is now more relevant than ever before [49, 43]. Cur-
rent studies show that numerous publications have already looked at it from a
variety of perspectives [43, 14, 31]. The technological possibilities offer new op-
portunities in the application of existing concepts as well as the need for new
theories in this context [35]. Extracting relevant information at the right time
can have a major impact on business success. It seems obvious to use machine
learning (ML) for these tasks, as it has already demonstrated its enormous po-
tential in numerous use cases e.g. innovation management [46]. For this reason,
Giuggioli and Pellegrini [12] also describe a need for further research in the area
of entrepreneurial opportunity. This demand can also be supported by the pub-
lication by Obschonka and Audretsch [36], according to which the influence of
artificial intelligence in the research field of entrepreneurship has hardly been
scientifically investigated so far. This motivates the following article to take a
closer look at the following research question: "How are ML methods currently
used to identify entrepreneurial opportunities in science?".
2 Szafarski and Fischer
2 Theoretical background
2.1 Entrepreneurship
As mentioned in the introduction, the field of entrepreneurship is viewed from
different perspectives and is an extremely popular and diverse field of research.
For this reason, there is as yet no clear, generally accepted definition of the term
[45, 4]. Newer definitions in particular view entrepreneurship as a multi-stage
process in which individuals identify and evaluate an entrepreneurial opportunity
and then exploit it in the form of a business start-up or innovation [42, 48, 8].
New innovative entrepreneurship means observing the market closely, thinking
in a networked and lateral way, questioning existing products and developing
new solutions that satisfy customer needs and thus create new value [3, 2].
2.2 Entrepreneurial Opportunity
An opportunity itself can refer to the creation and introduction of an innova-
tive product or the founding of a new company [8, 3]. According to Shane and
Venkataraman [42], an entrepreneurial opportunity exists if a product can be
sold on the market at a higher price than its resources and processing costs.
At the same time, a business idea itself as well as its further development and
modification [5, 7] and the arise from technological possibilities [41] can be un-
derstood as an entrepreneurial opportunity. While an idea is a creative result,
only its examination and development leads to a corresponding opportunity [7,
15]. Research associates the term primarily with the areas of creation, discovery
and recognition of further processes and concepts [43]. There are two basic per-
spectives [1]. One view is based on the discovery of an existing opportunity in
the market, whereas the other is based on the systematic creation of new oppor-
tunities through a creative process [2]. The search is usually understood as an
active process that is stimulated by previous experience, market and customer
knowledge and access to relevant information, among other things [2, 3].
2.3 Machine Learning
Depending on the method used, the terms deep learning, machine learning or ar-
tificial intelligence (AI) are used. AI can be understood as a term for advanced,
intelligent computer systems. This term has become a buzzword in science and
practice, which has led to the synonymous use of AI and ML [11,13]. ML is
a mathematical algorithm that has the ability to recognize and independently
learn complex patterns in large data sets and therefore solve different problems
and situations. Depending on the data used, the task can be specified more
precisely. Traditionally, supervised and unsupervised learning are differentiated,
although there are now also more advanced forms such as reinforcement learn-
ing and hybrid forms such as self-supervised learning [34, 16]. One of the key
success factors is the development of artificial neural networks (ANN) based on
the Multilayer Perceptron towards deeper ANNs, which have improved learning
capabilities and are summarized under the term deep learning [34, 25, 16].
Machine Learning for forecasting entrepreneurial opportunities 3
3 Methodology
This work is based on the research design according to Webster and Watson
[52] as well as Fettke et al.[9] in order to present the current state of the art.
Public availability and the German or English language were used as selection
criteria and conducted using a search string. For this purpose, the six literature
databases ACM Digital Library, AIS Electronic Library, IEEE Xplore Digital Li-
brary, SpringerLink, Google Scholar and Science Direct were examined. In total,
420 potentially relevant publications were identified. Based on their abstracts, a
forward and backward search for further relevant publications and the removal of
duplicates, a final corpus of 25 publications remained for this analysis. The pub-
lications were then analyzed further using qualitative content analysis according
to Mayring [32]. In order to answer the research question the dimensions shown
in table 1 were applied according to a previously created coding guide.
4 Results
4.1 Types and causes of entrepreneurial opportunities
In 64% of the publications examined, the opportunity is considered in connection
with the further development and modification of an existing product portfolio
in established companies. The existing product range and corresponding patents
and property rights are used as a starting point for the conceptualization of
new ideas. Examples span various industries, including chinese elderly care [30],
mobile payments [37], and the turkish textile industry [6], with specific cases
like Thermo Fisher Scientific [33] and the Samsung Galaxy Note 5 [18]. Only
three studies focus on entrepreneurship, all related to further development. Ad-
ditionally, 44% of studies explore opportunities in new technologies, with a few
considering their application in established firms.
Different concepts are recognizable in the publications with regard to the
cause of an entrepreneurial opportunity. In close to half publications (40%), the
opportunity or idea is gained on the basis of new technological progress. This is
followed by the generation of new ideas through the identification of gaps and the
combination of two concepts in 28% of cases. The combination of two concepts
with technological progress can be identified three times, which demonstrates the
adaptation of existing, successful trends to other use cases. Market dynamics and
internal capabilities account for 24% and 20% of opportunities, respectively, with
16% adapting existing patterns to new situations. Remarkably, only one study
focuses on identifying new customer needs for idea generation.
4.2 Data basis and support by ML
Patents are by far the most important (64%) basis for generating new ideas.
However, different aspects of patents are used depending on the publication.
In the sample, company or product-specific information was explicitly used six
4 Szafarski and Fischer
Table 1. Concept matrix for mapping the current state of the art
Machine Learning for forecasting entrepreneurial opportunities 5
times (24%). Social media data and trademarks were used in three cases each.
Project data was used twice. In addition, six other publications use data sources
that are only used in this one case. These include data on design property rights
and scientific publications, which are semitically similar to patents, as well as
advertising, news and external platforms such as Bloomberg.
Regarding the degree of support, 64% of the studies focus on preparation,
while 36% are aimed at proposal, indicating most don’t offer explicit recom-
mendations. Instead, they process the identified data sources to aid end users
in spotting entrepreneurial opportunities, either by classifying patents based on
relevance and novelty or through visual graphics.
4.3 ML methods for identifying opportunities
A total of 28 different methods were differentiated as part of the analysis. At
the same time, almost half (48%) refer to Text Mining and five contributions
(20%) use link prediction as a supercategory, which can be realized on the ba-
sis of several concrete methods. Specifically, 28% of the publications use Latent
Dirichlet Allocation (LDA), a method for modeling topics that makes it possi-
ble to identify the distribution and affiliation of words to topic clusters from a
text corpus. Some publications apply LDA for topic and keyword identification,
grouping related patents [19, 23, 54]. The Vector Space Model (VSM), semantic
similarity and similarity measurement are applied five times (32%). For the re-
verse use case of identifying data points that do not fit into an existing graph, the
Angle-based Outlier Detection, the Local Outlier Factor (LOF) method and the
Structural Hole Theory are used in three cases. Support Vector Machine (SVM),
logistic regression and Random Forest are each used twice, with Park and Geum
[37] comparing their effectiveness. Additional methods from diverse application
contexts were noted but not detailed due to the reason of scope.
4.4 Evaluation and Performance of the ML-techniques
A general comparison of performance between the publications examined is not
possible due to their individuality, e.g. because of different data sources. The
majority (64%) develop a method that is applied to a specific use case. Here,
the functionality is usually checked by the plausibility of the concrete results in
the specific application case which, however, does not allow any statement to be
made about the general quality. In turn, 9 publications (36%) explicitly measure
and evaluate performance. For this purpose, widely used quantitative metrics in
the field of ML such as Accuracy, Error Rate, Precision and Recall are used. As
an example, Jin et al. [20], compared with a manual approach, and achieved an
accuracy of 80.64% , although a low recall value of 48.65%. This indicates that the
approach is only poorly able to identify all connections of the manual process.
However, the high precision score of 62.07% is also worth mentioning, which
suggests that the approach is well able to identify connections that remained
undetected by the experts. Further authors compare their performance with
other methods in order to compare the performance of the new approach.
6 Szafarski and Fischer
4.5 Challenges of the publications and the ML methods
Six main key challenges were identified. Out of the 25 publications examined,
20 address several challenges, while the remaining 5 do not mention any and
will not be discussed further. On average, one publication addresses challenges
in more than half of the identified areas (3.6). In 13 publications, the main chal-
lenges lie in the focus area Research. These include, above all, the weaknesses of
research with regard to validity and generalizability, most of which are listed as
limitations. Challenges with the concept are revealed equally frequently. Here,
the authors reveal concept gaps and weaknesses in both usability and results.
Challenges with the data basis are also pointed out 13 times. These publica-
tions address various topics, including data quality, domain-specific factors, and
thresholds for narrowing down relevant data points. Additionally, the text dis-
cusses the time delay between discovery and patent granting. The hindered us-
ability and generalization of these tools pose a difficulty for their intended usage.
Eleven citations point to problems with operationalization, since, for example,
many cases require initial input by experts as well as the fact that not all con-
nections are found by ML or possible to scale. Eight publications also describe
challenges in connection with the technical implementation of ML methods while
Seven see further challenges in the area of the domain.
5 Conclusion
In summary, the systematic analysis of 25 publications made it possible to iden-
tify the state of the art in the identification of entrepreneurial opportunities
with the help of ML. The main focus of current research is on the analysis of
patents in the context of identifying technological opportunities. At the same
time, entrepreneurial opportunities in the context of entrepreneurship are rarely
explicitly promoted. A closer look at the data basis revealed a focus on natu-
ral language texts, and their processing using NLP methods. Reference is often
made to the general methods of text mining, whereby the methods of LDA and
semantic similarity are widely used in current research. For further identifica-
tion of entrepreneurial opportunity, relevant entities are extracted from the data
sources and linked in a graph structure. With the help of advanced, diverse net-
work analysis algorithms, opportunities are extracted on this basis, primarily
on the basis of technological progress. Challenges with a focus on data sources,
concept and operationalization have been identified, which question the holistic
nature and generalizability of the approaches. Nevertheless, the isolated per-
formance metrics of the publications show the effectiveness of the approaches,
which legitimizes further interest and hope. However, the findings presented have
limitations, include incomplete coverage and variability in methodological detail,
complicating direct comparisons. Future research should validate these findings,
explore universal models or novel approaches like Large Language Models (LLM)
or generative AI, and address the operational challenges identified.
Machine Learning for forecasting entrepreneurial opportunities 7
References
1. Alvarez, S.A., Barney, J.B.: Discovery and creation: alternative theories of en-
trepreneurial action. Strategic Entrepreneurship Journal 1(1-2), 11–26 (nov 2007)
2. Ardichvili, A., Cardozo, R., Ray, S.: A theory of entrepreneurial opportunity iden-
tification and development. Journal of Business Venturing 18(1), 105–123 (2003)
3. Baron, R.A.: Opportunity Recognition as Pattern Recognition: How Entrepreneurs
"Connect the Dots" to Identify New Business Opportunities. Academy of Manage-
ment Perspectives 20(1), 104–119 (2006)
4. Baumol, W.J.: Formal entrepreneurship theory in economics: Existence and
bounds. Journal of Business Venturing 8(3), 197–210 (1993)
5. Davidsson, P.: Entrepreneurial opportunities and the entrepreneurship nexus: A
re-conceptualization. Journal of Business Venturing 30(5), 674–695 (2015)
6. Dereli, T., Durmuşoˇglu, A.: Classifying technology patents to identify trends: Ap-
plying a fuzzy-based clustering approach in the Turkish textile industry. Technol-
ogy in Society 31(3), 263–272 (2009)
7. Dimov, D.: Beyond the Single-Person, Single-Insight Attribution in Understanding
Entrepreneurial Opportunities. Entrepreneurship Theory and Practice 31(5), 713–
731 (sep 2007)
8. Eckhardt, J.T., Shane, S.A.: Opportunities and entrepreneurship. Journal of Man-
agement 29(3), 333–349 (2003)
9. Fettke, P.: State-of-the-art des state-of-the-art: Eine untersuchung der
Forschungsmethode "review" innerhalb der Wirtschaftsinformatik. Wirtschaftsin-
formatik 4(48), 257–266 (2006)
10. Gaglio, C.M., Katz, J.A.: The Psychological Basis of Opportunity Identification:
Entrepreneurial Alertness. mall Business Economics 16, 95–111 (2001)
11. Gbadegeshin, S.A., Al Natsheh, A., Ghafel, K., Tikkanen, J., Gray, A., Rimpiläi-
nen, A., Kuoppala, A., Kalermo-Poranen, J., Hirvonen, N.: What Is an Artificial
Intelligence (Ai): a Simple Buzzword or a Worthwhile Inevitability? ICERI2021
Proceedings 1(November), 468–479 (2021)
12. Giuggioli, G., Pellegrini, M.M.: Artificial intelligence as an enabler for en-
trepreneurs: a systematic literature review and an agenda for future research. In-
ternational Journal of Entrepreneurial Behavior Research (mar 2022)
13. Grewal, D.S.: A Critical Conceptual Analysis of Definitions of Artificial Intelligence
as Applicable to Computer Engineering. IOSR Journal of Computer Engineering
16(2), 09–13 (2014)
14. Hansen, D.J., Monllor, J., Shrader, R.C.: Identifying the elements of en-
trepreneurial opportunity constructs: Recognizing what scholars are really exam-
ining. International Journal of Entrepreneurship and Innovation 17(4), 240–255
(2016)
15. Hsieh, C., Nickerson, J.A., Zenger, T.R.: Opportunity discovery, problem solving
and a theory of the entrepreneurial firm. Journal of Management Studies 44(7),
1255–1277 (2007)
16. Janiesch, C., Zschech, P., Heinrich, K.: Machine learning and deep learning. Elec-
tronic Markets 31(3), 685–695 (sep 2021)
17. Jeong, B., Ko, N., Son, C., Yoon, J.: Trademark-based framework to uncover busi-
ness diversification opportunities: Application of deep link prediction and compet-
itive intelligence analysis. Computers in Industry 124, 103356 (2021)
18. Jeong, B., Yoon, J., Lee, J.M.: Social media mining for product planning: A product
opportunity mining approach based on topic modeling and sentiment analysis.
8 Szafarski and Fischer
International Journal of Information Management 48(September 2017), 280–290
(2019)
19. Jeong, Y., Park, I., Yoon, B.: Identifying emerging Research and Business Devel-
opment (RBD) areas based on topic modeling and visualization with intellectual
property right data. Technological Forecasting and Social Change 146(May 2018),
655–672 (2019)
20. Jin, G., Jeong, Y., Yoon, B.: Technology-driven roadmaps for identifying new prod-
uct/market opportunities: Use of text mining and quality function deployment.
Advanced Engineering Informatics 29(1), 126–138 (2015)
21. Kim, B., Gazzola, G., Lee, J.M., Kim, D., Kim, K., Jeong, M.K.: Inter-cluster
connectivity analysis for technology opportunity discovery. Scientometrics 98(3),
1811–1825 (2014)
22. Kim, J.E., Choi, Y.J.: A Methodology of Predicting Market Convergence Oppor-
tunity Using Machine Learning for Small and Mid-Size Enterprises 142(Dta), 146–
149 (2016)
23. Kim, M., Park, Y., Yoon, J.: Generating patent development maps for technol-
ogy monitoring using semantic patent-topic analysis. Computers and Industrial
Engineering 98, 289–299 (2016)
24. Kirzner, I.M.: Entrepreneurial Discovery and the Competitive Market Process: An
Austrian Approach. Journal of Economic Literature 35(1), 60–85 (1997)
25. Lecun, Y., Bengio, Y., Hinton, G.: Deep learning. Nature 521(7553), 436–444
(2015)
26. Lee, C., Lee, G.: Technology opportunity analysis based on recombinant search:
patent landscape analysis for idea generation. Scientometrics 121(2), 603–632
(2019)
27. Lee, J., Ko, N., Yoon, J., Son, C.: An approach for discovering firm-specific technol-
ogy opportunities: Application of link prediction to F-term networks. Technological
Forecasting and Social Change 168(July 2020), 120746 (2021)
28. Lee, M., Lee, S.: Identifying new business opportunities from competitor intelli-
gence: An integrated use of patent and trademark databases. Technological Fore-
casting and Social Change 119, 170–183 (2017)
29. Lee, O.J., Park, S.Y., Kim, J.T.: IdeaNet: Potential opportunity discovery for
business innovation. CEUR Workshop Proceedings 2794, 5–8 (2020)
30. Li, C.: Market Opportunity and Policy Support for Chinese Old Aging Industry:
An Application of Text Mining. SHS Web of Conferences 39, 01015 (2017)
31. Mary George, N., Parida, V., Lahti, T., Wincent, J.: A systematic literature re-
view of entrepreneurial opportunity recognition: insights on influencing factors.
International Entrepreneurship and Management Journal 12(2), 309–350 (2016)
32. Mayring, P., Fenzl, T.: Qualitative inhaltsanalyse. In: Baur, N., Blasius, J.H.
(eds.) Handbuch Methoden der empirischen Sozialforschung, pp. 543–556. Springer,
Wiesbaden (2014)
33. Mun, C., Kim, Y., Yoo, D., Yoon, S., Hyun, H., Raghavan, N., Park, H.: Dis-
covering business diversification opportunities using patent information and open
innovation cases. Technological Forecasting and Social Change 139(May 2018),
144–154 (2019)
34. Murphy, K.P.: Machine learning: a probabilistic perspective. MIT press (2012)
35. Nambisan, S.: Digital Entrepreneurship: Toward a Digital Technology Perspec-
tive of Entrepreneurship. Entrepreneurship: Theory and Practice 41(6), 1029–1055
(2017)
36. Obschonka, M., Audretsch, D.B.: Artificial intelligence and big data in en-
trepreneurship. Small Business Economics 55(3), 529–539 (2020)
Machine Learning for forecasting entrepreneurial opportunities 9
37. Park, M., Geum, Y.: Two-stage technology opportunity discovery for firm-level
decision making: GCN-based link-prediction approach. Technological Forecasting
and Social Change 183(July), 121934 (2022)
38. Pournemat, M., Weiss, M.: Identifying business opportunities using topic modeling
and chance discovery. In: The ISPIM Innovation Conference – Innovating Our
Common Future. No. June (2021)
39. Russo, D., Spreafico, M., Precorvi, A.: Discovering new business opportunities with
dependent semantic parsers. Computers in Industry 123, 103330 (2020)
40. Seo, W., Yoon, J., Park, H., youl Coh, B., Lee, J.M., Kwon, O.J.: Product opportu-
nity identification based on internal capabilities using text mining and association
rule mining. Technological Forecasting and Social Change 105, 94–104 (2016)
41. Shane, S.: Technological opportunities and new firm creation. Management Science
47(2), 205–220 (2001)
42. Shane, S., Venkataraman, S.: The Promise of Entrepreneurship as a field of reseanc.
Academy ol Managemenl fleview 25(1), 217–226 (2000)
43. Short, J.C., Ketchen, D.J., Shook, C.L., Ireland, R.D.: The concept of "Opportu-
nity" in entrepreneurship research: Past accomplishments and future challenges.
Journal of Management 36(1), 40–65 (2010)
44. Song, K., Kim, K.S., Lee, S.: Discovering new technology opportunities based on
patents: Text-mining and F-term analysis. Technovation 60-61(January), 1–14
(2017)
45. Stevenson, H.H., Jarillo, J.C.: A Paradigm of Entrepreneurship: Entrepreneurial
Management. Strategic Management Journal 11(Corporate Entrepreneurship
(Summer, 1990)), 17–27 (1990)
46. Szafarski, D., Beckmann, H.: Einsatz von Machine Learning im Innovationsman-
agement. Angewandte Forschung in der Wirtschaftsinformatik 2022 pp. 273–286
(2022)
47. Talbot, B.G., Talbot, L.M.: Business Radar : Opportunity Analysis and Metric
Estimation Using a Fuzzy Grade-of-Membership Model. Technology Review (June
2014), 87–112 (2003)
48. Timmons, J., Spinelli, S., Prescott, E.: New Venture Creation: Entrepreneurship
for the 21st Century (2010)
49. Venkataraman, S.: The Distinctive Domain of Entrepreneurship Research. Ad-
vances in Entrepreneurship, Firm Emergence and Growth 3(January) (1997)
50. Wang, J., Chen, Y.J.: A novelty detection patent mining approach for analyzing
technological opportunities. Advanced Engineering Informatics 42(June), 100941
(2019)
51. Wang, M.Y., Fang, S.C., Chang, Y.H.: Exploring technological opportunities by
mining the gaps between science and technology: Microalgal biofuels. Technological
Forecasting and Social Change 92, 182–195 (2015)
52. Webster, J., Watson, R.T.: Analyzing the Past To Prepare for the Future : Writing
a Literature Review. MIS Quarterly 2(26), 13–23 (2002)
53. Yoon, B., Magee, C.L.: Exploring technology opportunities by visualizing patent
information based on generative topographic mapping and link prediction. Tech-
nological Forecasting and Social Change 132(July 2017), 105–117 (2018)
54. Yoon, J., Seo, W., Coh, B.Y., Song, I., Lee, J.M.: Identifying product opportuni-
ties using collaborative filtering-based patent analysis. Computers and Industrial
Engineering 107, 376–387 (2017)
55. Zhang, C., Wang, H., Xu, F., Hu, X.: IdeaGraph plus: A topic-based algorithm for
perceiving unnoticed events. Proceedings - IEEE 13th International Conference on
Data Mining Workshops, ICDMW 2013 pp. 735–741 (2013)
