JIMMA UNIVERSITY JIMMA INSTITUTE

OF TECHNOLOGY FACULTY OF COMPUTING AND

INFORMATICS

Submission Final Thesis Titles

Name: Mohamed Abdiwahab ID: RM/0098

January 2025

Jimma Ethiopia
 Title One

"Sentiment Analysis of Somali-Language Customer Feedback on Fintech

Services Using Advanced Machine Learning Approaches: A Case Study of
Dahabshil Bank, Hargeisa, Somaliland "
ABSTRACT

This Research aims to conduct an extensive analysis of customer feedback on fintech services
offered by Dahabshil Bank in Somaliland, utilizing machine learning approaches. The focal point
of this study is the sentiment analysis of feedback gathered in the Somali language. With the
rapid expansion of fintech, understanding customer satisfaction and feedback has become crucial
for continuous service improvement. By leveraging sophisticated machine learning algorithms,
this research seeks to explore and understand customer sentiments, identify predominant themes,
and uncover key areas for enhancement.

Dahabshil Bank, a prominent financial institution in Somaliland, has been at the forefront of
providing innovative fintech solutions. However, to maintain and advance these services, it is
essential to delve into customer feedback, particularly in the native language of the customers.
The proposed study will collect feedback from diverse sources, including surveys, social media
platforms, and online reviews. Machine learning techniques, such as supervised learning
algorithms like Support Vector Machines and Recurrent Neural Networks (RNN), will be
employed to analyze the sentiment and extract valuable insights.

The expected outcomes of this research include a comprehensive understanding of customer

satisfaction levels, identification of critical improvement areas, and generation of actionable
recommendations to enhance Dahabshil Bank's fintech services. Moreover, the findings will
contribute to the broader knowledge of sentiment analysis in the context of Somali-language
feedback and fintech services, offering valuable implications for both academia and the industry.

Research Concept Note

Background of the Proposed Research
Natural Language Processing (NLP) is a subfield of artificial intelligence that focuses on
enabling computers to understand, interpret, and generate human language. Recent innovations,
such as transformer models, have led to significant improvements in the accuracy and
capabilities of NLP systems, especially in tasks like machine translation, text summarization, and
sentiment analysis (Vaswani et al., n.d.). NLP has a diverse applications, including sentiment
analysis, machine translation, and conversational agents. Sentiment analysis, specifically,
involves categorizing textual feedback into positive, negative, or neutral sentiments, allowing
organizations to assess customer satisfaction and make data-driven decisions (Londhe, 2022).

The fintech industry, which integrates technology-driven innovations such as mobile banking,
blockchain, and AI-powered financial tools into traditional financial services, which
encompasses technological innovations in financial services, has experienced rapid growth
globally, becoming a key driver of financial inclusion, particularly in emerging markets (Sahay
et al., 2020). Institutions like Dahabshil Bank in Somaliland play a vital role in providing
essential financial services, including mobile banking, online transfers, and remittances, catering
to Somali-speaking customers. Despite the growing importance of customer feedback in
improving service quality, there is a significant gap in automating the analysis of Somali-
language feedback due to the scarcity of annotated datasets and computational tools tailored to
this low-resource language (Nekoto et al., 2020).

Existing research has shown that African languages, including Somali, are underrepresented in
NLP, with limited models and resources available for tasks like sentiment analysis (Muhammad
et al., 2023). This disparity hinders the effective utilization of customer insights in fintech
services. For Dahabshil Bank, this challenge is particularly pronounced as the bank receives
substantial customer feedback in Somali, yet manual analysis methods are inefficient and not
scalable for real-time decision-making

Addressing these gaps, this research aims to leverage advanced machine learning techniques to
develop a sentiment analysis model for Somali. By creating annotated datasets and designing
models tailored to the linguistic and contextual peculiarities of the Somali language, this study
seeks to provide actionable insights for Dahabshil Bank. This approach will enable the institution
to enhance customer satisfaction, improve service delivery, and set a precedent for NLP
applications in low-resource languages and regions.
Problem Statement

The rapid growth of financial technology fintech services in Somaliland presents new
opportunities for improving customer experience, but the challenge of effectively analyzing
customer feedback remains. Dahabshil Bank, a key player in this space, receives substantial
customer feedback primarily in Somali, but the bank currently lacks an automated, scalable
method to understand the sentiment behind this feedback. This hinders the bank's ability to
proactively address customer concerns, improve services, and enhance overall customer
satisfaction.

The problem is compounded by the fact that the Somali language lacks sufficient resources and
models for natural language processing (NLP), making it difficult to accurately perform
sentiment analysis. There is also a lack of sentiment analysis tools tailored to the specific context
of fintech services in Somaliland, leading to an underutilization of valuable customer insights.

Research Questions

1. How can a sentiment analysis model be developed for the Somali language, given its lack
of computational resources and NLP tools?
2. What are the challenges in analyzing Somali-language customer feedback in the fintech
domain, and how can they be addressed through machine learning?
3. How effective is machine learning-based sentiment analysis in accurately classifying
customer feedback as positive, neutral, or negative in the context of Dahabshil Bank's
services?
4. What actionable insights can be drawn from the sentiment analysis results that will help
improve customer experience and service quality at Dahabshil Bank?

General Objective

To develop and implement an advanced machine learning-based sentiment analysis model for
Somali-language customer feedback, enabling Dahabshil Bank to identify sentiment patterns,
evaluate customer satisfaction, and improve fintech services

Specific Objectives

 To construct a high-quality, annotated dataset of Somali-language customer feedback on

fintech services at Dahabshil Bank.
  To design and implement an advanced machine learning model capable of accurately
analyzing the sentiment of Somali-language customer feedback.
 To evaluate the performance of the sentiment analysis model in terms of accuracy,
efficiency, and scalability.
 To provide actionable insights to Dahabshil Bank, enabling the institution to better
understand customer sentiments, improve service delivery, and enhance customer
satisfaction.

Research Variables and Objectives

This study focuses on sentiment analysis of Somali-language customer feedback for fintech
services at Dahabshil Bank, identifying the key variables and research objectives.

Independent Variables (IVs)

Factors influencing sentiment analysis and model training:

1. Customer Feedback (Textual Data):

 Length of Feedback (short, medium, long).

 Feedback Context (e.g., mobile banking, customer service).

2. Language Features:

 Syntax (sentence structure).

 Lexicon (choice of words, fintech terminology).
 Sentiment-Bearing Words (positive, neutral, negative).

3. Machine Learning Model:

 Model Type (e.g., SVM, BERT, LSTM).

 Training Algorithm (e.g., supervised learning).

Dependent Variables (DVs)

Outcomes measured:

1. Sentiment Classification:

 Positive, Neutral, or Negative feedback.

 2. Model Accuracy:

 How well the model classifies sentiment.

3. Actionable Insights:

 Trends in customer satisfaction.

 Recommendations for service improvements.

Significance of the Research

This research holds significant potential in addressing several key issues in both the fintech
sector and the broader health systems field in Somaliland, as well as offering broader policy
implications for data-driven decision-making in developing regions. Below are the key areas
where the research will have a beneficial impact.

Proposed Methodology

This study aims to analyze Somali-language customer feedback on Dahabshil Bank’s fintech
services using machine learning-based sentiment analysis. The methodology involves five key
stages:

1. Study Setting and Population

 Setting: Focused on Dahabshil Bank’s feedback systems and online platforms.

 Population: Somali-language feedback from customers using fintech services such as
mobile banking and online transfers.

2. Sampling

 Method: Stratified random sampling to capture diverse sentiments (positive, neutral,

negative).
 Sample Size: 1,000 feedback entries, proportional to the sentiment distribution in the
dataset.
 3. Data Collection

 Source: Feedback from mobile apps, customer service, and online platforms. google
forms,
 Process: Automated text scraping and preprocessing (e.g., tokenization, stop-word
removal, lemmatization).
 Annotation: Sentiment labels (positive, neutral, negative) (Wanaagsan, Dhexdhexaad,
Xun) assigned by a native Somali speaker for model training.

4. Data Analysis

 Approach: Supervised machine learning to classify sentiment.

 Steps: Feature extraction (e.g., TF-IDF, embeddings) model selection (e.g., SVM,
Logistic Regression, LSTM, BERT), and evaluation using metrics like accuracy,
precision, recall, and F1-score.
 Insights: Extract actionable feedback themes for service improvements.

5. Timeline and Outcomes

 Duration: 5–7 months (data collection, model development, evaluation, and insights
generation).
 Expected Outcome: A sentiment analysis model tailored for Somali-language feedback,
delivering insights to enhance customer satisfaction and fintech service quality.

Conclusion

The proposed research will have broad-ranging benefits in improving fintech services, financial
inclusion, public health systems, and policy-making in Somaliland. By automating sentiment
analysis for Somali-language feedback, the study can also serve as a model for similar initiatives
in low-resource languages and regions, helping shape a more inclusive, data-driven society.

Reference

Londhe, S. (2022). International e-journal of Library Science: Editorial Advisory Committee: :

Communication Address : 10.

Muhammad, S. H., Abdulmumin, I., Ayele, A. A., Ousidhoum, N., Adelani, D. I., Yimam, S. M.,

Ahmad, I. S., Beloucif, M., Mohammad, S. M., Ruder, S., Hourrane, O., Brazdil, P.,

Jorge, A., Ali, F. D. M. A., David, D., Osei, S., Bello, B. S., Ibrahim, F., Gwadabe, T., …

Arthur, S. (2023). AfriSenti: A Twitter Sentiment Analysis Benchmark for African

Languages. EMNLP 2023 - 2023 Conference on Empirical Methods in Natural Language

Processing, Proceedings, 13968–13981. https://doi.org/10.18653/v1/2023.emnlp-

main.862

Nekoto, W., Marivate, V., Matsila, T., Fasubaa, T., Kolawole, T., Fagbohungbe, T., Akinola, S.

O., Muhammad, S. H., Kabongo, S., Osei, S., Freshia, S., Niyongabo, R. A., Macharm,

R., Ogayo, P., Ahia, O., Meressa, M., Adeyemi, M., Mokgesi-Selinga, M., Okegbemi, L.,

… Bashir, A. (2020). Participatory research for low-resourced machine translation: A

case study in African languages. Findings of the Association for Computational

Linguistics Findings of ACL: EMNLP 2020, 2144–2160.

https://doi.org/10.18653/v1/2020.findings-emnlp.195

Sahay, R., Allmen, U., Lahreche, A., Khera, P., Ogawa, S., & Beaton, K. (2020). The Promise of

Fintech Financial Inclusion in the Post COVID-19 Era. Departmental Papers, 2020.

https://doi.org/10.5089/9781513512242.087

Vaswani, A., Shazeer, N., Parmar, N., Uszkoreit, J., Jones, L., Gomez, A. N., Kaiser, Ł., &

Polosukhin, I. (n.d.). Attention is All you Need.

Title Two
 " Examining the Evolution of Financial Technology (Fintech) Solutions at
Dahabshil Bank International and Their Role in Driving Innovation Using
Machine Learning Approaches "

Research Concept Note

1. Introduction

The rapid growth of financial technology (fintech) has revolutionized traditional banking,
enabling innovative solutions to enhance customer experiences, operational efficiency, and
financial inclusion. Fintech has become a critical tool for bridging gaps in financial access and
modernizing banking systems globally. Dahabshil Bank International, a leading financial
institution in Somaliland, has embraced fintech to meet the unique needs of its diverse clientele.

Operating in a region where traditional financial systems face significant challenges, Dahabshil
Bank has implemented fintech solutions that provide secure, accessible, and efficient services.
The adoption of machine learning techniques has further enhanced these innovations, allowing
the bank to offer personalized services, detect fraud, and make data-driven decisions. Machine
learning's ability to analyze large datasets and predict trends has proven invaluable in creating
more responsive and secure financial systems.

Despite the progress, there is limited research on the evolution of fintech in Somaliland and its
impact on innovation in local institutions like Dahabshil Bank. Most studies focus on fintech
developments in advanced economies, leaving a gap in understanding the specific dynamics of
emerging markets. Additionally, the integration of machine learning into fintech solutions in
these contexts remains underexplored.

This study aims to address these gaps by examining the evolution of fintech solutions at
Dahabshil Bank International and assessing their role in driving innovation through machine
learning approaches. By analyzing the historical development, current practices, and impact of
these solutions, the research seeks to highlight the transformative potential of fintech in
advancing financial services and fostering economic growth in Somaliland.

2. Statement of the Problem

Despite the significant strides made in adopting financial technology (fintech) solutions, a
noticeable gap exists in the academic and practical understanding of how these solutions
have evolved and contributed to innovation within the specific context of Somaliland.
Fintech has been a transformative force globally, yet its development and impact in emerging
markets like Somaliland have not been thoroughly documented or analyzed. The local
financial landscape presents unique challenges, such as limited infrastructure, regulatory
constraints, and varying levels of digital literacy, which influence the adoption and evolution
of fintech solutions.
In this context, Dahabshil Bank International has emerged as a pioneer, leading the way in
integrating fintech into its operations to address these challenges. However, the processes,
strategies, and outcomes of this integration remain underexplored. This lack of research hinders a
comprehensive understanding of how fintech innovations contribute to fostering financial
inclusion, improving operational efficiency, and driving customer satisfaction in Somaliland’s
financial sector.

Additionally, the role of machine learning in analyzing and enhancing fintech solutions is
underutilized and poorly understood in this region. Machine learning, with its capacity to process
large volumes of data, identify patterns, and deliver predictive insights, has immense potential to
optimize fintech innovations. Its application could provide significant value in areas such as
fraud detection, customer behavior analysis, and personalized financial services. However,
limited exploration of its capabilities within local financial institutions like Dahabshil Bank has
constrained the ability to fully harness its benefits.

Understanding these dynamics is crucial for fostering sustainable development and maximizing
the impact of fintech innovations in Somaliland. A deeper analysis of how these technologies are
implemented, the challenges encountered, and their outcomes could provide valuable insights for
other financial institutions, policymakers, and stakeholders. Furthermore, it could pave the way
for more effective integration of advanced technologies like machine learning to drive innovation
and strengthen the financial ecosystem in emerging markets.

3. Research Questions

This study will address the following questions:

1. How have fintech solutions evolved at Dahabshil Bank International over the years?
2. What role has machine learning played in driving innovation within these fintech
solutions?
3. What challenges and opportunities exist in implementing machine learning for fintech
innovation in Somaliland?

4. Objectives

General Objective

To examine the evolution of fintech solutions at Dahabshil Bank International and their role in
driving innovation through machine learning approaches.

Specific Objectives

1. To analyze the historical development of fintech solutions at Dahabshil Bank

International.
2. To evaluate the contribution of machine learning techniques to fintech innovation.
 3. To identify challenges and opportunities in adopting machine learning for fintech
solutions in the Somaliland context.

5. Methodology

Research Design

The study will adopt a mixed-methods approach, combining qualitative and quantitative
techniques.

Data Collection Methods

1. Primary Data:
o Interviews with key stakeholders (e.g., bank executives, IT specialists).
o Surveys targeting customers and employees of Dahabshil Bank.
2. Secondary Data:
o Review of existing literature, annual reports, and case studies on fintech and
machine learning.

Data Analysis Methods

 Qualitative Data: Thematic analysis for interview and survey responses.

 Quantitative Data: Statistical analysis and machine learning models to assess patterns
and trends in fintech innovation.

Study Scope

The study will focus on Dahabshil Bank International's fintech solutions and their evolution from
inception to current implementations.

6. References

Brynjolfsson, E., & McAfee, A. (2017). Machine, Platform, Crowd: Harnessing Our Digital
Future. W. W. Norton & Company.
Goldstein, I., Jiang, W., & Karolyi, G. A. (2019). The Impact of Fintech on the Financial
Industry. Annual Review of Financial Economics, 11(1), 225-247.

World Bank. (2022). The Role of Fintech in Financial Inclusion. Retrieved from World Bank
website.

Dahabshil Bank International. (2023). Annual Report.