Addis Ababa University

Master’s Thesis Proposal

Name of Student: Melat Engdashet Kabtimer

College: Natural Sciences

Department: Computer Science

Major: Data and Web Engineering

Advisor: Dr. Minale Ashagire (PhD)

Title of Thesis: zkTranscript: Privacy-Preserving Academic Credential Verification

Framework Using Blockchain Technology

October, 2025
Table of Contents

1. Introduction ................................................................................................................. 1
2. Motivation ................................................................................................................... 2
3. Statement of the Problem ............................................................................................ 3
4. Objective ..................................................................................................................... 4
5. Methods ....................................................................................................................... 5
6. Related Work ............................................................................................................... 6
7. Scope and Limitations ................................................................................................. 9
8. Application of Result................................................................................................... 9
Annexes ............................................................................................................................ 10
Annex A: Timetable ..................................................................................................... 10
Annex B: Cost............................................................................................................... 10
References......................................................................................................................... 11

i
List of Acronyms

Acronym Full Term

CGPA cumulative Grade Point Average
CNN Convolutional Neural Network
CSV Comma-Separated Values
DID Decentralized Identifier
DL Deep Learning
DSR Design Science Research
GPA Grade Point Average
IPFS Interplanetary File System
JSON JavaScript Object Notation
ML Machine Learning
PoS Proof of Stake
QR Quick Response (Code)
ZKP Zero-knowledge Proof
zk-SNARK Zero-knowledge Succinct Non-interactive Argument of Knowledge
zk-STARK Zero-knowledge Scalable Transparent Argument of Knowledge
Zkcert Zero-knowledge Certificate Verification Framework
ZKBAR-V Zero-knowledge Blockchain Academic Record verification Syst

ii
1. Introduction

Academic credentials such as degrees and transcripts are essentials for validating an
individual’s educational achievements. Academic credentials are mostly used to secure
Employability, professional and academic advancement. However, in today’s digital era,
the high incr

ease of forgery, misrepresentation, and inefficient verification mechanism of academic

credentials is becoming a serious issue. Global Education Monitoring Report [1],
fraudulent credentials and diploma mills cost the global economy over US $600 million
annually, revealing how vulnerable current systems remain to manipulation.

The challenges are more severe in developing countries, where systems that verify
credentials are manual and resource constrained. In Ethiopia, an audit conducted by
SICPA/CERTUS exposed 921 forged degrees among 18,000 degrees (5%). Such
falsifications compromise institutional reputation, recruitment of unqualified individuals,
undermine service quality and productivity. Manual verifications relying on inter-
institutional consultation, takes several weeks to complete. Centralized databases remain
susceptible to tampering, insider attacks, and data breaches [3,4].

Existing early digital approaches aimed to automate verification. Machine Learning (ML),
Deep Learning (DL) models, and Ontology-based frameworks can be applied to detect
anomalies and identify forged academic documents [5, 6, 7]. These systems improved
fraud detection, reduced human oversight, they are still dependent on trusted
intermediaries, and lack mechanisms for validation without the need for trust.

blockchain technology, a decentralized ledger system that securely records transactions

across a distributed network has emerged as a promising solution [4, 8, 9]. Systems such
as Blockcerts and EduCTX illustrate that distributed ledgers can enhance data integrity,
transparency, and auditability by eliminating central authorities. Nevertheless,
blockchain’s inherent transparency creates privacy challenges, since sensitive academic
details could be publicly visible unless additional cryptographic safeguards are applied
[10, 11].

This privacy issues can be solved by Zero-knowledge Proofs (ZKPs) i.e., a cryptographic
method to verify a statement without the need to reveal underlying information. ZKP allow
a verifier to confirm the authenticity of a document without revealing the underlying data

1
[12] (for e.g. student x passes course A without revealing the exact grade). This property
of selective disclosure makes ZKPs highly suitable for privacy-preserving academic
credential verification. Specific frameworks—zkCert [10], ZKBAR-V [16], and the
privacy-centric framework of El Haji et al. [15]—demonstrated that ZKPs can be
integrated with blockchain to achieve both security and confidentiality.

Despite these advances, the above-mentioned systems focus on simple, single-attribute

proofs. (e.g. verifying a degree) thus, not supporting verification of multi-attribute
academic transcripts. This contains interdependent data such as courses, grades, credits,
and CGPA. Moreover, little research has quantitatively assessed the performance and
scalability of different ZKP protocols in this domain [10, 11, 16].

This study, entitled zkTranscript, addresses the mentioned gaps by designing a privacy-
preserving academic credential verification framework that leverages transcript-specific
ZKP circuits built on zk-SNARK and zk-STARK technologies. This research focuses on
developing and validating a blockchain based system that enables secure, selective, and
multi-attribute credential verification without exposing sensitive data using ZKP. The
framework will attempt to show how privacy-preserving mechanisms can improve
confidentiality, integrity, and trust in digital credential verification through experimental
implementation.

2. Motivation

In recent years, numerous reports have exposed that several individuals present forged
academic documents to secure employment or admission opportunities. This scenario
particularly prevails in developing countries with high unemployment rates for example,
in Ethiopia, the total unemployment rate is about 3.4 percent according to recent world
bank estimates [17]. The widespread circulation of fraudulent educational transcripts has
caused the recruitment of underqualified individuals, often leading to poor organizational
performance, compromised institutional integrity, and financial losses. This recurring
concern inspired this research to examine existing transcript verification mechanisms and
locate key research gaps where meaningful contribution can be achieved.

2
3. Statement of the Problem

The digitalization of education has inadvertently simplified the creation and circulation of
counterfeit academic credentials ranging from diplomas obtained from non-existent
universities to sophisticated forgery of degrees in reputable universities [1]. An estimated
100,000 fake degrees are sold each year in the United States alone, about one-third of
which are graduate degrees [18]. While in Ethiopia, a verification project by
SICPA/CERTUS discovered 921 counterfeit degrees out of 18,000 degrees [2]. However,
the project relies on printed certificates or security/QR codes thus, not providing strong
cryptographic or privacy protections.

Traditional verification systems, manual checks and centralized databases- are slow,
costly, and vulnerable [2, 8]. Manual verification may take weeks creating delays for
students and employers, while centralized databases create a single point of failure [2, 12]
This vulnerability can be further solidified by the 2017 Equifax breach incident, where
hackers were able to leak the personal data of about 147 million people [3] prompting the
need for decentralized, more resilient verification mechanism.

Several technologies including deep learning, machine learning, and blockchain are
adopted for verification of academic transcripts. Deep learning approach that use CNN-
based approach can localize and detect tampered regions in transcript [5]. However, their
reactive nature, i.e. detecting fraud after rather than before issuance prevents the approach
from being effective. Ontology-based, semantic web frameworks [7] rely on trusted third
issuers and lack cryptographic mechanisms to prevent forgery or alteration. Similarly, a
Machine learning approach that incorporates blockchain for reliability achieved high
accuracy [19]. However, such approach suffers from high computational complexity as
dataset size increases.

Blockchain technology for academic credential verification has emerged as a promising

solution, by offering a decentralized and immutable distributed ledger for credential
verification. However, the significant limitation of such systems is the privacy trade-off:
to verify a credential, requires the entire document to be disclosed. Zero-knowledge proofs
(ZKPs) are able to resolve this tension. zkCert proposed by [10], which combines zk-
SNARKs with Merkle commitments to enable selective disclosure of credential attributes.
While it is feasible, zkCert is limited to simple, single-attribute claims. The ZKBAR-V (a

3
dual-blockchain framework) advances the field by incorporating zk-SNARKs, DIDs and
IPFS to ensure privacy and provides credential verification [16]. Although the
implementation highlighted that the scope is confined to credential-level proofs and does
not evaluate the performance of ZKPs for complex, multi-attribute transcripts.

This research seeks to address the lack of privacy-preserving and multi-attribute credential
verification frameworks that can be effectively implemented using zero-knowledge proofs
(ZKPs). While existing studies have explored ZKP-based verification, few have
demonstrated domain-specific applications for academic credentials. Therefore, this study
aims to design and implement zkTranscript, a blockchain based framework that applies
transcript-specific ZKP circuits to enable secure and selective credential verification. The
framework will be validated in terms of its efficiency, and privacy compliance to
demonstrate its feasibility for real- word adoption in academic credential verification.

The following research questions can be derived from the statement of the problem:

1. How can privacy preserving transcript-specific circuits be designed and

implemented for multi-attribute academic transcript verification?
2. How efficiently can transcript-specific circuit designs support selective disclosure
of multiple transcript attributes?
3. How does the performance of Zero-Knowledge proof (ZKP) protocols differ based
on validation on evaluation metrics?

4. Objectives

General Objective

The general objective of this research is to design, implement, and validate zkTranscript,
a blockchain-based framework for privacy-preserving academic credential verification.

4
Specific Objectives

➢ To prepare a real academic transcript dataset or generate a synthetic dataset when

real data is unavailable.
➢ To design a transcript-specific ZKP circuits that allow selective proof of attributes
without revealing the full transcript.
➢ To implement the designed verification circuits using ZKP protocols.
➢ To validate and evaluate the performance of ZKP protocols in terms of specified
evaluation metrics.

5. Methods

This research follows a Design Science Research (DSR) methodology as outlined by

Hevner et al. [20] and extended by Peffers et al. [21]. This methodology explors the design,
implementation, and evaluation of a privacy-preserving academic transcript verification
framework which we call zkTranscript through the structure of the following five stages.

Literature Review

Technologies for academic credential verification have evolved significantly over the last
decade, driven by the need for reliability, scalability, and privacy in educational data
management. This section reviews major technological approaches relevant to the
zkTranscript framework, focusing on key domains: traditional systems, machine learning,
ontology-based frameworks, blockchain-based verification, and zero-knowledge proof
systems.

Dataset Preparation

A small representative dataset obtained from publicly available sources such as UCI or
KAGGLE will be used to simulate transcript records. If suitable anonymous data fails, a
synthetic dataset will be created to represent academic transcripts containing attributes
such as student ID, course codes, grades, credit hours, CGPA. A simulated record can be
generated using Python in JSON or CSV format in a real university schema.

5
Artifact Design

The zkTranscript will be organized into three architectural layers. The Issuance layer is
the layer that generates and commit transcript data cryptographically using tools such as
Python, Hashlib. The proof generation layer encodes the transcript attributes and generate
selective disclosure proofs using tools SnarkJS or Circom 2.1. In verification layer, ZKP
are verified via on-chain verifiers using tools such as Ganache, or StarkNet.

Prototype Implementation

This phase involves developing zk-SNARK and zk-STARK circuits, to generate proofs
and validate them via blockchain methods. The circuit will be implemented using Circom,
or SnarkJS. SNARK and STARK can be used to generate proofs.

The blockchain implementation and integration can be carried on ubuntu 22.04 with the
help of programming languages such as Python and [Link].

Evaluation and Validation

The zkTranscript framework will be validated on performance metrics such as proof

generation time, verification latency, proof size, and computational cost to ensure that zk-
SNARK and zk-STARK circuits function correctly and achieve privacy-preserving
verification, and to illustrate practical trade-offs between the two protocols. These metrics
will be observed under controlled conditions and interpreted comparatively, focusing on
trends rather than hardware-dependent precision.

6. Related Work

The authors in [4] proposed a blockchain-based credential verification framework The

system utilized a PoS consensus model and QR code integration that improves institutional
verification efficiency. The prototype was evaluated on a Kaggle dataset of about 1,000
academic records, achieving an average verification time of 5 seconds, and processing
capacity of over 1,000 verification per hour with a record of 34.67% error rate. Evaluation
metrics like proof size and computational cost were not tested. While the approach
enhanced operational efficiency, it required full credential disclosure during validation,
leaving privacy unaddressed, and the framework is limited to single -attribute verification.
6
The authors in [10] presented a framework which builds upon and extends earlier privacy-
preserving systems which is zkCert. While zkCert established the use of zk-SNARKs for
selective disclosure, this work extended the architecture by more deeply integrating the
proofs with smart contracts on the Ethereum blockchain. However, the dataset used is not
mentioned. The paper does not make analysis on performance metrics. While the authors
succeeded to advance privacy-preserving credential verification, the scope is limited to
single-attribute credentials rather than multi-attribute academic transcripts.

The authors in [11] proposed zk-creds, a protocol that explicitly aims to improve existing
anonymous credential schemes, as they are difficult to scale and deploy in practice, their
framework leverages zk-SNARKs and Merkle trees to create flexible credentials. The
system was implemented in Rust using the Arkworks library. However, the specific dataset
used for benchmarking is still not mentioned. While the paper illustrated flexibility and
advanced certain, its evaluation does not measure its performance against other ZKP types
for benchmark. The framework is designed for general-purpose credentials and fails to
extend to complex, multi-attribute academic transcripts.

The authors in [15] proposed a framework that extends the core principle of zkCert [10]
using ZKPs for selective disclosure- by integrating it more deeply with a blockchain smart
contract layer to automate and decentralize the verification process itself, positioning their
work as an architectural evolution. Implementation was carried on the Ethereum
blockchain using zk-SNARK proof. However, the specific dataset remains to be unknown.
The system was not measured on evaluation metrics. Its application is still confined to
simple, single-attribute credentials.

The authors in [16] proposed ZKBAR-V, a dual-blockchain framework integrating zk-

SNARKs and IPFS. A standard Ethereum smart contract verification was used as a
benchmark for the system. Evaluation results showed that the system can handle 280
concurrent requests and achieved verification speed of 120x faster than standard Ethereum
smart contract verification. However, the dataset used for the evaluation is not mentioned.
While the framework advanced privacy-preserving verification, its evaluation did not
include key metrics like proof generation time. Furthermore, the scope is confined to
single-attribute academic credential verification.

7
Similarly, the authors in [22] proposed BLS-MT-ZKP, a novel approach that enhances
selective disclosure mechanisms in digital credential by combining Boneh-Lynn-Shacham
(BLS) signatures with Merkle Tree-based Zero-Knowledge Proofs (ZKPs). The system
implemented in Python promised to have enhanced efficiency and smaller proof size
compared to traditional ZKP-based credential systems. The specific dataset used was not
mentioned. The study stated that the result indicated improvement in proof size and
verification efficiency, but does not provide the specific quantitative values. The
framework does not extend to complex, multi-attribute academic transcript.

In summary, while existing research has evolved from basic blockchain systems for
integrity to ZKP-based frameworks for privacy, a critical limitation persists across all
studies. All studies are confined to single-attribute credential verification and lack a
comprehensive, empirical performance evaluation for complex, multi-attribute academic
transcripts. Works such as zkCert [10], ZKBAR-V [16], and others [11, 15, 22]
successfully demonstrate selective disclosure but fail to benchmark ZKPs circuits against
metrics like proof generation time and computational cost. This unified gap underscores
the necessity of the present study, which evaluate the performance of zk-SNARK and zk-
STARK protocols to determine their feasibility for multi-attribute academic verification.

8
7. Scope and Limitations

This research focuses on the design, implementation evaluation of ZKPs circuits for
privacy preserving academic transcripts. the framework integrates Zero-Knowledge proofs
with blockchain technology to ensure authenticity and confidentiality of educational
records. The study is confined in using zk-SNARK and zk-STARK protocols. Evaluation
will concentrate on computational efficiency, proof generation time, verification latency,
and efficiency as input data increases under controlled experimental settings. User
interface design and large-scale network deployment fall outside of the scope.

8. Application of Result

The outcome of this research will show the feasibility of using ZKPs for privacy-
preserving verification of multi-attribute academic transcripts. The findings will be able to
strengthen the integrity, confidentiality, and efficiency of digital credential management.
Thus, enabling universities, employers, and government agencies to be able to verify
academic records without the need to expose the underlying or sensitive data. The
proposed system can be extended to national digital ID and e-governance systems where
privacy-preserving authentication with selective disclosure are essential. The research
contributes to building credible digital ecosystems that support education, employment
and governance transformation.

9
Annexes
Annex A: Timetable

Figure 1: Timetable

Annex B: Cost
Table 1: Cost and budget

Expense Unit Total (ETB)

Internet 1300 * 8 month 10,400

Transport 1000 * 8 months 8,000

Printing, Photocopying, 500 * 3 copy 1,500

& Binding

Stationary Flash drives, notebooks 1,000

Contingency (10%) - 2,090

Total Estimated Cost 22,990

10
Reference
[1] UNESCO, Global Education Monitoring Report 2022: Building Credibility in
Academic Qualifications, Paris, France: UNESCO, 2022.

[2] SICPA/CERTUS, Credential Integrity Audit Report: Ethiopia Higher Education

Verification, Addis Ababa, Ethiopia: SICPA, 2024.

[3] A. Puig, "Equifax Data Breach Settlement: What You Should Know," Federal Trade
Commission, Jul. 22, 2019.

[4] N. and Y. Xu, “Development of Blockchain-Based Academic Credential

Verification System,” OAlib, vol. 11, no. 09, pp. 1–20, 2024.

[5] P. Yang, W. Fang, F. Zhang, L. Bai, and Y. Gao, “Document Image Forgery
Detection Based on Deep Learning Models,” in Proceedings - 2022 International
Symposium on Electrical, Electronics and Information Engineering, ISEEIE 2022,
Institute of Electrical and Electronics Engineers Inc., 2022, pp. 36–41.

[6] S. Murugesan, M. B. Lakshminarasaiah, S. Ramaiah, and R. Balarama, “Document

Recognition in Education Sector Using Machine Learning Algorithms,”
Mathematical Modelling of Engineering Problems, vol. 12, no. 4, pp. 1335–1342,
2025.

[7] D. Pinto, F. Vargas, and N. Piedra, “Ontology for open digital educational
credentials: A step towards decentralized learning paths,” in Proceedings of the
International Conference on Innovative Technologies and Learning (ICITL 2023),
Singapore: Springer, 2023.

[8] A. Grech and A. Camilleri, Blockchain in Education, Luxembourg: Publications

Office of the European Union, 2017.

[9] T. Rahman, S. I. Mouno, A. M. Raatul, A. K. Al Azad, and N. Mansoor, “Verifi-

Chain: A Credentials Verifier using Blockchain and IPFS,” Jul. 2023.

11
[10] R. Q. Saramago, H. Meling, and L. N. Jehl, “A Privacy-Preserving and Transparent
Certification System for Digital Credentials,” in Leibniz International Proceedings
in Informatics, LIPIcs, Schloss Dagstuhl- Leibniz-Zentrum fur Informatik GmbH,
Dagstuhl Publishing, Feb. 2023.

[11] M. Rosenberg, J. White, C. Garman, and I. Miers, “zk-creds: Flexible Anonymous

Credentials from zkSNARKs and Existing Identity Infrastructure.”

[12] E. Ben-Sasson, A. Chiesa, D. Genkin, E. Tromer, and M. Virza, “SNARKs for C:

Verifying program executions succinctly and in zero knowledge,” in Proc. 33rd
Annu. Int. Cryptology Conf. (CRYPTO), Santa Barbara, CA, USA, 2013, pp. 90–
108.

[13] N. Naik and P. Jenkins, “zkCert: Zero-knowledge proof-based digital certificate

verification on blockchain,” IEEE Access, vol. 8, pp. 227138–227152, 2020.

[14] J. Wang, L. Chen, and W. Yu, “ZKBAR-V: A privacy-preserving blockchain

authentication and verification system using zero-knowledge proofs,” Computers
& Security, vol. 113, p. 102546, 2022.

[15] R. El Haji, R. Karam, and A. Chehab, “A privacy-preserving blockchain framework

for academic credential verification using zero-knowledge proofs,” International
Journal of Information Security and Privacy, vol. 17, no. 4, pp. 1–16, 2023.

[16] J. A. Berrios Moya, J. Ayoade, and M. A. Uddin, “A Zero-Knowledge Proof-

Enabled Blockchain-Based Academic Record Verification System,” Sensors, vol.
25, no. 11, Jun. 2025.

[17] World Bank, “Employment to population ratio, 15+, total (%) – Ethiopia,” World
Bank Data, 2024.

[18] SICPA, Facts and Figures – Document Protection and Verification, Lausanne,
Switzerland: SICPA, 2025.

12
[19] D. Shah, D. Patel, J. Adesara, P. Hingu, and M. Shah, “Integrating machine learning
and blockchain to develop a system to veto the forgeries and provide efficient
results in education sector,” Vis Comput Ind Biomed Art, vol. 4, no. 1, Dec. 2021.

[20] A. R. Hevner, S. T. March, J. Park, and S. Ram, "Design Science in Information

Systems Research," MIS Quarterly, vol. 28, no. 1, pp. 75–105, Mar. 2004.

[21] K. Peffers, T. Tuunanen, M. A. Rothenberger, and S. Chatterjee, "A Design Science

Research Methodology for Information Systems Research," Journal of
Management Information Systems, vol. 24, no. 3, pp. 45–77, Dec. 2007.

[22] Š. B. Ramić, I. Prazina, D. Pozderac, R. T. Mulahasanović, and S. Mrdović, “BLS-

MT-ZKP: A novel approach to selective disclosure of claims from digital
credentials,” Dec. 2024.

13
Submitted by:

___________________ _______________ _____________

Name Signature Date

Approved by:

1. _________________ _____________ _____________

Advisor Signature Date

2. __________________ _____________ _____________

Chairman Signature Date

Department’s Graduate Committee

14