hh.sePublications
Planned maintenance
A system upgrade is planned for 17/11-2026, at 12:00-13:00. During this time DiVA will be unavailable.
Change search
Link to record
Permanent link

Direct link
Picazo-Sanchez, PabloORCID iD iconorcid.org/0000-0002-0303-3858
Publications (10 of 11) Show all publications
Picazo-Sanchez, P. & Ortiz-Martin, L. (2026). Evaluating the Integrity of LLM-Generated Citations: Prevalence and Risks of Fabricated References in Scientific Literature. Data, 11(5), 1-24, Article ID 122.
Open this publication in new window or tab >>Evaluating the Integrity of LLM-Generated Citations: Prevalence and Risks of Fabricated References in Scientific Literature
2026 (English)In: Data, E-ISSN 2306-5729, Vol. 11, no 5, p. 1-24, article id 122Article in journal (Refereed) Published
Abstract [en]

Large Language Models have become important in our lives, and academia is not agnostic to this trend, offering tools like text rephrasing and summarisation. However, this integration raises significant concerns regarding the integrity of science. In this paper, we investigate hallucinations of LLMs when generating scientific references. Using nine LLMs, we generated a dataset of 74,196 BIBTEX references to quantify and analyse fabricated references, focusing on distinguishing between intrinsic and extrinsic hallucinations. Also, we extracted and analysed 127,063 references from 3541 published papers in 2023 to assess the prevalence of fake bibliographic data. Our manual verification process identified eight instances of fabricated references. While the overall rate is statistically low, the mere existence of fabricated content in the peer-reviewed literature is a critical integrity issue, demonstrating a vulnerability in current academic validation systems. The significance of our finding is not the statistical prevalence but rather the necessity for rigorous, human-validated processes to prevent the injection of spurious citations regardless of their source. © 2026 by the authors.

Place, publisher, year, edition, pages
Basel: MDPI, 2026
Keywords
hallucinations, scientific references, NLP, trust
National Category
Computer and Information Sciences
Identifiers
urn:nbn:se:hh:diva-60071 (URN)10.3390/data11050122 (DOI)001775420000001 ()2-s2.0-105041620873 (Scopus ID)
Available from: 2026-07-17 Created: 2026-07-17 Last updated: 2026-07-17Bibliographically approved
Flemark, A., Paulander, A. & Picazo-Sanchez, P. (2026). From comments to threats: leveraging NLP to detect malicious browser extensions. Journal of Computer Virology and Hacking Techniques, 22(1), 1-15, Article ID 8.
Open this publication in new window or tab >>From comments to threats: leveraging NLP to detect malicious browser extensions
2026 (English)In: Journal of Computer Virology and Hacking Techniques, ISSN 2263-8733, Vol. 22, no 1, p. 1-15, article id 8Article in journal (Refereed) Published
Abstract [en]

With the increasing popularity of browser extensions, there is a growing concern about malicious actors exploiting this software to distribute malware. Existing solutions rely on manual review processes or traditional methods like static, dynamic or hybrid analysis, which are insufficient in detecting complex and evolving threats. In this study, we investigate the potential of Natural Language Processing (NLP) for automatically classifying users’ comments in the Chrome Web Store, the public repository where extensions are stored and distributed, to identify malicious extensions. We propose a novel framework called CoTH that leverages NLP techniques to analyse the textual feedback provided by users and detect patterns indicative of malicious activity. We evaluate the accuracy of our model using a dataset of user reviews and demonstrate its effectiveness in identifying malicious extensions. Our findings suggest that NLP-based comment analysis can be a valuable addition to existing security measures, providing an opportunity for more accurate and efficient detection of malware in the Chrome Web Store. © The Author(s) 2025.

Place, publisher, year, edition, pages
Cham: Springer, 2026
Keywords
Browser extensions, Malware, NLP, Web security
National Category
Computer Sciences Computer Systems
Research subject
Smart Cities and Communities, LEADS
Identifiers
urn:nbn:se:hh:diva-58126 (URN)10.1007/s11416-025-00578-w (DOI)001642761100001 ()2-s2.0-105025349479 (Scopus ID)
Available from: 2026-01-09 Created: 2026-01-09 Last updated: 2026-01-12Bibliographically approved
Picazo-Sanchez, P. (2025). Breaking the silence: the rise of hate speech in Spanish politics. Journal of Computational Social Science, 8(3), 1-26, Article ID 81.
Open this publication in new window or tab >>Breaking the silence: the rise of hate speech in Spanish politics
2025 (English)In: Journal of Computational Social Science, ISSN 2432-2717, E-ISSN 2432-2725, Vol. 8, no 3, p. 1-26, article id 81Article in journal (Refereed) Published
Abstract [en]

In this paper, we analyse the evolving role of hate speech in political rhetoric in Spain over the past three decades, focusing on its impact on electoral outcomes. Through a longitudinal analysis of political discourse in Parliament using Natural Language Processing (NLP), we calculate the likelihood of hate speech and we find significant changes in the frequency and intensity of inflammatory speeches made by politicians. Our findings suggest that hate speech is linked to higher voter mobilisation, indicating that such rhetoric may influence voter turnout and election dynamics. Furthermore, we show that since the rise of the extreme-right party in Parliament in 2019, that tendency has shifted, and parties that refrained from using hate speech tended to attract more voters, highlighting the potential electoral cost of employing divisive language. Additionally, we show how this party influenced others to escalate their rhetoric, increasing in some cases the likelihood of hate speech by around 200% in less than 5 years. This study contributes to understanding the use of hate speech by politicians in Parliament in polarising political discourse and marginalising minority groups. © The Author(s) 2025.

Place, publisher, year, edition, pages
Springer, 2025
Keywords
Hate speech, Political interventions, NLP
National Category
Political Science (Excluding Peace and Conflict Studies)
Research subject
Smart Cities and Communities, LEADS
Identifiers
urn:nbn:se:hh:diva-58385 (URN)10.1007/s42001-025-00413-8 (DOI)
Available from: 2026-02-11 Created: 2026-02-11 Last updated: 2026-02-17Bibliographically approved
Changrampadi, M. H., Almgren, M., Picazo-Sanchez, P. & Ali-Eldin, A. (2025). Snort Meets Transformers: Accelerating Transformer-Based Network Traffic Classification for Real-Time Performance. In: Proceedings Of The 18th European Workshop On Systems Security, Eurosec 2025: . Paper presented at 18th European Workshop on Systems Security, Rotterdam, The Netherlands, 31 Marsch, 2025 (pp. 33-39). New York: Association for Computing Machinery (ACM)
Open this publication in new window or tab >>Snort Meets Transformers: Accelerating Transformer-Based Network Traffic Classification for Real-Time Performance
2025 (English)In: Proceedings Of The 18th European Workshop On Systems Security, Eurosec 2025, New York: Association for Computing Machinery (ACM), 2025, p. 33-39Conference paper, Published paper (Refereed)
Abstract [en]

Transformer-based models have emerged as a powerful solution for network traffic classification, achieving high accuracy by autonomously learning patterns in raw traffic data. However, their high computational costs make real-time deployment impractical. In contrast, industry-proven tools like Snort and Suricata offer efficient network analysis but rely on manually crafted signatures, resulting in slower updates and limited adaptability to emerging threats. In this work, we propose a cascading model that leverages the strengths of both approaches. During training, a transformer-based model learns traffic patterns, which are then extracted using SHAP analysis to enhance the knowledge base of a signature-based IDS. In deployment, the IDS handles routine classifications, while only complex cases are escalated to the transformer model. Our experiments combining the analysis of ET-BERT with Snort demonstrate a four-fold performance improvement over running only ET-BERT without compromising false positive or false negative rates.

Place, publisher, year, edition, pages
New York: Association for Computing Machinery (ACM), 2025
Keywords
Network Traffic Analysis, Network Pre-trained Models, Intrusion Detection Systems (IDS)
National Category
Computer and Information Sciences
Identifiers
urn:nbn:se:hh:diva-56062 (URN)10.1145/3722041.3723098 (DOI)001477796300005 ()2-s2.0-105007283634 (Scopus ID)979-8-4007-1563-1 (ISBN)
Conference
18th European Workshop on Systems Security, Rotterdam, The Netherlands, 31 Marsch, 2025
Available from: 2025-07-14 Created: 2025-07-14 Last updated: 2025-10-01Bibliographically approved
Picazo-Sanchez, P. & Ortiz-Martin, L. (2024). Analysing the impact of ChatGPT in research. Applied intelligence (Boston), 54(5), 4172-4188
Open this publication in new window or tab >>Analysing the impact of ChatGPT in research
2024 (English)In: Applied intelligence (Boston), ISSN 0924-669X, E-ISSN 1573-7497, Vol. 54, no 5, p. 4172-4188Article in journal (Refereed) Published
Abstract [en]

Large Language Models (LLMs) are a type of machine learning that handles a wide range of Natural Language Processing (NLP) scenarios. Recently, in December 2022, a company called OpenAI released ChatGPT, a tool that, within a few months, became the most representative example of LLMs, automatically generating unique and coherent text on many topics, summarising and rewriting it, or even translating it to other languages. ChatGPT originated some controversy in academia since students can generate unique text for writing assessments being sometimes extremely difficult to distinguish whether it comes from ChatGPT or a person. In research, some journals specifically banned ChatGPT in scientific papers. However, when used correctly, it becomes a powerful tool to rewrite, for instance, scientific papers and, thus, deliver researchers’ messages in a better way. In this paper, we conduct an empirical study of the impact of ChatGPT in research. We downloaded the abstract of over 45,000 papers from over 300 journals from Dec 2022 and Feb 2023 belonging to different research editorials. We use four of the most known ChatGPT detection tools and conclude that ChatGPT played a role in around 10% of the papers published in every editorial, showing that authors from different fields have rapidly adopted such a tool in their research. © The Author(s) 2024.

Place, publisher, year, edition, pages
New York, NY: Springer, 2024
Keywords
ChatGPT, Large language models (LLMs), Natural language processing (NLP), Research
National Category
Natural Language Processing
Identifiers
urn:nbn:se:hh:diva-53179 (URN)10.1007/s10489-024-05298-0 (DOI)001190059700001 ()2-s2.0-85188165139 (Scopus ID)
Funder
Halmstad University
Note

Funding: Open access funding provided by Halmstad University.

Available from: 2024-04-12 Created: 2024-04-12 Last updated: 2025-10-01Bibliographically approved
Olsson, E., Eriksson, B., Picazo-Sanchez, P., Andersson, L. & Sabelfeld, A. (2024). FakeX: A Framework for Detecting Fake Reviews of Browser Extensions. In: ASIA CCS '24: Proceedings of the 19th ACM Asia Conference on Computer and Communications Security. Paper presented at 19th ACM Asia Conference on Computer and Communications Security, AsiaCCS 2024, Singapore, Singapore, 1-5 July, , 2024 (pp. 769-784). New York, NY: Association for Computing Machinery (ACM)
Open this publication in new window or tab >>FakeX: A Framework for Detecting Fake Reviews of Browser Extensions
Show others...
2024 (English)In: ASIA CCS '24: Proceedings of the 19th ACM Asia Conference on Computer and Communications Security, New York, NY: Association for Computing Machinery (ACM), 2024, p. 769-784Conference paper, Published paper (Refereed)
Abstract [en]

Browser extensions boost user experience on the web. Similarly to smartphone app stores, browsers like Chrome distribute browser extensions via their Web Store, enabling a thriving market of third-party developed extensions. The Web Store incorporates a user review system to help users decide which extensions to install. Unfortunately, the open nature of the review system is subject to reputation manipulation. As browser vendors fight reputation manipulation, attackers employ more sophisticated methods to stay under the radar. Focusing on fake reviews, we identify several techniques attackers use: fake accounts, disjoint sets of fake accounts for different extensions, automation of generated reviews, and focusing on reviews rather than ratings. We present FakeX, a framework to detect fake reviews by focusing on inference from review metadata. FakeX employs five distinct methods, including temporal distribution analysis, relationship clustering, and ratio-based assessments, to unveil patterns indicative of fake reviews. Evaluation of over 1.7 million reviews reveals the effectiveness of FakeX in identifying hundreds of fake review campaigns. Furthermore, our investigation of these fake reviews uncovers 86 malicious extensions, mounting attacks that range from data-stealing to monetization, impacting over 64 million users. In addition, we collaborate with Adblock Plus and Avast to demonstrate FakeX in action, expanding a seed list of newly detected malicious extensions to discover a further 16 malicious extensions with millions of users, where, in some cases, attackers tried to improve malicious code. © 2024 Copyright held by the owner/author(s).

Place, publisher, year, edition, pages
New York, NY: Association for Computing Machinery (ACM), 2024
Keywords
Browser Extensions, Fake Reviews, Web Security
National Category
Computer Sciences
Identifiers
urn:nbn:se:hh:diva-54373 (URN)10.1145/3634737.3656999 (DOI)2-s2.0-85199257289 (Scopus ID)9798400704826 (ISBN)
Conference
19th ACM Asia Conference on Computer and Communications Security, AsiaCCS 2024, Singapore, Singapore, 1-5 July, , 2024
Available from: 2024-08-05 Created: 2024-08-05 Last updated: 2025-10-01Bibliographically approved
Palmqvist, O., Jiremalm, J. & Picazo-Sanchez, P. (2024). Policy Parsing Panthers at Touché: Ideology and Power Identification in Parliamentary Debates. In: Guglielmo Faggioli; Nicola Ferro; Petra Galuščáková; Alba García Seco de Herrera (Ed.), CLEF 2024 Working Notes: Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2024): Grenoble, France, 9-12 September, 2024. Paper presented at 25th Working Notes of the Conference and Labs of the Evaluation Forum, CLEF 2024, Grenoble, France, 9-12 September, 2024 (pp. 3451-3477). Aachen: RWTH Aachen University, 3740
Open this publication in new window or tab >>Policy Parsing Panthers at Touché: Ideology and Power Identification in Parliamentary Debates
2024 (English)In: CLEF 2024 Working Notes: Working Notes of the Conference and Labs of the Evaluation Forum (CLEF 2024): Grenoble, France, 9-12 September, 2024 / [ed] Guglielmo Faggioli; Nicola Ferro; Petra Galuščáková; Alba García Seco de Herrera, Aachen: RWTH Aachen University , 2024, Vol. 3740, p. 3451-3477Conference paper, Published paper (Refereed)
Abstract [en]

Political debates are vital in shaping public opinion and influencing policy decisions. However, understanding the complex linguistic structures used by politicians to ascertain their orientations and power dynamics can be challenging. In this paper we explore Natural Language Processing techniques for identifying political orientation and power structures in parliamentary debates. We introduce a Located Missing Labels-loss in order to train jointly to predict both power and ideology. Furthermore, our proposed method also trains to predict a third synthetically generated polarity label. Finally, we combine this training method with pre-processing steps including back-translation and meta data inclusion. Our results show that our method manages to improve upon conventional methods of fine-tuning. © 2024 Copyright for this paper by its authors.

Place, publisher, year, edition, pages
Aachen: RWTH Aachen University, 2024
Series
CEUR Workshop Proceedings, E-ISSN 1613-0073 ; 3740
Keywords
CLEF, NLP, Political Debates, Touché
National Category
Natural Language Processing
Identifiers
urn:nbn:se:hh:diva-54563 (URN)2-s2.0-85201616600 (Scopus ID)
Conference
25th Working Notes of the Conference and Labs of the Evaluation Forum, CLEF 2024, Grenoble, France, 9-12 September, 2024
Available from: 2024-09-05 Created: 2024-09-05 Last updated: 2025-10-01Bibliographically approved
Arshad, H., Picazo-Sanchez, P., Johansen, C. & Schneider, G. (2023). Attribute-based encryption with enforceable obligations. Journal of Cryptographic Engineering, 13(3), 343-371
Open this publication in new window or tab >>Attribute-based encryption with enforceable obligations
2023 (English)In: Journal of Cryptographic Engineering, ISSN 2190-8516, Vol. 13, no 3, p. 343-371Article in journal (Refereed) Published
Abstract [en]

Attribute-based encryption (ABE) is a cryptographic mechanism that provides fine-grained access control to encrypted data, which can thus be stored in, e.g., public clouds. However, ABE schemes lack the notion of obligations, which is common in attribute-based access control systems such as eXtensible Access Control Markup Language and Usage Control. Obligations are used to define and enforce extra constraints that happen before approving or denying an access request. In this paper, we propose OB-ABE, a system for extending any classical ABE with enforceable obligations. Our system architecture has as core component trusted hardware enclaves, implemented with SGX, used for enforcing obligations. We employ ProVerif to formally model OB-ABE and verify its main property called “enforceable obligations,” i.e., if a message is encrypted along with an obligation, then the message can be decrypted only after enforcing the attached obligation. OB-ABE has two more properties: (i) OB-ABE is a “conservative extension” of the underlying ABE scheme, preserving its security properties; (ii) OB-ABE is “backward compatible” in the sense that any ciphertext produced by an ABE scheme can be decrypted by its extended OB-ABE version, and moreover, a ciphertext produced by an OB-ABE scheme can be decrypted by its underlying ABE scheme provided that the ciphertext does not have obligations attached. We also implement in C using Intel SGX a prototype of an OB-ABE extending the well-known ciphertext-policy ABE. © 2023, The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature.

Place, publisher, year, edition, pages
Heidelberg: Springer, 2023
Keywords
Attribute-based encryption, Enforceable obligations, Intel SGX, Security, Trusted hardware enclaves
National Category
Computer Systems
Identifiers
urn:nbn:se:hh:diva-51967 (URN)10.1007/s13389-023-00317-1 (DOI)000978486800001 ()2-s2.0-85153764879 (Scopus ID)
Funder
Swedish Foundation for Strategic ResearchSwedish Research Council
Available from: 2023-11-13 Created: 2023-11-13 Last updated: 2025-10-01Bibliographically approved
Riera, F. B., Almgren, M., Picazo-Sanchez, P. & Rohner, C. (2023). Clipaha: A Scheme to Perform Password Stretching on the Client. In: Paolo Mori; Gabriele Lenzini; Steven Furnell (Ed.), Proceedings of the 9th International Conference on Information Systems Security and Privacy - ICISSP: . Paper presented at 9th International Conference on Information Systems Security and Privacy, ICISSP 2023, Lisbon, Portugal, 22-24 February, 2023 (pp. 58-69). Science and Technology Publications, Lda
Open this publication in new window or tab >>Clipaha: A Scheme to Perform Password Stretching on the Client
2023 (English)In: Proceedings of the 9th International Conference on Information Systems Security and Privacy - ICISSP / [ed] Paolo Mori; Gabriele Lenzini; Steven Furnell, Science and Technology Publications, Lda , 2023, p. 58-69Conference paper, Published paper (Refereed)
Abstract [en]

Password security relies heavily on the choice of password by the user but also on the one-way hash functions used to protect stored passwords. To compensate for the increased computing power of attackers, modern password hash functions like Argon2, have been made more complex in terms of computational power and memory requirements. Nowadays, the computation of such hash functions is performed usually by the server (or authenticator) instead of the client. Therefore, constrained Internet of Things devices cannot use such functions when authenticating users. Additionally, the load of computing such functions may expose servers to denial of service attacks. In this work, we discuss client-side hashing as an alternative. We propose Clipaha, a client-side hashing scheme that allows using high-security password hashing even on highly constrained server devices. Clipaha is robust to a broader range of attacks compared to previous work and covers important and complex usage scenarios. Our evaluation discusses critical aspects involved in client-side hashing. We also provide an implementation of Clipaha in the form of a web library1 and benchmark the library on different systems to understand its mixed JavaScript and WebAssembly approach’s limitations. Benchmarks show that our library is 50% faster than similar libraries and can run on some devices where previous work fails. © 2023 by SCITEPRESS – Science and Technology Publications, Lda. Under CC license (CC BY-NC-ND 4.0).

Place, publisher, year, edition, pages
Science and Technology Publications, Lda, 2023
Series
ICISSP, ISSN 2184-4356
Keywords
Argon2, IoT Security, Password Stretching, Password-based Authentication, Server Relief, Web Security
National Category
Computer Sciences
Identifiers
urn:nbn:se:hh:diva-52116 (URN)10.5220/0011653200003405 (DOI)2-s2.0-85176332335 (Scopus ID)9789897586248 (ISBN)
Conference
9th International Conference on Information Systems Security and Privacy, ICISSP 2023, Lisbon, Portugal, 22-24 February, 2023
Available from: 2023-11-24 Created: 2023-11-24 Last updated: 2025-10-01Bibliographically approved
Andersson, T. & Picazo-Sanchez, P. (2023). Closing the Gap: Automated Distractor Generation in Japanese Language Testing. Education Sciences, 13(12), Article ID 1203.
Open this publication in new window or tab >>Closing the Gap: Automated Distractor Generation in Japanese Language Testing
2023 (English)In: Education Sciences, E-ISSN 2227-7102, Vol. 13, no 12, article id 1203Article in journal (Refereed) Published
Abstract [en]

Recent advances in natural language processing have increased interest in automatic question generation, particularly in education (e.g., math, biology, law, medicine, and languages) due to its efficiency in assessing comprehension. Specifically, multiple-choice questions have become popular, especially in standardized language proficiency tests. However, manually creating high-quality tests is time-consuming and challenging. Distractor generation, a critical aspect of multiple-choice question creation, is often overlooked, yet it plays a crucial role in test quality. Generating appropriate distractors requires ensuring they are incorrect but related to the correct answer (semantically or contextually), are grammatically correct, and of similar length to the target word. While various languages have seen research in automatic distractor generation, Japanese has received limited attention. This paper addresses this gap by automatically generating cloze tests, including distractors, for Japanese language proficiency tests, evaluating the generated questions’ quality, difficulty, and preferred distractor types, and comparing them to human-made questions through automatic and manual evaluations. © 2023 by the authors.

Place, publisher, year, edition, pages
Basel: MDPI, 2023
Keywords
cloze tests, education, Japanese, NLP
National Category
General Language Studies and Linguistics
Identifiers
urn:nbn:se:hh:diva-52405 (URN)10.3390/educsci13121203 (DOI)001131157300001 ()2-s2.0-85180644354 (Scopus ID)
Available from: 2024-01-15 Created: 2024-01-15 Last updated: 2025-10-01Bibliographically approved
Organisations
Identifiers
ORCID iD: ORCID iD iconorcid.org/0000-0002-0303-3858

Search in DiVA

Show all publications