LUTPub (LUT University)
Not a member yet
24632 research outputs found
Sort by
Malicious infiltration in open source projects and methods of prevention
Get PDFThe thesis focuses on how malicious actors gain access to open source projects in order to inject malicious code, factors that make some projects more susceptible to these types of attacks than others, and steps that can be taken to mitigate the risk of successful social engineering-based malicious infiltration attacks.
The research was conducted through a multivocal literature review and multiple case studies of real world incidents including XZ Utils, event-stream, OpenJS, and ESLint.
The findings include both technical solutions, such as reproducible builds, usage of SCA tools and secure authentication, as well as human resource factors that are critical to the security of a project, such as addressing maintainer burnout.
The thesis concludes that both technical measures as well as addressing the human element, particularly through increased support for maintainers, are essential to mitigating the risks of an open source project being successfully attacked
Tree-of-thoughts kehotesuunnittelu viitekehys tilinpäätösanalyysille
No full textLarge Language Models are increasingly capable of complex tasks and advanced reasoning. Use cases have been suggested for many high-stakes fields, such as law, medicine and finance. However, deploying these models for real world analytical tasks is still difficult and risky, as there are many considerations for model performance and result evaluation. Tasks also vary in context attributes, as for example legal documents often consist mainly of text data. Therefore, domain specificity is an important consideration, for example with business analytics tasks, where the context often consists mainly of structured numerical data.
This research aims to develop a prompting framework specifically for business analytics domain, and it uses a set of financial statements data from 110 Finnish SMEs to test the analytical consistency and performance of the OpenAI o3 model with a series of prompts compiled according to that framework. The results indicate that at best the analytical performance of the current models could be useful for real-world uses with well-developed prompt for the specific task. However, the results clearly show that human evaluation of the results cannot be bypassed when using the models.
This research emerged from the desire to develop low-threshold ways to adopt LLMs for recurrent and clearly structured analytics tasks, and to develop similarly low-threshold ways to evaluate and validate the model performance. The results of the thesis form a decent basis for such purposes.Kielimallit kykenevät yhä monimutkaisempiin tehtäviin ja kehittyneempään päättelyyn. Harkittuja käyttökohteita löytyy monilta merkittävien seurausten aloilta, kuten juridiikasta, lääketieteestä ja rahoituksesta. Mallien käyttöönotto oikean elämän analytiikkatehtävissä on kuitenkin yhä vaikeaa ja riskialtista, koska niiden suorituskykyyn ja tulosten arviointiin liittyy merkittäviä haasteita. Tehtävien datakontekstit vaihtelevat, esimerkiksi oikeudelliset asiakirjat ovat usein pääosin tekstiä, kun taas liiketoiminta-analytiikassa konteksti koostuu pääasiassa taulukoidusta numeerisesta datasta. Toimivat sovellutukset ovat siis hyvin alakohtaisia.
Tämä tutkimus pyrkii kehittämään nimenomaan liiketoiminta-analytiikkaan suunnitellun viitekehyksen kehotteille. Viitekehystä testattiin 110 suomalaisen pk-yrityksen tilinpäätösaineistolla, arvioiden OpenAI o3-mallin analyyttistä johdonmukaisuutta ja suorituskykyä sarjalla kyseisen viitekehyksen mukaisesti laadittuja kehotteita. Tulokset osoittavat, että parhaimmillaan nykyiset mallit voivat suorittaa oikeita analyysejä hyödynnettävällä tasolla, kun kehotteet on huolellisesti räätälöity kyseiseen tehtävään. Samalla tulokset kuitenkin korostavat, ettei ihmisen tekemää arviointia voida sivuuttaa kielimalleja käytettäessä.
Tutkimus sai alkunsa halusta kehittää matalan kynnyksen tapoja ottaa kielimalleja käyttöön toistuvissa ja selkeästi jäsennellyissä analytiikkatehtävissä sekä luoda vastaavasti matalan kynnyksen menetelmiä mallien suorituskyvyn arviointiin ja todentamiseen. Tutkimuksen tulokset muodostavat kohtuullisen lähtökohdan näihin tarkoituksiin
Effective and efficient data collection in designer workflows for environmental certifications in building design
Get PDFThis thesis investigates how environmental certification systems—specifically BREEAM, LEED, and RTS—affect the mechanical, electrical, and plumbing (MEP) design process, with a particular focus on the effectiveness of certification requirements and the efficiency of data collection. The research was conducted in collaboration with the Granlund company and LUT University and combines qualitative methods, including three expert interviews, project document analysis, and participation in a certification-focused industry workshop.
The study explores three core research questions: how information collection related to certifications can be organized efficiently; how certification requirements can be integrated to support effective design and implementation; and how designers can minimize repetitive investigative work and share learning across projects. Findings highlight that while international systems like BREEAM and LEED are comprehensive, they often present usability challenges due to fragmented evidence requirements, unclear responsibilities, and limited compatibility with Finnish design practices. In contrast, the RTS certification system offers clearer requirements, stronger alignment with local design documentation, and a more efficient evidence collection process.
The thesis concludes with targeted recommendations for design firms, certification coordinators, and system developers. These include improving early-phase planning, developing internal evidence templates, and enhancing knowledge transfer between projects. Overall, the research emphasizes that environmental certification can support sustainable building goals more effectively when integrated into design workflows, aligned with local practices, and supported by structured organizational learning
Power electronic converters as source of virtual inertia for future power grids
Get PDFThe increased inverter-based generation and the growing penetration of renewable energy sources into modern power grids have significantly reduced the natural inertia of the system, making frequency stability a critical challenge. This thesis investigates two control methodologies for frequency support in low-inertia grids: inverter-based virtual inertia and reserve-based virtual inertia. The grid-connected synchronous generator was modeled in MATLAB Simulink to represent the core of the power system, and an additional current injection unit was developed to represent renewable sources of solar, BESS, etc. A dedicated control unit was designed in accordance with Fingrid specifications to emulate virtual inertia and deliver structured frequency response using FCR-N, FCR-D and aFRR.
Inverter-based control is a direct approach for stabilizing the system frequency, while the reserve-based approach strictly follows Fingrid specifications and activates a particular reserve based on the frequency band. Each control strategy is simulated under different case scenarios and varied grid-inertia coefficients. The results highlight the strengths and limitations of each individual control method, emphasizing their roles in enhancing frequency stability in low inertia grids
Sector coupling, energy systems and 100% renewable energy in Eastern Africa : a transition to 2050
No full textThis study presents a comprehensive analysis of the potential of Eastern Africa to transition to a 100% renewable energy system by 2050. Using high-resolution scenario modelling, five distinct pathways, including three Best Policy Scenarios, a Delayed Policy Scenario, and a Current Policy Scenario, were evaluated for their techno-economic feasibility. Results highlight that ambitious adoption of solar PV and wind power, combined with deep electrification and sector coupling, can achieve complete defossilisation with lower long-term costs and zero CO₂ emissions by mid-century. Storage technologies and power-to-X applications play critical roles in enabling system flexibility and deep defossilisation of the hard-to-abate energy-industry segments. Despite challenges including financing gaps and limited regional integration, proactive policies that incentivise investment, support decentralised solutions, and promote regional cooperation can unlock Eastern Africa’s vast renewable energy potential, fostering sustainable development, and climate resilience
Agent-based approaches in social prototyping and software development : synthesizing core elements for implementation and gaps of evaluation methods
Get PDFThis thesis synthesizes the core architectural elements of agent-based modelling in social prototyping and AI systems for software development by studying the implementation architectures and the possible gaps that existing evaluation methods for the implementation of agent-based modelling should address in overcoming the criticisms. Publications and public information on implementation architectures, evaluation methods and implementation criticisms were studied, and common patterns were synthesized to come up with theories on core architectural elements and evaluation gaps. Both fields have two architectural elements in common – agent and software scaffold. Agent has a promoted LLM and an intelligence part in software development systems and agent-based modelling, respectively. The software scaffolding supports efficient agent operations. Software scaffolds are outside the agents in software development systems, both inside and outside the agents in agent-based modelling. The evaluation methods in agent-based modelling do not address the efficiency of change or intervention implementation in simulated societies, and the correctness of the theory on which the simulation is based. The findings of this study are synthesized from recurring patterns that provide hypotheses which can be empirically validated in further research
Kiertotalouden lähtöinen kriittisten raaka-aineiden talteenotto käytetystä litiumioniakuista sähkökemiallisella- ja avustetulla liuotuksella
No full textThis thesis investigates the structural composition and recycling potential of lithium-ion batteries (LIBs), with a particular focus on the electrochemical leaching (ECL) of critical raw materials (CRMs) from end-of-life (EoL) LIBs. The work is divided into two main sections: the theoretical section offers a comprehensive review of LIB chemistry (e.g., LiCoO2, LiFePO4, LiMn2O4, NMC), component architecture, and operational mechanisms, emphasising the roles of cathode, anode, electrolyte, separator, and binder materials. It highlights the European Union’s circular economy (CE) framework, including targets and sector-specific challenges, and discusses the current status of LIB recycling, incentives, and risks associated with CRMs. The thesis also summarizes existing LIB recycling techniques and commercial processes worldwide, detailing their underlying principles.
The experimental section focuses on disassembly, characterisation, and leaching of the active materials black mass (BM) derived from end-of-life laptop spent LIBs. The BM was milled and sieved into five particle fractions, and ICP-MS and SEM-EDS analysis identified that the <75 μm fraction had the highest cobalt (38.9 wt%) and lithium (5.77 wt%). Three chemical leaching (CL) and two ECL tests were performed. CL achieved up to 89.3% Li and 37.3% Co leaching efficiencies, whereas ECL yielded 72.3% Li and 33.0% Co. These results indicate that higher acid concentrations and temperatures predominantly enhance Li and Co dissolution, whereas electrochemically-assisted leaching (EAL) improves Mn and Ni extraction. Overall, the findings underscore the promise of small-scale ECL reactors for selective CRM recovery and highlight the need for design optimization — alternative electrode materials, flow configurations, and pre-treatments — to enhance efficiency and scalability for sustainable LIB recycling.Tämä diplomityö tutkii litiumioniakkujen (LIB) rakenteellista koostumusta ja kierrätyspotentiaalia, erityisesti elinkaarensa päässä olevista (EoL) poistuvista litiumioniakuista peräisin olevien kriittisten raaka-aineiden (CRM) sähkökemiallista liuotusta (ECL). Työ on jaettu kahteen pääosaan: teoreettisessa osiossa tarjotaan kattava katsaus litiumioniakkujen kemiaan (esim. LiCoO2, LiFePO4, LiMn2O4, NMC), komponenttien arkkitehtuuriin ja toimintamekanismeihin, korostaen katodin, anodin, elektrolyytin, erotuskalvon ja sideaineiden rooleja. Siinä tuodaan esiin Euroopan unionin kiertotalouskehyksen (CE) rakenne, mukaan lukien tavoitteet ja toimialakohtaiset haasteet, sekä käsitellään litiumioniakkujen kierrätyksen nykytilaa, kannustimia ja CRM liittyviä riskejä. Diplomityö tiivistää myös olemassa olevat litiumioniakkujen kierrätystekniikat ja kaupalliset prosessit maailmanlaajuisesti, kuvaillen niiden taustalla olevia periaatteita.
Kokeellisessa osassa keskitytään litiumioniakkujen aktiivimateriaalien mustan massan (BM) purkamiseen, karakterisointiin ja liuotukseen. BM on peräisin käytöstä poistuneista kannettavien tietokoneiden litiumioniakuista. BM jauhettiin ja seulottiin viiteen raekokojaksoon. ICP-MS- ja SEM-EDS-analyysien perusteella havaittiin, että alle 75 μm:n fraktiossa havaittiin korkein Co pitoisuus (38,9 painoprosenttia) ja Li (5,77 painoprosenttia) pitoisuudet. Kokeellisessa työssä suoritettiin kolme kemiallista liuotuskokeilua (CL) ja kaksi ECL. CL kokeissa Li liotustehokkuus ylsi enimmillään 89,3 %:iin ja Co 37,3 %:iin, kun taas ECL:ssä vastaavat luvut olivat 72,3 % Li:lle ja 33,0 % Co:lle. Tulokset osoittavat, että korkeammat happopitoisuudet ja lämpötilat edistävät erityisesti Li ja Co liukenemista, kun taas sähkökemiallisesti avustettu liotus (EAL) tehostaa Mn ja Ni talteenottoa. Kaiken kaikkiaan tulokset korostavat pienten ECL-reaktoreiden lupaavuutta CRM selektiivisessä talteenotossa ja osoittavat tarpeen optimoida reaktorin suunnittelua — hyödyntämällä vaihtoehtoisia elektrodimateriaaleja, erilaisia virtausasetelmia ja esikäsittelyjä — litiumioniakkujen kierrätyksen tehokkuuden ja skaalautuvuuden parantamiseksi
Assessing the health impact of indoor air purification using health data obtained from Apple Watch wearers
No full textPM2.5 poses significant health risks. While indoor air purification is a proven mitigation approach, most studies have used medical devices with limited usability and practicality. Thus, we used Apple Watch health metrics to assess short-term physiological responses in a randomized, double-blind, crossover real-world indoor purification intervention – first of its kind in the UAE. 41 healthy, non-smoking young adults underwent two 2-week interventions (real and sham filters), separated by a 1-week washout. Collected Apple Watch metrics include: heart rate (HR), heart rate variability (HRV), (daily) resting heart rate (RHR), sleep respiratory rate (RR), and blood oxygen saturation (SpO₂). Daily timelocation surveys enabled personal PM2.5 exposure estimation. Real purification reduced home PM2.5 exposure from 8.1 μg/m3 to 1.7 μg/m3, associated with significant changes in HR (–0.68%), sleep RR (–0.25%), and SpO₂ (+0.10%). When accounting for exposure across all locations, an IQR reduction in PM2.5 exposure was significantly associated with changes in HR (–1.02%), RHR (–1.07%), sleep RR (–0.39%), and HRV (+1.86%). SpO₂ showed a near-significant increase (+0.04%, p = 0.110). Our findings supported the use of indoor air purification to reduce PM2.5 exposure, linked to health metrics improvements, and the feasibility of using the Apple Watch in environmental interventions
The Role of Actor’s Self-efficacy in AI Utilization
No full textThe aim of this study was to explore the role of self-efficacy in the adoption and utilization of AI technologies within organizational settings through a metasummary approach. By analyzing thematic findings across multiple studies, the research identified three key themes: catalyst for AI adoption, attitudes and emotions, and performance. The findings revealed that self-efficacy plays a significant role in promoting AI adoption, fostering positive emotional responses, and motivating learning behaviors, although certain areas, such as creativity and decision-making, were less frequently mentioned. From a theoretical perspective, the study confirmed that self-efficacy enhances Human-Computer Interaction (HCI) and AI utilization. Practically, the results suggest that organizations should focus on improving employees’ self-efficacy through training and development to achieve better AI integration. Limitations include the reliance on existing literature and the lack of qualitative insights into certain areas. Future research should address these gaps by examining underexplored aspects of self-efficacy in AI environments to optimize both HCI and AI utilizations.Post-print / Final draf
The usability influence of user navigation paths regarding the integration of artificial intelligence in software platforms : an investigation through the lens of sustainability
Get PDFThe prevalence of artificial intelligence (AI) has seen a dramatic rise in recent years, with its integration being implemented in a variety of software systems. Human computer interaction research, centred around the potential usability related influences that the integration of artificial intelligence has on software systems exists on a broad level. Despite this, a gap has been identified regarding the integration of guidance-oriented AI. This refers to AI whose goal is to help in guiding and assisting the user via the relevant steps to complete some task, rather than primarily providing them with a solution, or results. Therefore, it is an AI agent whose main focus is to ensure the user remains on the intended navigation path. Furthermore, another increasingly important and prevalent field is that of sustainability, with various individuals and organizations striving to be aware of their sustainability impacts. This research aimed to not only contribute to the understanding of the influence that the integration of such an AI has on a software system, but to also bridge the gap between these influences and the potential corresponding sustainability impacts.
To evaluate the usability influences, a guidance-oriented AI was created for this study. Existing literature was thoroughly examined to understand what integration aspects are the most impactful from a usability perspective, regarding guidance-oriented AI. These findings were used to create an AI chatbot that focused on assisting users by guiding and directing the path they took within some system. This AI was integrated, and specifically trained to assist users on a third-party site, which was made possible via a research-industry partnership. Two sets of usability tests, being a baseline and benchmark group, were conducted with different participants. This included providing participants with set tasks that they were to complete under observation, followed by oral interviews and question periods. Both qualitative data (such as perceived feelings of frustration, trust, and usefulness for example) were collected, as well as quantitative data (such as task completion time).
It was found that the integration of guidance and advisory oriented AI had a positive influence on the perceived system usability, with the perceived efficiency, satisfaction, usefulness, reliability, accuracy, and the rate of successful task completion improving. This came with a trade-off of perceived system trust and transparency decreasing, as well as the time spent on tasks lengthening. Furthermore, despite participants indicating they felt they required less help overall with the AI’s integration, in reality more assistance was requested.
These findings were mapped to the defined sustainability dimensions, which found that there was a strong perceived positive influence on the technical dimension, with the potential for negative perceived consequences regarding both the social and environmental dimensions. The environmental dimension findings are a strong avenue for future research, due to the complex relationship numerous variables could have on the results. The economical dimension was determined to be case dependent, with both positive and negative influences being present depending on the requirements and goals set. Lastly, a complex relationship was identified regarding the potential influence on the individual dimension, with no clear conclusion being present. It should be noted that the majority of identified influences were mapped to the individual and technical dimensions. Therefore, the conclusions drawn for these dimensions are the strongest