User feedback plays a significant role in helping recommendation systems to make personalized and accurate predictions. Despite the fact that many methods of collecting user feedback have been proposed, little research exists that addresses both the breadth and depth of data collected. In this study, we incorporate personal reflection into traditional crowdsourcing tasks and investigate how it facilitates people's reflection on their usage of YouTube. We present a novel crowdsourcing approach with personal reflection integration based on several design principles, which allows participants to reflect on contextual factors and personal values of using YouTube through guided context recall and evaluations, therefore gathering deeper insights into people's preferences and behaviors on a large scale. We conducted a user study involving 20 participants and explored the insights generated from their textual answers and the role of design principles in the reflection process. This approach successfully enables multiple participants to conduct the study simultaneously, thereby reflecting on their watching behaviors and preferences. Based on the quantitative and qualitative analysis of the findings, we sum up the implications of this approach to provide guidance for the YouTube recommendation system and point to the directions for the design of similar studies in the future.

This thesis investigates the potential of solving large-scale linear systems of equations (LSEs) in the Tensor Train (TT) format. First, we study when this approach can outperform traditional solvers like Conjugate Gradient (CG) and Gaussian Elimination (GE). In turn, we examine the time complexity of the TT-solve technique when applied to ill-conditioned, sparse, and symmetric positive definite (SPD) matrices. This enables us to assess the scenarios where TT-solve may offer significant advantages.
During this exploration, we have also identified and proposed solutions for a handful of research gaps that could further optimize the TT-solve process. These include challenges related to poorly factorizing matrix sizes, as well as the need for a deeper understanding of the trade-offs between TT-matrix (TTM) rank and maximal mode size during the decomposition process. Additionally, the study examined the potential for altering the coefficient matrix through techniques like variable reordering and partial Gaussian elimination (PG) to achieve lower TTM-ranks without affecting the solution of the LSE.

Cluster analysis in high dimensional data is a difficult but desirable task. Many existing methods fail to cluster high dimensional data due to what is known as the curse of dimensionality. Therefore, sophisticated clustering methods are in wide development. Along these lines, spectral modularity maximization emerged from the theory of random matrices and graph modularity. The method is based on a filtering of the spectral decomposition of similarity matrices. Despite the recent success of this method, we uncover a fundamental challenge of spectral modularity: the spectral modularity breaks down as the number of groups in a data set grows. To mitigate this challenge, we propose two solutions: one solution based on a regularization and one solution based on a normalization. We perform a thorough empirical analysis of the clustering performance of the solutions and find that, not only do our methods resolve the breakdown of spectral modularity, but they also outperform existing clustering methods in
a variety of settings.

To assist smokers in potentially quitting their habit, this paper investigates digital eHealth applications. Based on a dataset provided by research into an eHealth application, it aims to determine if persuasive activities can convince users of the usefulness of competencies determined to assist in quitting smoking. A thematic analysis was used on the participant's qualitative responses to the activities. Using this and the quantitative data derived, insights into the efficacy of persuasion were determined.
It was determined that engagement with the optional qualitative aspect of the data produced similar utility perspectives on the competencies to those who did not comment. It was noted that the general perspective of the competencies rose after completing the activity, however not to a significant degree. Additionally, no notable correlations between age, gender or educational level and increased perception of the competency arose. Several interesting remarks from participants were analysed to offer considerations for any future research in this field.

Smoking has been one of the great threats to health in recent years, being strongly correlated with multiple negative health consequences, including lung cancer. Recent research suggests that artificial intelligence chatbots can be effective in persuading healthy behavior change. However, these chatbots usually rely on persuasive techniques to achieve their goal. Such techniques depend on identifying and meeting users' needs to be effective. To help improve understanding of the domain of healthy behavior change, we proposed a study which analysed the needs of daily smokers as they emerged from their interactions with a chatbot specifically designed to help them quit. The study performed a thematic analysis on users' free-text feedback, from which a set of 8 themes that directly correspond to 8 different needs were observed. The user needs were correlated with their genders, ages and highest completed education levels. While most of the results indicate that there is no significant correlation between the needs and user characteristics, which suggests that user needs are evenly distributed, certain correlations were highlighted for further analysis.

Unhealthy behaviors such as smoking is the major cause for premature deaths and changing behaviors by oneself can be difficult. That is where eHealth applications come into rescue. One of the recent research explored the possibility of using a Reinforcement Learning model to choose persuasive types for a virtual coach to adopt to persuade people to prepare for smoke-quitting and it has shown advantages. However, there are still more aspects to investigate in this context except the persuasive types of the messages, and this paper intended to further look into using reinforcement learning to choose activities for preparing the users to quit smoking, To be more specific, we implemented and evaluated a reinforcement learning model to choose activities to optimize both the effort spent by the users and also the likelihood of them staying for the next session. The result suggests that reinforcement learning is a promising approach to choose activities for people to prepare for quitting smoking and it can move the users to states that they are more likely to spend a good effort on the activities and are more likely to come back to the next session.

Contract review is a critical yet time-consuming process in legal practice, with significant financial implications when errors occur. While Large Language Models (LLMs) have shown promise in legal document processing, they still face challenges with lengthy contracts and complex legal relationships. This research presents an advanced approach to automated contract review by integrating knowledge graphs into Retrieval-Augmented Generation (RAG) frameworks, addressing the limitations of current methodologies.

Through a comprehensive literature review of contract review automation and RAG systems, we conducted systematic experiments comparing RAG approaches with LLMs' in-context learning capabilities. Our empirical analysis validated that RAG-based methods significantly enhance long-context text analysis and information extraction in legal documents, particularly in terms of accuracy and consistency.

Building on these findings, we extensively investigated optimization techniques for the RAG retrieval phase, recognizing its critical role in contract review accuracy. Our experimental evaluation encompassed various chunking strategies, query expansion methods, and re-ranking approaches, establishing best practices for legal document processing.

Our primary contribution is a novel KG-RAG system that enhances contextual understanding in legal document analysis. We evaluate our approach using the Contract Understanding Atticus Dataset (CUAD) and ContractNLI dataset, demonstrating improved performance over traditional RAG implementations and long-context models. The research also explores optimal chunking strategies and investigates the efficiency-effectiveness trade-offs between different model architectures.

Results indicate that our KG-enhanced RAG framework achieves superior performance in identifying and analyzing complex legal relationships while maintaining computational efficiency. The integration of knowledge graphs particularly excels in capturing hierarchical and cross-referential relationships within legal documents, a crucial aspect often overlooked by conventional approaches.

This work advances the field of legal AI by providing a more robust and context-aware approach to contract review, while offering practical insights for implementing AI systems in legal practice. Our findings suggest promising directions for future research in legal document processing, particularly in areas requiring deep contextual understanding and relationship modeling.

Immersive technologies, such as head-mounted devices for Virtual Reality (VR), have experienced increased availability on the consumer market. This opens up numerous possibilities for scientific research and industrial applications. But, configuring and combining data from various sources to create an effective visualization to explore, collaborate in and interact with, has only been researched in limited aspects. This thesis aims to demonstrate the feasibility of integrating these diverse aspects into a single application while preserving their individual potential.

To achieve this objective, an extensive analysis of the literature was done to establish the state of the art. All aspects necessary to create a single application were investigated to define how much of the literature fully or partly implemented them. Findings of the literature were summarized through definition, progress, limitations and improvements.

Based on the analysis, a proof of concept was designed and implemented, which established a virtual reality framework for real-time analysis and exploration of heterogeneous urban data. The framework involved the development of both a client and server applications with real-time communication. The server application processes and caches input data based on a structured configuration file, while the client application renders and visualizes points and shapes based on instructions from the server. The layered processing approach in this proof of concept facilitated real-time data handling and exploration. Which also makes it possible to collaborate with multiple people working on or manipulating the same data at the same time.

Evaluation of the server application showed its capability of processing over 1.000.000 objects from GeoJSON data. After initial rendering, the client application could update over 4.000 objects with a latency of just 200ms, which is considered to be interactive. Feedback from experts further supported the validation of the proof of concept application as a viable solution to the research question.

This research established the validity of a configurable real-time analysis and exploration solution in virtual reality. With further development and research, this tool holds potential for being used in diverse environments dealing with urban data.

This thesis aims to assess robustness of networks by evaluating the performance of node attack strategies, the applicability and accuracy of different approaches, and to increase robustness of networks through analysing protecting methods, including link addition and node protection strategies. To be specific, the relative size of the Largest Connected Component (rLCC) and Average Two-Terminal Reliability (ATTR) are chosen to be the performance metrics to evaluate the robustness of networks. In the assessment of network robustness, simulations are employed for ten distinct attack strategies, which include random attack, non-updated and updated degree attacks, non-updated and updated stochastic degree attacks, non-updated and updated betweenness attacks, and greedy attack. Three analytical approximation methods are implemented and discussed, alongside the testing of a machine learning-based approach. Regarding the enhancement of network robustness, the comparative analysis involves thirteen link addition strategies and four node protection strategies. In all node removal scenarios, it is observed that the updated betweenness attack emerges as the most harmful strategy. Moreover, analytical approximations are proved to be effective means of evaluating network robustness in the scenario of predicting robustness for synthetic networks under random attack. Machine learning based methodologies show high accuracy in predicting robustness of synthetic networks, exhibiting acceptable error rates in predicting robustness of real-world networks. Besides, the updated betweenness link addition strategy and targeted betweenness-based node protection exhibit the highest efficacy in protecting networks against the most harmful attacks.

In real-world scenarios, users provide invaluable data; however, this data is inherently incoherent, incomplete, and duplicated, i.e., different data rows refer to the same real-world object. Merging duplications to a single entry broadens the knowledge of a given real-world object represented within a data set. Applying a straightforward cross-join operation to check for duplications is infeasible and intractable in big data scenarios. Instead of comparing all rows, the similarity can be approximated by Locality-Sensitive Hashing (LSH). LSH’s MinHash creates colliding hashes on common tokens. With enough matching hashes, two rows can be deemed similar. The applications of LSH within an Entity Resolution (ER) pipeline on incoherent and incomplete big data are thoroughly investigated in this work. An ER pipeline is introduced to make the deduplication problem tractable. This pipeline consists of four phases: preprocessing, blocking, matching, and clustering. In the preprocessing, the data is made coherent. In the blocking, the data is divided into smaller main blocks by blocking on complete properties. These blocks contain a majority of True-Matches. The created blocks can be augmented with additional blocks that rely on incomplete properties. In the matching phase, only the Matches found within the blocks are compared, significantly reducing the required number of comparisons. The incomplete properties cannot be transformed into features for every Match; therefore, classifiers are trained per property combination, allowing for optimal Match classification. In clustering, the duplicated rows are merged into a single Entity. These Entities are created with disconnected sub-graphs of Positive-Matches. These sub-graphs are particularly susceptible to noise. Therefore, novel implementations of MinHash are used for noise removal and finding missed True-Matches. The created Entities were tested on an existingmodel. Blocks constructed with MinHash on N-gram tokens allow spelling mistakes to be overcome within the user data, which improves the Pairs Completeness (PC) at the cost of Pairs Quality (PQ) and Reduction Ratio (RR). Introducing augmentative blocks increases the PC significantly at the cost of PQ and a low cost of RR. Multiple runs are investigated throughout the pipeline. The runs relying on MinHash showed increased PC at the cost of PQ and RR. A novel concept of a double-pass pipeline is introduced, in which the baseline ER pipeline should be initially applied, followed by a more expensive error-correctiveMinHash ER pipeline that can be used on Entities created by the first pass, combining the strengths of the investigated runs.

This thesis’ research concerns the time-dynamics of a complex geographical network of municipalities, i.e. the Dutch Municipality Network over the period 1830-2019. By analysing 190 years of socioeconomic statistical data and applying contemporary tooling from network science and geographic information systems (GIS), the findings from this research can provide a new approach and supportive methods for policymakers, statistical offices, researchers and businesses (to decide when and where to invest).

Multiple Instance Learning (MIL) is a type of semi-supervised machine learning used recently in medical and multi-media fields. In MIL, instead of a single feature vector, a set of feature vectors has to be classified. Standard MIL algorithms assume that only some of these vectors are useful for building a classifier. This paper extends the standard MIL assumption by combining propositional logic and classical MIL classifiers. Adding propositional logic allows for increased interpretability as it establishes an if-then relationship between the input data and the output classes. This combination of logic and classical MIL classifiers will be called Concept Rule Induction (CRI). CRI is tested on several artificial and real-life bird song data. CRI is shown to work for these data sets, and the rules produced by propositional logic can be interpreted

Suicide prevention is an important global topic, since suicide is being a worldwide serious health issue for decades. In the meantime, people start to pay more attention to mental health issues and Artificial Intelligence is developed and applied in many different fields. Recently, there are many studies been done on using Conversational Agents for mental illness. There are also applications of preventing suicide with Conversational Agents. In the Netherlands, 10.4 suicides happened per 100 thousand inhabitants in 2019. 113 Suicide Prevention is a Dutch national organization with a mission of pursing zero suicides. They are providing multiple services to help people with suicidal thoughts, including a self-test for evaluating suicidal risks anonymously. Even though, people who have taken the self-test receive some suggestions from the 113 websites, they might not contact the 113 helpline for further help. However, it is significant for people with suicide ideation to disclose their thoughts in order to reduce their intention of committing suicide. Therefore, a conversational agent was designed and developed to upgrade the suicide self-test service, which was aiming to resolve test-takers barriers towards contacting the helpline for human help. With the service, the conversational agent interacts with users to help them take the self-test and persuade them to reach for further help by relieving their worriers. A double-blind mixed designed experiment was conducted with 147 healthy general participants and two conditions (traditional questionnaire and interactive service) to verify the hypotheses and evaluate the service. It was hypothesised that participants would have more motivation to contact a helpline for further professional help after taking the self-test in an interactive way than taking the self-test in a traditional questionnaire way. The results showed a significant increase in the traditional questionnaire condition comparing to the interactive service condition. For participants' satisfaction and perceived usefulness, these are higher in the traditional questionnaire condition than the ones in the interactive service condition. In conclusion, the proposed interactive self-test service did not show an expected potential of increasing people's motivation to contact a helpline for professional help. However, the traditional self-test with a textual barrier intervention section showed the potential to motivate people’s help-seeking behavior comparing to the proposed interactive self-test service based on the experiment.

TOPdesk is a service management software provider in a wide variety of domains and industries. TOPdesk also offers consultancy to their customers that aims to continuously assess and improve the customer’s experience and service efficiency. TOPdesk offers a Mini Health Check (MHC) to their customers in which aconsultant analyzes how efficiently the customer uses their software based on six Key Performance Indicators (KPI). However, the process of creating an MHC report is very time-consuming as it requires performing a lot of manual steps. Also, the norms used for the KPIs provide little meaning as they are arbitrarily chosen and not specific to the customer’s industry. This report aims to improve the current process of performing an MHC. Research has been done on how the MHC is performed, identifying the suitable technologies and learning the currently existing infrastructure that helped us pave the way to create our product. During our project we managed to create a product that automates the MHC. Through user testing we found that this process now takes about two minutes, where the manual process took about two hours. To create more meaningful norms for the KPIs, we also implemented a benchmarking feature. This allows a company to compare the results of their MHC to other TOPdesk customers in the same sector, country or of similar size. We have some recommendations for TOPdesk for the further development of our product. The MHC process could be streamlined in a few ways, most importantly with respect to the process for getting access to customer data. Benchmarking could become even more useful if data can be more easily gathered from more TOPdesk customers.

Parkinson's disease (PD) diagnosis is based on clinical criteria, i.e. bradykinesia, rest tremor, rigidity, etc. Assessment of the severity of PD symptoms, however, is subject to inter-rater variability. In this paper, we propose a deep learning based automatic PD diagnosis method using videos recorded during the assessment with the Movement Disorders Society - Unified PD rating scale (MDS-UPDRS) part III. Seven tasks from the MDS-UPDRS III are investigated, which show the symptoms of bradykinesia and postural tremors. We demonstrate the effectiveness of automatic classification of PD severity using 3D Convolutional Neural Network (CNN) and the PD severity classification can benefit from non-medical datasets for transfer learning. We further design a temporal self-attention (TSA) model to focus on the subtle temporal vision changes in our PD video dataset. The temporal relative self-attention-based 3D CNN classifier gives promising classification results on task-level videos. We also propose a task-assembling method to predict the patient-level severity through stacking classifiers. We show the effectiveness of TSA and task-assembling method on our PD video dataset empirically.

The M.O.R.S.E. system is a tool for creating and managing large escape events, mainly used for local escape events. The tool is designed for only a limited range of puzzle types and styling options because most of the puzzles require physical items in order to solve a puzzle and only the answers have to be entered in M.O.R.S.E. Because of this design, it is really difficult to create online escape experiences, especially rich and immersive ones. It also requires a lot of programming outside of the M.O.R.S.E. system to do so. Raccoon Serious Games , the client, does not have many employees with programming experience and, therefore, it is not feasible for them to create the rich and immersive online escape experiences they want. To be able to create such immersive experiences, we are extending M.O.R.S.E. with editable domains and web pages. Game designers can add domains and web pages to the existing event schedule and then puzzles can be created for web pages. Players can view one or multiple of these domains and for each domain, the active web page will be served. Web pages can be created and stored in the domains, but the actual contents of the web pages still have to be made. Because making web pages is often a programming intensive task, a page builder has been created in M.O.R.S.E. This page builder allows the user to load and save web pages created in the M.O.R.S.E.
system. It uses a drag-and-drop system to place building-block elements inside the web pages and allows for directly visible styling of those elements. Because of this, the user does not need programming knowledge of the underlying implementation of the web pages. It also facilitates the linking between M.O.R.S.E. features and the domains such as puzzles and triggers for buttons. Using the import and export functionality, users can easily copy previous web pages created with the page builder. This is not only limited to internal web pages but can also be used to import external code from outside the page builder. With user-friendly features such as the ability to undo and redo changes, the page builder tries to make creating web pages as easy as possible. An important aspect of the escape games hosted by Raccoon Serious Games is team building. We extend upon this notion by adding roles and a leaderboard screen to M.O.R.S.E., both of which increase the need and opportunity for interaction between players. The addition of roles allows game designers to enforce cooperation in their escape events, by restricting the access to resources required for solving a puzzle to only a subset of the players in a team. This way they have to cooperate and combine their information and resources to solve all puzzles. The addition of leaderboards is also an extra incentive for a player in a team to work together efficiently because this will positively impact their score and, therefore, ranking on the leaderboard.

ScenWise is an innovative company that specializes in data science revolving around traffic management. ScenWise strives to use the newest and best technologies and practices when it comes to web applications, data science and traffic management. The reason for this is that they provide tools to analyse and visualise a variety of situations that occur in traffic management. One such tool is SmartRoads 1.0, which allows users to analyse traffic data and situations via a web application. Unfortunately SmartRoads 1.0 does not perform as desired. Additionally, ScenWise itself has the problem of not being able to integrate previously made products by student groups into their own existing products. During the research aimed to resolve these problems another issue arose; the software development life cycle of ScenWise is very lacking. Research on the SmartRoads 1.0 performance problem showed that the bottleneck of its performance is due to the front-end. The outdated SmartRoads 1.0 front-end was thus replaced with a new and better SmartRoads 2.0 front-end. The integration problem and development life cycle problem are both addressed in the Longterm evolution (LTE) design found in appendix I. This LTE design contains the architecture migration plan. This plan will transform the current software architecture to a Service-oriented architecture (SOA) providing a solution for the current integration problems. A result of the first steps of this architecture migration plan is the Application Programming Interface (API) Gateway, which has been implemented in the aforementioned SmartRoads 2.0. Next to the migration plan, guidelines for ScenWise to improve their software development life cycle are elaborated in the LTE design. In this report the identified problems, their solutions and executions are explained, discussed and evaluated.

Due to increasing pressure on the system, the functioning of public transport networks (PTN) in metropolitan areas is crucial for future mobility. Within these networks, hierarchical levels can be distinguished where different levels have different functions. These hierarchical levels can be analyzed but there is no way to quantitatively determine the hierarchy in a PTN. Therefore, this paper presents a metric to quantify the hierarchy in PTN to increase the understanding of these complex networks. In order to determine the hierarchy, a metric is developed based on a combined topological and empirical approach. The metric is a multiplication of three different elements which are the topological influence, (non-)redundancy and transfer potential. Together these elements are applied to define the hierarchical degree of nodes in a network. Furthermore, to determine the hierarchy of a network as a whole, a hierarchical coefficient, based on the distribution and inequality of the hierarchical degree in the network is developed. The metric has been applied to case-studies for the Dutch cities of Amsterdam and Rotterdam which allows for different state and cross-network comparison. The results show some expected yet non-trivial results identifying different patterns in network structures for network states and different spatial distribution of hierarchy between networks. Furthermore, by dividing the network into functional levels, a hierarchical structure can be identified.
Throughout this study, a new method to quantify hierarchy in PTN, based on different approaches, is developed which can be seen as the most important contribution of this research. While this study explores the implications of this metric, it can be applied in numerous different contexts. Furthermore, in potential the metric has numerous network related applications such decreasing vulnerability and solving bottlenecks.

Authentication mechanisms play an important role in everyday digital interactions and allow users to prove their identity to others. Privacy-preserving Attribute-Based Credential Systems (PABCSs) allow users to authenticate by presenting their credential attributes, while multiple presentations remain unlinkable and untraceable. A revocation mechanism allows the credential issuer to revoke a credential, for example when the attributes of a user change. Verifiers can in turn determine the revocation status of the presented credential. This thesis considers situations in which a verifier needs to determine the revocation status of a credential after the presentation. This stronger revocation level, also described as forward-looking consistency, has not been researched before in PABCSs. To address this gap, we present the PABC-FLC design, a PABCS with forward-looking consistency. Our design allows users to remain offline after presentation, without compromising on the privacy properties. To avoid an impossibility result, we introduce a new participant called the Non-Revocation Prover (NRP). The NRP facilitates a verifier in determining the revocation status of a credential, even though neither the verifier nor the issuer have to trust the NRP. We show that our PABC-FLC design has a comparable experimental runtime to a credential system without revocation mechanism. This experiment shows our design has a runtime overhead of +20 ms (+32%) during presentation and +20 ms (+34%) during verification. Concluding that our PABC-FLC design is a feasible PABCS with forward-looking consistency, while remaining unlinkable and untraceable.

Intrusion detection problem in Industrial Control Systems(ICS), such as water treat- ment plant and power grid, is an important real-world problem. Real-time anomaly detection have been proposed to minimize the risk of cyber attack. In this study, two different kind of intrusion detection mode-based approach are learned from normal behaviour of an ICS, SWaT. One model is inspired by TABOR, a offline graphical model-based approach of CPS intrusion detection. Timed automaton, Bayesian net- work and Out of Alphabet are combined to find anomalies and localize the abnormal sensors and actuators. The other model proposed is based on two time slice Bayesian Network(2-TBN) with the motivation simplifying multi-model approach into a single model one. In this way, cost of computing power and time is reduced in order to meet the requirement of real-time operation testing.
Experimental results demonstrate the model’s close performance to TABOR and slight advantage on time. The study about difference between offline testing and real-time operation is another topic in this study. The generic underlying idea and experience is also applicable to cyber physical system in other industrial control systems.