This thesis investigates whether motion data from IMU embedded drinking cups can provide movement cues for multimodal analysis of natural conversation. Using a manually annotated subset of the COSI-Lab conference living lab dataset, ten five-minute participant recordings are aligned between overhead video and a glass-mounted three-axis accelerometer sampled at 50 Hz. Video observations are used to define and annotate five pilot movement categories: Still, Gesture, Drinking, Toasting, and Nodding. The annotations are mapped to a dense accelerometer timeline and used to compare three complementary modelling families: conventional window based ML/DL classifiers, supervised inertial temporal action localization models, and LLM based localization from numerical prompt representations.

The results show that smart cup motion contains useful movement information, but reliable temporal localization remains difficult under natural interaction conditions. Window based baselines reveal local class-discriminative information but are limited by fixed boundaries and severe class imbalance. Supervised temporal action localization provides the most stable sample-level performance, whereas few-shot LLM prompting can sometimes identify useful candidate intervals but remains sensitive to reference construction and recording variation.

The findings support a cautious role for smart cup IMU data as an auxiliary cue for locating movement-related intervals in long multimodal recordings, rather than as independent evidence of conversational meaning. Future work should refine the behaviour category system, expand annotations, and integrate cup motion with video, speech, body keypoints, and interaction context. ... Read more

In inclusive classrooms, the physical presence of children with hearing loss (HL) does not guarantee social inclusion with their typically hearing (TH) peers. Self-report is biased and manual observation is too labour-intensive to cover a full school day at fine temporal resolution. This study adapts a multimodal sensing pipeline that combines ultra-wideband (UWB) spatial tracking with Language Environment Analysis (LENA) audio recordings, applied in a single inclusive preschool with 13 children (6 HL, 7 TH).

A Gaussian Mixture Model fit on four room-level features (adult word count, auditory overlap, displacement, and teacher distance) and selected by a stability-aware rule (lowest mean BIC among K values whose cluster assignments reproduce across random initialisations, mean pairwise Adjusted Rand Index >= 0.80) over 2 to 20 components recovers six latent activity contexts, each with a distinct sensor profile. Each context is then given a post-hoc descriptive label drawn from activity types familiar in inclusive preschool classrooms: dispersed transition, peer-driven activity, independent / parallel work, adult-scaffolded peer activity, seated guided work, and whole-class instruction / read-aloud. These labels describe the recovered clusters and are not validated against an external ground truth.

Within each context, a linear mixed model with a per-child random intercept compares HL and TH children on three sensor-derivable behavioural markers of inclusion: peer co-presence (time in spatial groups, where a spatial group is a set of children simultaneously co-present and mutually body-oriented; §5.2.3), vocal participation rate (utterances per minute), and peer affiliation patterns (time in same-diagnosis vs. mixed-diagnosis spatial groups).

The HL/TH signal distributes differently across contexts for each marker. Peer co-presence shows no substantial HL/TH difference in three of six contexts. HL children exceed TH children in independent / parallel work (+7.8% of the minute, q < 0.0001, d = +1.59) and fall below TH children in peer-driven activity (-4.3% of the minute, q = 0.02) and seated guided work (-5.3% of the minute, q = 0.002). Vocal participation rate shows an HL deficit concentrated in the contexts that combine high peer communicative demand with reduced adult scaffolding (peer-driven activity q = 0.03 and adult-scaffolded peer activity q = 0.05, both significant after false-discovery-rate correction; dispersed transition in the same direction but only marginal, q = 0.07), but not in the three contexts where high adult word count or very low auditory overlap slows the pace of exchange. Peer affiliation is the most consistent asymmetry: TH children concentrate grouped time in TH-only spatial groups more than HL children concentrate theirs in HL-only spatial groups in five of six contexts (significantly in four), and HL children spend more grouped time in mixed spatial groups than TH children do across all six (significantly in four). The 6:7 cohort composition produces a baseline TH-above-HL gap of approximately 0.08 on the homophily index under random affiliation; the four significant clusters exceed this baseline (observed gap 0.12-0.19, of which 0.04-0.11 is preference beyond availability), while the two non-significant clusters are within the magnitude expected from cohort composition alone. ... Read more

The recent surge in machine learning (ML) research has led to a record number of paper submissions, overwhelming the traditional peer-review process. Although conferences like NeurIPS have introduced reproducibility checklists to maintain scientific standards, manual verification of these claims is time-consuming and inconsistent. This study investigates the feasibility of using Large Language Models (LLMs) to automate the evaluation of paper reproducibility. By creating a ground-truth dataset through the manual annotation of NeurIPS papers, this study assesses the accuracy of LLMs in verifying author claims regarding code availability, hyperparameter transparency, and compute resources. The results compare LLM performance with manual labels to identify where automated tools succeed and where they fail to capture technical nuances. Ultimately, this research demonstrates that while LLMs can act as highly efficient administrative filters to streamline initial screening, they fail to reliably predict execution viability, highlighting the remaining boundaries of automated verification. ... Read more

The rapid growth of Machine Learning research has overwhelmed traditional peer-review systems, leading to concerns regarding reviewer fatigue and the consistency of scientific evaluation. While Large Language Models (LLMs) are being explored as potential assistants for quality assessment, their ability to objectively verify specific scientific criteria—such as those in the NeurIPS Paper Checklist—remains unproven. This checklist serves as a structured self-auditing framework that mandates authors to explicitly declare critical details, including potential negative societal impacts, exact hyperparameter tuning ranges, and clear definitions of theoretical assumptions or limitations. This study investigates the core question: “How well can an LLM extract the limitations described in scientific papers?” Using a manually annotated dataset of 78 papers, this research evaluates the accuracy of LLMs in extracting limitations stated by authors. Our findings reveal that while the LLM achieves perfect accuracy (100\%) in detecting the presence of dedicated limitation sections, its performance in textual extraction is more nuanced. For explicit limitations, the model demonstrates high recall (0.91) but moderate precision (0.71), frequently over-extracting context. Furthermore, when tasked with extracting implicit limitations from papers lacking dedicated sections, both recall (0.71) and precision (0.69) decline. Notably, we found that a major bottleneck in unstructured text is getting the LLM to look at the actual weakness instead of getting distracted by subsequent sentences talking about future work. By comparing LLM performance against a human-verified ground truth, this work provides a feasibility study for automating high-stakes research quality assessments and identifies current bottlenecks in LLM reasoning for scientific auditing. ... Read more

As machine learning conferences such as NeurIPS expand rapidly, the manual classi-
fication and evaluation of responsible research checklists impose a significant burden on
reviewers. This study investigates the ability of Large Language Models (LLMs) to au-
tomatically classify research papers as empirical, theoretical, or hybrid, and to extract
checklist compliance data. Using a dataset of publicly available NeurIPS papers, we
designed an automated pipeline and evaluated its outputs against a human-annotated
ground truth. Our results demonstrate that the LLM achieves high accuracy in the
core classification task, reliably distinguishing the papers core methodology by iden-
tifying clear structural indicators like mathematical proofs and benchmark datasets.
Furthermore, the model excels at extracting objective checklist elements, performing
well on close-ended extraction tasks that rely on clear structural indicators. However,
performance noticeably decreased on structurally scattered or subjective criteria, such
as broader impacts and the declaration of AI usage. This drop highlights a limitation in
the model’s broader reading comprehension, as it struggles to merge contextual infor-
mation without explicit headers. Notably, this automated failure closely mirrors human
task ambiguity, as these exact subjective items also generated the lower inter-annotator
agreement among human annotators. Conclusively, while LLMs provide a highly con-
sistent baseline for classifying paper typologies and extracting explicit methodological
data, their reliance on structural cues indicates they should serve as assistive screening
tools rather than autonomous evaluators in academic peer review. ... Read more

Scientific peer review is an important part of the scientific process. However, the growing number of submissions has sparked interest in automated review tools. Recent work has shown that Large Language Models (LLMs) can generate reviews and evaluate author-provided checklists, yet it is unclear to what extent they can independently identify the scientific claims that are made in papers and perform structured reviews. This thesis investigates whether an LLM can automatically extract scientific claims from research papers in the machine learning field and then complete the NeurIPS Checklist without relying on author-written justifications. The evaluation focuses on claim extraction accuracy, preserving the semantic meaning of claims, and agreement between LLM-generated checklist annotations and human judgment. Gemini 3 Flash's claim extraction and checklist annotations are compared against human ground-truth annotations on NeurIPS 2024 papers. The results show that the model successfully identifies primary claims of papers, with a recall of 0.99 and precision of 0.75. Most errors are caused by over-segmentation or incorrect classification. For checklist annotation, the system achieves a mean accuracy of 0.85 and a mean Cohen's Kappa of 0.58 compared to human annotations. Agreement is strongest for objective checklist criteria. These findings indicate that LLMs can effectively support claim-based scientific review, but are not advanced enough to fully replace expert reviewers. ... Read more

Results show that combined dual-sensor features consistently outperform single-sensor variants, and that coarser grids yield higher exact accuracy while mean physical error remains stable across resolutions at approximately 70–80 cm. At 0.5 m resolution, the best configuration places 71% of predictions within 50 cm of the true location, approaching the accuracy of a UWB baseline system while requiring no installed infrastructure. Zone merging and hexagonal tessellation do not provide consistent improvements over the plain square grid, suggesting that magnetic ambiguity rather than data imbalance or cell geometry is the dominant source of error. The findings demonstrate that infrastructure-free magnetic fingerprinting is a practically viable approach for coarse spatial awareness in socially dynamic indoor environments. ... Read more

The design of data collection scenarios is critical for evaluating intelligent systems for social intention recognition in aviation. Identical aircraft behaviors can generate multiple equally plausible intention interpretations depending on situational context and the observer’s professional perspective, yet existing research offers limited guidance for constructing scenarios that preserve this interpretive open-endedness. This study addresses this gap through an exploratory, literature-based investigation of how contextual factors shape intention interpretation across aviation roles. An integrated framework combining the 3Cs model of situational analysis and script theory is proposed to identify contextual dimensions influencing interpretation. Through qualitative synthesis of aviation literature, the framework demonstrates how variations in cues, classes, characteristics, and internal-external script configurations can produce divergent but valid intention narratives for the same observable behavior. The resulting scenario-first methodology provides structured guidance for designing aviation scenarios that support role-dependent intention annotation and evaluate intelligent systems. As a conceptual contribution, the framework requires empirical validation by aviation professionals. ... Read more

Inferring social intention in everyday settings is challenging because the same observable behavior can support multiple plausible interpretations. This issue is pronounced in restaurants, where roles and norms structure interaction but do not uniquely determine what is socially “meant.” The research question addressed is: How can scenarios in a restaurant setting be created that allow investigation into how humans and intelligent systems construct multiple plausible narratives of social intention?A literature-grounded scenario design method is presented that separates observable cues from inferred psychological meanings and situation classes, and frames interactions using external scripts (norms, roles, scenes) and internal script variants (observer-dependent interpretations). Two short scenarios with three controlled variations each are specified to modulate openendedness through cue completeness, norm clarity, and perspective. It is concluded that openendedness can be designed systematically by controlling observability and introducing norm tensions and perspective differences that keep multiple narratives simultaneously plausible. ... Read more

Understanding how people move and interact within shared spaces is important for applications such as behaviour analysis, smart environments, and social interaction research. Existing human sensing approaches, including vision-based systems, wearables, and wireless sensing, can provide detailed information but often face limitations related to privacy, deployment complexity, cost, or scalability. Floor-based sensing, and smart carpets in particular, offer a privacy-preserving alternative by capturing interaction patterns through pressure measurements without recording visual or auditory data. However, many existing smart carpet systems remain constrained by fixed dimensions, high manufacturing costs, or limited scalability.

This thesis presents the design, implementation, and evaluation of a low-cost, modular smart carpet for human behaviour tracking. The proposed system uses a resistive matrix sensing approach based on Velostat and copper electrodes, combined with a tile-based architecture that enables flexible deployment over larger areas. A custom-designed printed circuit board centralises signal routing and data acquisition, reducing wiring complexity while avoiding the use of active electronics within individual tiles. To support scalability, the system separates real-time data acquisition from offline signal processing, where cross-talk mitigation, filtering, and visualisation are performed.

The system is evaluated through a series of qualitative and quantitative experiments that examine its ability to capture pressure distributions, detect multiple contact regions, and represent dynamic interactions such as weight shifting and walking. The results show that the system reliably captures coarse spatial interaction patterns, but that limitations in spatial resolution and material behaviour reduce its effectiveness for distinguishing subtle pressure differences. These findings indicate that, while the system provides useful information for activity and behaviour analysis in shared spaces, its sensing capabilities are less expressive than initially anticipated for fine-grained interpretation.

Overall, this work explores a design point that prioritises affordability, modularity, and privacy, and demonstrates how these priorities shape the trade-offs observed in large-area, pressure-based human sensing systems. ... Read more

Intelligent systems are being increasingly deployed alongside humans, yet their ability to understand human intentions in order to perform tasks better is far from perfect. In this paper a literature survey will be performed to obtain existing examples of narratives, with and without intelligent systems,that take place in a Hospital - a setting with a variation in situation openness, where a framework of cues, characteristics, classes, internal scripts, external scripts, and memory and associative triggers all play an equally important role in determining an individual’s intention. Dimensions were extract from these scenarios and were analysed based on the three roles: Patients, Clinicians, and Family Members. Results and findings were used to develop a new case study involving an intelligent system assisting a doctor with diagnosing a patient and communicating it to them. Variations of contextual cues, amid other dimensions, were made to demonstrate multiple plausible narratives and its effects on intention. ... Read more

Intelligent systems in autonomous driving increasingly require the ability to infer social intentions to ensure safe and fluid interactions with human road users. However, current approaches typically frame this problem as objective trajectory prediction or fixed classification, ignoring the open-ended nature of human interpretation where a single physical behaviour can generate multiple plausible narratives. To address the gap between trajectory forecasting and narrative understanding, this research investigates how to systematically map the dimensions of variation in driving situations to the range of intention narratives they generate. A literature survey was conducted to distinguish between foundational human social norms and current algorithmic approaches. By integrating script theory with the 3Cs framework (Cues, Characteristics, Classes), this study developed a dimension extraction framework to analyse where objective observations diverge into subjective interpretations. Through comparative analysis of prototypical scenarios (lane merging and pedestrian negotiation), results revealed that current intelligent systems operate predominantly in geometric space, optimizing for physical feasibility, whereas human drivers operate in social space governed by normative scripts. The research concludes that narrative open-endedness is inversely proportional to the strength of physical and social constraints. That is, when constraints are weak, human internal scripts diverge from machine logic, leading to critical prediction errors. Consequently, future systems must shift from raw trajectory output to semantic narrative understanding to explicitly model this uncertainty and align machine reasoning with human expectations. ... Read more

We propose a method for detecting laughter in spontaneous social interactions using chest-worn accelerometers. Our approach compares three segmentation strategies—padded, centered, different sliding win- dowssizesandevaluatesannotationmodalities: No Audio, Only Audio, and With Audio. Using time-domain features and Random Forests, we reach up to 0.962 macro F1-score. Longer windows and multimodal annotations improve performance and generalizability. Key features include axis-wise means, deriva- tives, and inter-axis correlations. These results support the potential of motion-based laughter detection in privacy-sensitive environments, while highlighting the importance of segment and label design. ... Read more

Consumer-grade fitness trackers can produce unreliable physiological data due to sensor errors. The same holds for cycling data from Wahoo Fitness, where heart rate (HR) and power readings are essential for training and performance analysis. This thesis presents a prediction-based anomaly detection framework tailored to multivariate time-series cycling data. The approach reframes anomaly detection as a personalized physiological HR prediction problem. We define anomalies as deviations between measured sensor values and their predicted values, based on contextual activity metrics (e.g., power, cadence, speed, altitude, and gradient) and user-specific embeddings. The system combines ordinary differential equations (ODEs) modeling heart rate dynamics with machine learning techniques to capture non-linear, non-stationary, and individualized relationships. The model not only detects implausible values but reconstructs them with physiologically consistent alternatives. Compared to reconstruction-based methods, which are mostly used for anomaly detection in time series data, this physiologically grounded approach better differentiates between normal variation and true anomalies. Experimental results demonstrate effective identification and correction of HR and power anomalies, contributing to improved data quality and reliability in wearable fitness applications. ... Read more

This study investigates how genre preferences and sentiment influence fanfiction popularity across multiple languages, focusing on English, Mandarin, Russian, and Spanish datasets. Leveraging advanced natural language processing techniques, including multilingual sentiment analysis, genre classification, and topic modeling, this research explores the interplay between cultural and linguistic factors in storytelling. Preprocessing steps, such as translation and named entity recognition, ensured consistency and reduced noise across the multilingual dataset. Key findings reveal cross-linguistic patterns, such as the popularity of genres like Alter- nate Universe and Romance, alongside cultural distinctions in sentiment and engagement. This work contributes to computational fan studies by demonstrating how linguistic and cultural factors influence storytelling trends and audience preferences in fanfiction. ... Read more

Natural language processing, specifically style, is not explored significantly in the context of fan-fiction. By using function word frequency analysis, this paper explores the similarity in style between original works and fan-fictions derived from them as well as the impact of those stylistic similarities on the fan-fictions' successes. Investigating the works of Worm and Narnia and utilising a control set, this hypothesis is examined. Similarity in style seems to be variable, but it is suggested that certain stylistic features lead to success more than other regardless of the original work's style. More research would be needed to confirm those numbers. ... Read more

The introduction of Pretrained Language Models (PLMs) has revolutionised the field of Natural Language Processing (NLP) and paved the way for many new, exciting large-scale studies for various areas of research. One such field presents itself in the emerging digital literary corpus that is fanfiction, providing research opportunities within the fields of (NLP), Computational (Socio-) Linguistics, the Social Sciences and Digital Humanities. However, because of the unique linguistic characteristics of this literary domain many modern NLP solutions utilizing PLMs encounter difficulties when applied on fanfiction texts. This paper aims to indicate that the performance of various NLP tasks performed by PLMs on fanfiction texts can be improved by applying Domain Adaptive Pre-Training (DAPT) to PLMs. A case-study is performed to show that the performance of a BERT-based PLM can be improved for the downstream NLP task of Named Entity Recognition (NER) by applying supervised domain specific fine-tuning. While we gain a 6% increase in F1 score performance, we are sceptical about these results due to the limited amount of annotated data available leading to the model overfitting and show a lack of capacity to generalize to unseen data from the CoNLL NER dataset. ... Read more

This paper explores how the portrayal of female characters in fanfiction evolved in response to the #MeToo movement and fourth-wave feminism, with the aim of assessing whether the impact of the awareness of the campaign was broad enough to visibly alter how the average author portrays women in narrative contexts. To analyze these trends, fanfiction data from Archive of Our Own (AO3) spanning 2015–2019 was parsed, and two Natural Language Processing (NLP) pipelines — Word2Vec and GloVe, and BERT — were developed. The study finds that bias scores, aggregated through formulas created to compare gendered associations, show a stronger stereotypization of women before 2017 compared to after. Furthermore, a similar trend is discovered in the representation of women in fanfiction. While the BERT pipeline proved most effective for capturing contextual nuances, it is significantly limited by its reliance on binary labels and computational intensity. This further indicates the need for more inclusive and sustainable methods, making the Word2Vec/GloVe models more appropriate for this task. The paper concludes with recommendations for future work, including broader representation, longer-term analysis, and enhanced detection of evolving language patterns. ... Read more

Fanfiction writers always look for ways to make their stories more engaging. Analyzing what influences the popularity of fanfiction provides insights into readers' preferences and allows writers to tailor to these. This paper attempts to find linear correlations between fanfiction stories and the emotional journey of their characters. It does so by computationally extracting these journeys from 319 Good Omens fanfiction stories, defining and extracting several features from them and using simple linear regression to determine their correlation to fanfiction popularity. Five features were found to have a significant influence on fanfiction popularity. It was also determined that readers prefer characters that have low emotional fluctuations in their behavior. ... Read more

Laughter is a complex multimodal behavior and one of the most essential aspects of social interactions. Although previous research has used both auditory and facial cues for laughter detection, these approaches are commonly afflicted with difficulties in noisy, occluded, and privacy-sensitive settings. This paper explores the potential of using body posture alone—captured through 2D keypoint estimation as a robust signal for automatic laughter detection in naturalistic settings. We create a machine learning pipeline using the ConfLab dataset, which segments pose data, extracts motion-based features, and trains Random Forest classifiers on various annotation modalities (audio-only, video-only, and audiovisual) and segmentation methods (fixed and variable length). We show that, while variable-length segmentation yields optimal performance, it leads to overfitting. On the other hand, fixed-duration segmentation with three-second windows and audiovisual annotations achieves a pragmatic compromise and reaches F1-scores (65\%) comparable to earlier efforts in ideal environments. Upper-body movement, especially head and arm motion, is seen to be salient cues to laughter via feature importance analysis. Annotation modality is also found to significantly affect both classification performance and relative pose feature importance. These findings demonstrate the viability of pose-based laughter detection and reveal how annotation choices shape model behavior, offering insights for affective computing in the wild. ... Read more