Information Integration and Web Intelligence

In the era of AI, pandemics, and frequent disasters, informatics is no longer a matter of efficiency alone but of survival. We introduce Survival Informatics, a novel academic approach that emphasizes human well-being and societal resilience. Survival Informatics goes beyond traditional informatics by addressing fundamental issues of trust, inclusiveness, and sustainability, and by integrating ethical, legal, social, and economic perspectives into technical development. As one representative field where Survival Informatics can be practically implemented, the analysis of people’s reactions on social media provides a powerful means to capture public perceptions in real time and to link informatics research directly with societal well-being. Motivated by this perspective, we develop a stepwise technical framework that operationalizes the principles of Survival Informatics, highlighting a two-stage clustering approach for large-scale discourse analysis. Through a comprehensive case study of COVID-19 vaccine discourse in Japan, analyzing 32 million tweets, we demonstrate methodological innovations in scalability, reproducibility, and consistency. Finally, we discuss the broader implications of Survival Informatics for social implementation, including real-time public opinion monitoring, misinformation detection, and policy integration.

In this study, we propose a method to help understand products that don’t have many reviews. Specifically, we use an LLM to retrieve reviews of similar products. This helps users get more useful information when they are thinking of buying a product. First, when the user inputs a product and a question, the system uses the LLM to find the attributes that are related to the user’s question. Next, the system calculates the similarity of the attributes and find similar products. After that, by using an LLM, the system retrieves opinions from reviews of the similar products that are related to these attributes. Finally, the system ranks the opinions that were judged to be related to the attribute queries and shows them as the search results. This adds more useful information to support the target product. In the experiment, we compared the similarity between the opinions retrieved by the proposed method and the actual opinions that the target product has. As a result, opinions retrieved from similar products were slightly more similar to the actual opinions than those from randomly selected products.

In this study, we propose a method for generating a comparative table by summarizing product reviews using a Large Language Model (LLM). A comparative table summarizes, for each aspect of two products, the evaluations present in reviews and the number of reviews for each evaluation. By using an LLM to create the table, it becomes possible to generate the table on demand and to change the aspects to be compared depending on the products. The proposed method employs an LLM to extract aspects and their associated evaluations from reviews. These evaluations are summarized for each aspect to create a table mapping evaluations to their respective aspects. We used a review dataset from Rakuten Ichiba and automatically summarized reviews using an LLM to generate comparative tables. We conducted a user study to verify whether the tables are useful for comparing products. Based on the results of the user study, the proposed method was found to be more helpful for comparison than the baseline.

This paper proposes a method for retrieving diverse real-world user reviews that refer to a specific Aspect Query representing a user’s information need. Given a short Aspect Query, such as “practicality,” the system generates a variety of Sentence Fragment queries, e.g., “*able for da*” to retrieve phrases such as “suitable for daily use” or “comfortable for daytime work.” These Sentence Fragments act as wildcard-like queries and are particularly effective in languages like Japanese, where inflection and agglutinative structures make exact keyword matching challenging. To construct such fragments, we first use a large-scale language model (LLM) to generate a large number of synthetic Aspect Query–review sentence pairs. These pairs are filtered to retain only high-quality examples, which are subsequently used to fine-tune a lightweight local LLM. The fine-tuned model generates synthetic reviews for arbitrary Aspect Queries, from which Sentence Fragments that are frequent in the synthetic reviews but rare in general reviews are extracted and used as expanded queries. A user study on a real-world review dataset demonstrates that our method enables the retrieval of diverse reviews without compromising accuracy, effectively bridging the lexical gap between abstract Aspect Queries and concrete review expressions.

This paper proposes a document embedding method designed to obtain disentangled distributed representations. The resulting representations are expected to satisfy two key criteria: independence across dimensions and semantic interpretability of each dimension. We enhanced a classic shallow neural network-based embedding model with two modifications: 1) guidance task integration, where the network is trained to perform both a simple auxiliary metadata prediction task and a surrounding term prediction task simultaneously, and 2) loss regularization for independence, where the loss function includes both prediction accuracy and the independence across dimensions (i.e., the Kullback-Leibler divergence from a multivariate normal distribution). We evaluated the proposed method through both automatic and human-subject experiments using synthetic datasets and movie review texts. Experimental results show that even shallow neural networks can generate disentangled representations when dimensional independence is explicitly promoted.

Dajare, a form of Japanese pun, utilizes phonetically identical or similar words and phrases with different meanings to create a funny effect. This paper presents a method for generating interactive Dajare, a Japanese pun formed by a response that includes a segment phonetically similar to a part of the original utterance. Our approach involves retrieving candidate words and phrases based on phoneme similarity. We then leverage large language models (LLMs) to generate response sentences for these candidates, which are subsequently ranked according to conversational naturalness to select the most suitable interactive Dajare. Our experiment revealed that the proposed approach, by explicitly providing a phonetically similar segment, makes it easier for annotators to identify interactive Dajare compared to the baseline, which generates them without providing such a segment. This suggests that our method more effectively produces recognizable interactive Dajare. Additionally, the results showed that word selection positively influences the perceived cleverness (the ingenuity of transforming words with similar pronunciations) of interactive Dajare than on its funniness. Furthermore, current LLMs still fall significantly short of human capabilities in generating genuinely funny text.

This paper proposes a method for Japanese rhyme generation. A rhyme is defined as a pair of words with similar phonetic patterns and is widely used to enhance creative writing. To support creative writing, several rhyme search services are available on the web. However, these services typically rely on predefined word lists and search only for strict vowel-level matches. This approach limits their usefulness in creative applications. Therefore, this study proposes a Japanese rhyme generation method that supports not only strict vowel-level matches but also the generation of words outside predefined word lists. We use a GPT-2 model and control the token generation process to follow a given phoneme sequence, resulting in rhyme generation. The proposed method outperformed all other methods on 100 test inputs, achieving the best performance in both CER and mora similarity, and successfully generated rhymes for all test cases.

The complexity of mental healthcare billing enables anomalies, including fraud. While machine learning methods have been applied to anomaly detection, they often struggle with class imbalance, label scarcity, and complex sequential patterns. This study explores a hybrid deep learning approach combining Long Short-Term Memory (LSTM) networks and Transformers, with pseudo-labeling via Isolation Forests (iForest) and Autoencoders (AE). Prior work has not evaluated such hybrid models trained on pseudo-labeled data in the context of healthcare billing. The approach is evaluated on two real-world billing datasets related to mental healthcare. The iForest LSTM baseline achieves the highest recall (0.963) on declaration-level data. On the operation-level data, the hybrid iForest-based model achieves the highest recall (0.744), though at the cost of lower precision. These findings highlight the potential of combining pseudo-labeling with hybrid deep learning in complex, imbalanced anomaly detection settings.

The rapid adoption of electric vehicles (EVs) underscores the growing need for reliable battery health monitoring systems to ensure safety, optimize performance, and extend operational lifespan. In this paper, we introduce BATT2GRAPH, a novel approach that combines a temporal graph-based representation with a CNN-LSTM predictive model for accurate State-of-Health (SOH) estimation and anomaly detection in lithium-ion batteries (LIBs). On one hand, BATT2GRAPH constructs a temporal property graph using Neo4j to store enriched charge-discharge cycles with both raw time-series data and aggregated statistical indicators, enabling interpretable SOH monitoring and anomaly detection through expressive Cypher queries. On the other hand, a hybrid CNN-LSTM model is trained on this data to capture fine-grained variations and long-term degradation trends. Extensive experiments on the Stanford-MIT battery aging dataset demonstrate that our approach consistently outperforms existing baselines across multiple evaluation metrics.

Physiological sensing is essential for mental health monitoring, but models often degrade over time due to user behavior changes, sensor noise, and contextual variation. We propose CoRA (Continual and Regularized Adaptation), a lightweight continual learning framework that monitors latent feature drift using class-wise KL divergence and selectively retrains a downstream classifier with Elastic Weight Consolidation (EWC) to prevent forgetting. CoRA operates on top of a pretrained encoder, enabling efficient adaptation without storing raw past samples. In stress detection experiments on LifeSnaps, DAPPER, and WESAD, CoRA improves F1-score by up to 10.4% while reducing retraining overhead by over 40%, demonstrating a robust, personalized solution for real-world physiological monitoring.

Similarity search in high-dimensional spectral datasets is critical for applications in analytical chemistry, bioinformatics, and material science. Conventional methods often struggle with variability in peak positions, intensities, and noise, limiting their effectiveness for large-scale spectral comparison. In this paper, we propose a peak pattern based similarity search framework that abstracts spectra into robust peak representations and performs flexible, metric-based comparisons. The approach integrates preprocessing techniques such as noise filtering, normalization, and peak detection, followed by peak alignment with tunable tolerance N and matching threshold $$\delta $$ δ . Similarity is quantified using intensity-weighted metrics designed to accommodate spectral distortions and scaling variations. Experimental validation on real-world high-dimensional spectral datasets demonstrates that the framework achieves efficient and accurate retrieval of similar spectra. Parameter analysis highlights the impact of alignment tolerance and intensity weighting on retrieval performance, showing improved robustness against noise and peak shifts.

We propose error-range-guarantee approximate-aggregation methods called patch based encoding plus deterministic approximate querying (PBE+DAQ) and its extension, PBE+DAQ/WN (Wide-Narrow), which perform better than conventional DAQ by reducing the amount of data to be computed.With the increase in data volume, fast data analysis is required, and aggregation operations play an important role in data analysis. However, the larger the data volume, the longer time is required for aggregation operations. In many cases, while fast approximate-aggregation operations are required rather than accurate operations, the approximation error must be guaranteed. We use PBE for compressing the majority of data for faster aggregation operations and DAQ for error-guarantee approximation. We also developed cost models for the two methods as well as for conventional DAQ. We implemented the proposed methods and conducted experiments using real-world datasets. The experimental results indicate that the execution times of PBE+DAQ and PBE+DAQ/WN are 1.1x to 1.2x faster than that of DAQ while guaranteeing the error range of the aggregation results.

In this study, we propose a method for retrieving concrete reviews for a product, using an abstract review as a query. However, it is difficult to retrieve concrete reviews from keywords or abstract reviews, as simple sparse or dense retrieval methods cannot account for concreteness. Therefore, we propose a method called Query Rewriting with Retrieved Review Concretization (QR-ReReC), which utilizes retrieved reviews to rewrite the original query. For the experiment, we constructed a dataset to evaluate the effectiveness of QR-ReReC. The results showed that QR-ReReC is more effective for retrieving more concrete reviews than the retrieval methods without query rewriting and pseudo relevance feedback.

In this paper, we present a collection of 3,503 kanji with complete component hierarchies and representative images. The information is displayed as kanji cards to offer a versatile format that can be easily integrated into Web-based language learning applications. We will make the data publicly available in JSON format together with a simple Web application using templates to access and inspect the kanji cards. As two recent extensions we also introduce an iconic representation of the compositional layout and an evaluation environment focusing on handwritten kanji input by the learner.

Knowledge bases (KBs) are increasingly used in diverse knowledge-processing tasks. Widely used RDF-style KBs represent knowledge as subject – predicate – object triples, whereas many external sources are described in non-RDF formats. Therefore, applications often need to integrate KBs with such external sources. In our previous work, we proposed an integrated query environment named Knowledge Mediator (KM), in which external sources are accessed via SPARQL magic properties, letting users query them as if they were parts of the KB. KM performs entity linking (EL) to find corresponding KB entities for external objects. However, EL may yield multiple plausible entities, and the results may involve uncertainty. In this study, we propose a model that represents data integration while allowing for the expression of uncertainty in query results caused by EL. Furthermore, we propose an efficient query processing method that retrieves only those results whose likelihoods exceed a user-specified threshold.

Semantic Data Specifications (SDSs) define agreements for data exchange using semantic technologies. The authors of SDSs pick existing terms for reuse in their specific contexts, creating Application Profiles (APs). However, when the terms are reused directly, that is, without creating subclasses and subproperties, we run into a lack of machine-readable representation, especially when the AP editors still adjust labels, definitions, domains, and ranges. This leads to confusion, inefficiencies, and inconsistencies in the management and use of SDSs and APs. Based on our experience profiling SDSs in the data catalog domain, we present the Data Specification Vocabulary (DSV), our language for AP definition, supporting the specification of term reuse and refinement in SDSs. For improved machine readability of SDS metadata, DSV reuses the Profiles Vocabulary. We demonstrate DSV on DCAT, DCAT-AP and DSV itself, and we show that DSV can be used in software for SDS and AP management.

A transparent display allows the presentation of images and information while simultaneously transmitting background light. Unlike conventional displays, it enables users to obtain information while visually confirming their surroundings. For instance, a transparent display equipped with a speech transcription system allows users to read transcribed conversation content while observing the facial expressions and gestures of their conversation partners, thereby enhancing comprehension. However, because users frequently shift their visual focus between the background and the overlaid text, a time lag may occur in acquiring textual information, potentially hindering smooth communication. Therefore, it is essential to clarify which methods of presenting information remain easily recognizable even when out of focus. This study focuses on fonts, the fundamental components of textual information, to investigate which fonts are more easily recognized under defocused conditions. An experiment was conducted to evaluate the visibility of fonts on a transparent display at various focal distances, using two character types with differing visual complexity (Kanji and the Latin alphabet). The results revealed that the visibility of text on a transparent display is influenced by factors such as font type, degree of defocus, and character type. Notably, differences in font visibility became particularly pronounced under moderate defocus.

Linear regression is one of the widest used data analytics techniques in support of human decision-making. In this paper we discuss some theoretical foundations of confounder detection used in our web-based decision making platform GrandReport. In the context of linear regression, controlling confounding effects means to add further influencing, potentially confounding factors to the analysis. This paper exactly explains confounding effects in terms of the various involved coefficients by utilizing a conjecture on the noise-independent relationship between crude, adjusted, confounder and latent coefficients. We discuss, in how far our findings can improve the explainability of linear regression models as well as the maturity of their application in various contexts.

Organizations worldwide face significant challenges in translating privacy regulations into implementable technical requirements, creating a critical gap between legal privacy compliance and system development. This paper adapts KORA (Konkretisierung Rechtlicher Anforderungen - Concretization of Legal Requirements) methodology by incorporating established privacy patterns to systematically translate regulatory privacy requirements into applicable solutions. Applying this methodology, we examine Indonesia’s Personal Data Protection Law (UU-PDP) to propose technical solutions for privacy compliance. Our three-phase methodology systematically identifies regulatory requirements, maps them to established privacy objectives, including transparency, manageability, and intervenability, and connects them to implementable privacy patterns. Through rigorous analysis of the 76 articles in the UU-PDP, we extracted 183 distinct legal criteria in 59 articles, revealing that transparency, manageability, and intervenability emerge as predominant regulatory priorities. Our analysis identifies 53 applicable privacy patterns, with the implementation of just 10 key patterns addressing half of the regulatory requirements, providing an efficient pathway toward compliance for resource-constrained organizations. The research contributes a privacy-oriented regulatory engineering framework and empirical evidence that structured approaches can achieve substantial compliance coverage through targeted technical implementations.

While AI-driven financial advisory has become increasingly essential for individuals to navigate volatile markets that offer complex financial products, limited data and privacy concerns pose challenges in its development. This research explores the potential of combining Retrieval-Augmented Generation (RAG) with synthetic data, which imitates real-world data without compromising privacy, in enhancing financial question-answering systems. We propose a framework to generate and incorporate synthetic data into the RAG model with strategies to enhance retrieval processes. The proposed framework is compared with a baseline model without synthetic data to evaluate our approach using the Retrieval-Augmented Generation Assessment (RAGAS). The result shows that the integration of synthetic data can improve recall, precision, and faithfulness of the generated responses. However, relevancy can degrade due to the broader scope of retrieved data. The outcome demonstrates that synthetic data can enhance the accessibility and accuracy of financial data while safeguarding privacy in financial question-answering systems.

With the growing volume of audio and video content generated during both virtual and in-person meetings, there is an increasing need for efficient analysis tools to support decision-making. While recent research has addressed individual tasks such as transcription, speaker identification, and summarization, these components are often developed and applied independently. This work introduces a unified, machine learning-assisted, semi-automated pipeline that integrates these tasks into a cohesive system. The proposed method enables real-time transcription, speaker diarization, and summarization, offering a deeper understanding of meeting dynamics. Unlike traditional tools that lack adaptability and operate in isolation, our approach incorporates a Human-in-the-Loop (HITL) interface for user validation and refinement, enhancing both accuracy and flexibility. By leveraging state-of-the-art speech recognition, speaker embedding models, and topic modeling techniques, our system provides actionable insights from raw meeting recordings with minimal manual intervention. This integrated solution marks a significant advancement in meeting analysis by effectively combining automation with human oversight.

Internet of Things technologies offer the interconnection of various sensors through telecommunications networks, which facilitate their integration into IoT applications. This collaboration of diverse technologies provides an unprecedented opportunity to monitor processes with exceptional precision in human history. Currently, a wide spectrum of end devices contributes to the immense potential of these technologies across various industries. Tracking events and processes offers opportunities for better optimization, both in corporate settings and private life.For business, IoT can reveal inefficiencies in established processes, leading to improved performance. For instance, monitoring freezer temperatures and tracking door neglect can reduce cooling costs. On a personal level, IoT can enhance home security or enable remote control of smart devices.In our research, we focused on analysing the utilization of IoT technologies in companies within Slovakia as EU member state. We explored IoT network architecture, communication protocols, and archetypes of companies leveraging or offering IoT solutions. Our study involved collecting primary data through unstructured interviews with respondents from three companies operating in the telecommunications and IoT Technology industry. Secondary data were compiled from information gathered online portals (Statista, Eurostat) regarding the use of IoT technologies in various sectors and EU countries.The findings of case study are that organizational and business structures defined by theoretical models were found to be used in businesses and have direct impact on the number and services structure of IoT services offered by enterprises. Furthermore the local Slovak market adoption of an IoT services in comparison to EU market is limited and requires regionally balanced disbursement of Digital Europe and RRF funds.

The global economy's transition towards a Circular Economy (CE) is being accelerated by regulatory instruments like the European Union's Digital Product Passport (DPP), mandated under the Ecodesign for Sustainable Products Regulation (ESPR). While the DPP aims to enhance transparency and enable circular practices, its integration presents significant architectural and operational challenges for businesses, particularly within e-commerce. This paper addresses these challenges by proposing a conceptual architectural framework for a scalable, secure, and federated DPP system. Developed through a qualitative synthesis of regulatory analysis, standardization efforts, and expert workshop feedback, the framework outlines the necessary components for both the EU's registry-centric ecosystem and the corresponding enterprise-level implementation. A key finding is that traditional Product Information Management (PIM) systems are insufficient, requiring a more comprehensive, API-driven, and modular architecture. Furthermore, the paper concludes that AI is not merely an efficiency tool but a critical strategic capability. AI is essential for managing data quality at scale, enhancing security, and transforming the DPP from a compliance burden into a value-creating asset through predictive analytics and personalized consumer engagement. The research posits that successfully operationalizing the DPP marks a paradigm shift towards a product-intelligence economy, demanding a holistic architectural approach to realize its full potential.

Businesses are depending more and more on small adjustments to their eServices in the face of rapidly accelerated digital transformation in order to stay competitive and satisfy changing client demands. This study looks at how these small-scale developments affect customer experience (CX), pointing out both the advantages and disadvantages. Based on a thorough examination of the literature, the study investigates how usability, personalization, trust, and cultural variations influence how customers react to changes in digital services. The results highlight the importance of user-centered design, open communication, and agile innovation processes by indicating that even minor adjustments can have a big impact on CX. In addition to providing useful implications for matching innovation objectives with customer expectations, the paper adds to the current conversation on digital service management.

This study examined the effect of a Learning Analytics Dashboard (LAD) on metacognitive activities in self-directed learning environments. We developed a LAD system that visualizes detailed video clickstream analysis and quiz results to support learners’ self-reflection. Through qualitative analysis using questionnaires and interviews, we conducted a one-month experiment with six first-year university students learning linear algebra, measuring metacognitive activities with the Metacognitive Awareness Inventory (MAI).

In this paper, we introduce a new sentence annotation, which relies on visual elements to convey the semantic interpretation. This makes it perfectly suited for the use in language learning environments and other multilingual applications. By building upon the rich experience in the related field of augmentative and alternative communication (AAC), we organize the annotation as a grid of tiles with pictograms for concepts and emoji for roles. As first use case we developed a framework for Japanese based on a flexible, distributed architecture using augmented browsing and several dedicated high-performance servers.

In our study we present the basic theoretical background of On-the-Job training of university graduates employed in the IT sector and the results of research conducted in this specific segment of the Slovak labour market, focusing on the respondents’ experiences prior to On-the-Job training.

This paper proposes a method for generating expressive and distinctive recipe names by identifying each recipe’s unique features relative to others in the same collection. For example, when most recipes boil pasta in a pot, our method may generate a descriptive recipe name like “One-Pan Carbonara” for a recipe that completes the dish using a single frying pan only. The method detects ingredients, cooking procedures, and utensils that are statistical outliers, either significantly more or less frequent compared to the rest of the recipe set. These distinguishing features are then passed to a fine-tuned large language model, which generates the final recipe name. A user study showed that the proposed method produces accurate and appealing names that effectively highlight the distinctiveness of each recipe.

This study proposes a method for finding museum exhibits that visually unexpected shapes, aiming to enhance visitor engagement and memory in museum. The proposed method measures shape-based unexpectedness by combining shape similarity computation with outlier detection. An exhibit is considered unexpected if it is identified as a shape-based outlier. To compute shape similarity, images are first converted into feature vectors using either Vision Transformer (ViT) or Convolutional Neural Networks (CNN). The study also investigates how converting color images into monochrome or line drawings affects the measurement of shape unexpectedness. For outlier detection, two methods, DBSCAN and PageRank-based approach are evaluated. Experiments were conducted using images of exhibits from the National Museum of Ethnology, Japan. Among all tested combinations, the pairing of ConvNeXt, a type of CNN, and PageRank-based approach achieved the highest performance with an nDCG@3 of 0.794.

Skin condition changes throughout the day, yet current skincare does not adapt to these changes. Adaptive skincare requires sensing skin condition and applying appropriate products, ideally in an automatic way even during busy periods. This study proposes a new skincare concept that measures the skin condition and automatically applies lotion to maintain a stable condition. We developed a demo system using a glasses-type sensor device that assesses skin condition and applies a suitable facial mist. In an evaluation experiment, participants used the demo system and completed a questionnaire. Results showed that participants understood and valued the new skincare style, and based on these findings we discussed a possible system design to realize the concept.

This paper explores the potential of Large Language Models (LLMs) to enhance community detection algorithms, with a focus on the SIWO (Strong In, Weak Out) algorithm. By integrating LLMs into the algorithm development process, focusing on their multi-disciplinary knowledge as a potential advantage over human expertise, we demonstrate how LLMs (with the possible oversight of a human expert) can generate innovative algorithm modifications that lead to enhanced performance. Our study reveals substantial reductions in execution times by more than 50% for SIWO when utilizing these modifications. Motivated by these promising results within the domain of Social Networks Analysis, we briefly introduce the Algorithmic Enhancement Framework (AEF), designed to extend these methodologies for broader algorithm enhancement. AEF employs the collaborative use of LLMs to generate and refine solutions iteratively, offering a novel foundational approach for incorporating LLM capabilities for the refinement of algorithms across a broad range of computational domains.

On crowdsourcing platforms, the quality of collected data depends on the clarity of instructions, but requesters struggle to create instructions that capture their own implicit criteria. To address this issue, we propose a novel framework that uses two Large Language Models (LLMs) – a Creator and an Evaluator – to automatically explore the space of possible instructions. In this iterative process, the Creator LLM generates diverse instruction candidates, and the Evaluator LLM, acting as a proxy for human workers, assesses their performance on a task, providing a fitness score. Our experiments show that this exploratory approach is effective for discovering high-quality instructions, even if the process does not show monotonic improvement. Using the best-performing instruction created by our method with gemma3, we achieved 5.4% higher accuracy and 0.035 lower RMSE than when gemma used an instruction created by a requester.

Recent advances in large language models (LLMs) have enabled the replication of speech patterns and philosophies of prominent historical figures. However, generating dialogue that reflects the philosophies of prominent leaders in small communities, such as founders of local universities or small businesses, remains a challenge due to the limited availability of public data. Nevertheless, the philosophies of such leaders often serve as important educational and behavioral foundations for members of these communities. In this study, we propose a dialogue generation method that enables the sharing of a prominent local leader’s philosophy through natural conversation. Specifically, we classify sentences left behind by the leader—such as those in books or diaries—into four types: statements, thoughts, actions, and facts. We then perform two-stage fine-tuning using the statements and thoughts to generate dialogues that faithfully reflect the leader’s values and philosophy.

This paper proposes a method for generating and recommending fashionable outfit images based on a given image of a fashion item. The system uses image generation AI, specifically Stable Diffusion, to produce images of a person wearing the input item, leveraging inpainting techniques to complete the surrounding area. Two models were prepared: a fashionable model fine-tuned on highly rated outfit images from social media, and a normal model without fine-tuning. Both models generated multiple images featuring the input item, and object detection techniques (YOLO and CLIP) were used to identify and count frequently appearing items. Items that appeared more often in the outputs of the fashionable model were prioritized, and the corresponding images were ranked and presented as outfit recommendations. A subject experiment was conducted to evaluate the system, demonstrating that the proposed method can recommend stylish outfits and reflect query items more effectively than metadata-based recommendations.

Recently, multimodal large language models (LLMs) have gained attention due to their ability to handle various data types such as text, images, and audio. These models are useful for diverse tasks, especially through in-context learning, where task-specific performance can be improved by including a few examples in the prompt. However, effective methods for selecting few-shot samples in multimodal LLMs remain unclear. This study explores the impact of image samples on one-shot in-context learning using a violent image classification task. We investigate what kinds of image examples with associated labels help improve classification performance. Experimental results demonstrate that image samples can significantly affect the model’s performance, as shown by comparing zero-shot and one-shot settings. Furthermore, we analyze characteristics of image samples that lead to better or worse classification results. Our findings clarify the role of image examples in enhancing multimodal LLM performance in one-shot in-context learning scenarios.

The rapid advancement of Large Language Models (LLMs) raises fundamental questions about the limits of AI in software development. This paper investigates whether a complete and playable multi-platform web-based game can be developed using only natural language prompts with an advanced LLM, Gemini 2.5 Pro. To explore this, the study compares two opposing development methodologies without the use of a traditional game engine.The study concludes that creating a playable game with prompts alone is possible, but only through a structured, iterative process. This positions the AI not as an autonomous developer, but as a powerful co-pilot that requires skilled, step-by-step guidance from a human expert.