CyberSec Research

URI redirections are integral to web management, supporting structural changes, SEO optimization, and security. However, their complexities affect usability, SEO performance, and digital preservation. This study analyzed 11 million unique redirecting URIs, following redirections up to 10 hops per URI, to uncover patterns and implications of redirection practices. Our findings revealed that 50% of the URIs terminated successfully, while 50% resulted in errors, including 0.06% exceeding 10 hops. Canonical redirects, such as HTTP to HTTPS transitions, were prevalent, reflecting adherence to SEO best practices. Non-canonical redirects, often involving domain or path changes, highlighted significant web migrations, rebranding, and security risks. Notable patterns included "sink" URIs, where multiple redirects converged, ranging from traffic consolidation by global websites to deliberate "Rickrolling." The study also identified 62,000 custom 404 URIs, almost half being soft 404s, which could compromise SEO and user experience. These findings underscore the critical role of URI redirects in shaping the web while exposing challenges such as outdated URIs, server instability, and improper error handling. This research offers a detailed analysis of URI redirection practices, providing insights into their prevalence, types, and outcomes. By examining a large dataset, we highlight inefficiencies in redirection chains and examine patterns such as the use of "sink" URIs and custom error pages. This information can help webmasters, researchers, and digital archivists improve web usability, optimize resource allocation, and safeguard valuable online content.

Advances in embedding models for text, image, audio, and video drive progress across multiple domains, including retrieval-augmented generation, recommendation systems, vehicle/person reidentification, and face recognition. Many applications in these domains require an efficient method to retrieve items that are close to a given query in the embedding space while satisfying a filter condition based on the item's attributes, a problem known as Filtered Approximate Nearest Neighbor Search (FANNS). In this work, we present a comprehensive survey and taxonomy of FANNS methods and analyze how they are benchmarked in the literature. By doing so, we identify a key challenge in the current FANNS landscape: the lack of diverse and realistic datasets, particularly ones derived from the latest transformer-based text embedding models. To address this, we introduce a novel dataset consisting of embedding vectors for the abstracts of over 2.7 million research articles from the arXiv repository, accompanied by 11 real-world attributes such as authors and categories. We benchmark a wide range of FANNS methods on our novel dataset and find that each method has distinct strengths and limitations; no single approach performs best across all scenarios. ACORN, for example, supports various filter types and performs reliably across dataset scales but is often outperformed by more specialized methods. SeRF shows excellent performance for range filtering on ordered attributes but cannot handle categorical attributes. Filtered-DiskANN and UNG excel on the medium-scale dataset but fail on the large-scale dataset, highlighting the challenge posed by transformer-based embeddings, which are often more than an order of magnitude larger than earlier embeddings. We conclude that no universally best method exists.

Modern analytical pipelines routinely deploy multiple deep learning and retrieval models that rely on approximate nearest-neighbor (ANN) indexes to support efficient similarity-based search. While many state-of-the-art ANN-indexes are memory-based (e.g., HNSW and IVF), using multiple ANN indexes creates a competition for limited GPU/CPU memory resources, which in turn necessitates disk-based index structures (e.g., DiskANN or eCP). In typical index implementations, the main component is a complex data structure that is serialized to disk and is read either fully at startup time, for memory-based indexes, or incrementally at query time, for disk-based indexes. To visualize the index structure, or analyze its quality, complex coding is needed that is either embedded in the index implementation or replicates the code that reads the data structure. In this paper, we consider an alternative approach that maps the data structure to a file structure, using a file library, making the index easily readable for any programming language and even human-readable. The disadvantage is that the serialized index is verbose, leading to overhead of searching through the index. The question addressed in this paper is how severe this performance penalty is. To that end, this paper presents eCP-FS, a file-based implementation of eCP, a well-known disk-based ANN index. A comparison with state-of-the-art indexes shows that while eCP-FS is slower, the implementation is nevertheless somewhat competitive even when memory is not constrained. In a memory-constrained scenario, eCP-FS offers a minimal memory footprint, making it ideal for resource-constrained or multi-index environments.

As news reporting becomes increasingly global and decentralized online, tracking related events across multiple sources presents significant challenges. Existing news summarization methods typically utilizes Large Language Models and Graphical methods on article-based summaries. However, this is not effective since it only considers the textual content of similarly dated articles to understand the gist of the event. To counteract the lack of analysis on the parties involved, it is essential to come up with a novel framework to gauge the importance of stakeholders and the connection of related events through the relevant entities involved. Therefore, we present SUnSET: Synergistic Understanding of Stakeholder, Events and Time for the task of Timeline Summarization (TLS). We leverage powerful Large Language Models (LLMs) to build SET triplets and introduced the use of stakeholder-based ranking to construct a $Relevancy$ metric, which can be extended into general situations. Our experimental results outperform all prior baselines and emerged as the new State-of-the-Art, highlighting the impact of stakeholder information within news article.

Recommender systems often struggle with over-specialization, which severely limits users' exposure to diverse content and creates filter bubbles that reduce serendipitous discovery. To address this fundamental limitation, this paper introduces an adaptive clustering framework with user-controlled exploration that effectively balances personalization and diversity in movie recommendations. Our approach leverages sentence-transformer embeddings to group items into semantically coherent clusters through an online algorithm with dynamic thresholding, thereby creating a structured representation of the content space. Building upon this clustering foundation, we propose a novel exploration mechanism that empowers users to control recommendation diversity by strategically sampling from less-engaged clusters, thus expanding their content horizons while preserving relevance. Experiments on the MovieLens dataset demonstrate the system's effectiveness, showing that exploration significantly reduces intra-list similarity from 0.34 to 0.26 while simultaneously increasing unexpectedness to 0.73. Furthermore, our Large Language Model-based A/B testing methodology, conducted with 300 simulated users, reveals that 72.7% of long-term users prefer exploratory recommendations over purely exploitative ones, providing strong evidence for the system's ability to promote meaningful content discovery without sacrificing user satisfaction.

The importance of recommender systems on the web has grown, especially in the movie industry, with a vast selection of options to watch. To assist users in traversing available items and finding relevant results, recommender systems analyze operational data and investigate users' tastes and habits. Providing highly individualized suggestions can boost user engagement and satisfaction, which is one of the fundamental goals of the movie industry, significantly in online platforms. According to recent studies and research, using knowledge-based techniques and considering the semantic ideas of the textual data is a suitable way to get more appropriate results. This study provides a new method for building a knowledge graph based on semantic information. It uses the ChatGPT, as a large language model, to assess the brief descriptions of movies and extract their tone of voice. Results indicated that using the proposed method may significantly enhance accuracy rather than employing the explicit genres supplied by the publishers.

Recommendation systems often suffer from data sparsity caused by limited user-item interactions, which degrade their performance and amplify popularity bias in real-world scenarios. This paper proposes a novel data augmentation framework that leverages Large Language Models (LLMs) and item textual descriptions to enrich interaction data. By few-shot prompting LLMs multiple times to rerank items and aggregating the results via majority voting, we generate high-confidence synthetic user-item interactions, supported by theoretical guarantees based on the concentration of measure. To effectively leverage the augmented data in the context of a graph recommendation system, we integrate it into a graph contrastive learning framework to mitigate distributional shift and alleviate popularity bias. Extensive experiments show that our method improves accuracy and reduces popularity bias, outperforming strong baselines.

Vision Large Language Models (VLLMs) have improved multi-modal understanding and visual question answering (VQA), but still suffer from hallucinated answers. Multi-modal Retrieval-Augmented Generation (RAG) helps address these issues by incorporating external information, yet challenges remain in visual context comprehension, multi-source retrieval, and multi-turn interactions. To address these challenges, Meta constructed the CRAG-MM benchmark and launched the CRAG-MM Challenge at KDD Cup 2025, which consists of three tasks. This paper describes the solutions of all tasks in Meta KDD Cup'25 from BlackPearl team. We use a single model for each task, with key methods including data augmentation, RAG, reranking, and multi-task fine-tuning. Our solution achieve automatic evaluation rankings of 3rd, 3rd, and 1st on the three tasks, and win second place in Task3 after human evaluation.

This full research paper investigates the impact of generative AI (GenAI) on the learner experience, with a focus on how learners engage with and utilize the information it provides. In e-learning environments, learners often need to navigate a complex information space on their own. This challenge is further compounded in interdisciplinary fields like bioinformatics, due to the varied prior knowledge and backgrounds. In this paper, we studied how GenAI influences information search in bioinformatics research: (1) How do interactions with a GenAI chatbot influence learner orienteering behaviors?; and (2) How do learners identify information scent in GenAI chatbot responses? We adopted an autoethnographic approach to investigate these questions. GenAI was found to support orienteering once a learning plan was established, but it was counterproductive prior to that. Moreover, traditionally value-rich information sources such as bullet points and related terms proved less effective when applied to GenAI responses. Information scents were primarily recognized through the presence or absence of prior knowledge of the domain. These findings suggest that GenAI should be adopted into e-learning environments with caution, particularly in interdisciplinary learning contexts.

Despite success across diverse tasks, current artificial recurrent network architectures rely primarily on implicit hidden-state memories, limiting their interpretability and ability to model long-range dependencies. In contrast, biological neural systems employ explicit, associative memory traces (i.e., engrams) strengthened through Hebbian synaptic plasticity and activated sparsely during recall. Motivated by these neurobiological insights, we introduce the Engram Neural Network (ENN), a novel recurrent architecture incorporating an explicit, differentiable memory matrix with Hebbian plasticity and sparse, attention-driven retrieval mechanisms. The ENN explicitly models memory formation and recall through dynamic Hebbian traces, improving transparency and interpretability compared to conventional RNN variants. We evaluate the ENN architecture on three canonical benchmarks: MNIST digit classification, CIFAR-10 image sequence modeling, and WikiText-103 language modeling. Our empirical results demonstrate that the ENN achieves accuracy and generalization performance broadly comparable to classical RNN, GRU, and LSTM architectures, with all models converging to similar accuracy and perplexity on the large-scale WikiText-103 task. At the same time, the ENN offers significant enhancements in interpretability through observable memory dynamics. Hebbian trace visualizations further reveal biologically plausible, structured memory formation processes, validating the potential of neuroscience-inspired mechanisms to inform the development of more interpretable and robust deep learning models.

The growing popularity of short-form video content, such as YouTube Shorts, has transformed user engagement on digital platforms, raising critical questions about the role of recommendation algorithms in shaping user experiences. These algorithms significantly influence content consumption, yet concerns about biases, echo chambers, and content diversity persist. This study develops an efficient data collection framework to analyze YouTube's recommendation algorithms for both short-form and long-form videos, employing parallel computing and advanced scraping techniques to overcome limitations of YouTube's API. The analysis uncovers distinct behavioral patterns in recommendation algorithms across the two formats, with short-form videos showing a more immediate shift toward engaging yet less diverse content compared to long-form videos. Furthermore, a novel investigation into biases in politically sensitive topics, such as the South China Sea dispute, highlights the role of these algorithms in shaping narratives and amplifying specific viewpoints. By providing actionable insights for designing equitable and transparent recommendation systems, this research underscores the importance of responsible AI practices in the evolving digital media landscape.

Extracting structured information from text, such as key-value pairs that could augment tabular data, is quite useful in many enterprise use cases. Although large language models (LLMs) have enabled numerous automated pipelines for converting natural language into structured formats, there is still a lack of benchmarks for evaluating their extraction quality, especially in specific domains or focused documents specific to a given organization. Building such benchmarks by manual annotations is labour-intensive and limits the size and scalability of the benchmarks. In this work, we present StructText, an end-to-end framework for automatically generating high-fidelity benchmarks for key-value extraction from text using existing tabular data. It uses available tabular data as structured ground truth, and follows a two-stage ``plan-then-execute'' pipeline to synthetically generate corresponding natural-language text. To ensure alignment between text and structured source, we introduce a multi-dimensional evaluation strategy that combines (a) LLM-based judgments on factuality, hallucination, and coherence and (b) objective extraction metrics measuring numeric and temporal accuracy. We evaluated the proposed method on 71,539 examples across 49 datasets. Results reveal that while LLMs achieve strong factual accuracy and avoid hallucination, they struggle with narrative coherence in producing extractable text. Notably, models presume numerical and temporal information with high fidelity yet this information becomes embedded in narratives that resist automated extraction. We release a framework, including datasets, evaluation tools, and baseline extraction systems, to support continued research.

Modeling user behavior is critical across many industries where understanding preferences, intent, or decisions informs personalization, targeting, and strategic outcomes. Surveys have long served as a classical mechanism for collecting such behavioral data due to their interpretability, structure, and ease of deployment. However, surveys alone are inherently limited by user fatigue, incomplete responses, and practical constraints on their length making them insufficient for capturing user behavior. In this work, we present LANTERN (Late-Attentive Network for Enriched Response Modeling), a modular architecture for modeling user behavior by fusing adaptive survey responses with supplemental contextual signals. We demonstrate the architectural value of maintaining survey primacy through selective gating, residual connections and late fusion via cross-attention, treating survey data as the primary signal while incorporating external modalities only when relevant. LANTERN outperforms strong survey-only baselines in multi-label prediction of survey responses. We further investigate threshold sensitivity and the benefits of selective modality reliance through ablation and rare/frequent attribute analysis. LANTERN's modularity supports scalable integration of new encoders and evolving datasets. This work provides a practical and extensible blueprint for behavior modeling in survey-centric applications.

Large Language Model-based Time Series Forecasting (LLMTS) has shown remarkable promise in handling complex and diverse temporal data, representing a significant step toward foundation models for time series analysis. However, this emerging paradigm introduces two critical challenges. First, the substantial commercial potential and resource-intensive development raise urgent concerns about intellectual property (IP) protection. Second, their powerful time series forecasting capabilities may be misused to produce misleading or fabricated deepfake time series data. To address these concerns, we explore watermarking the outputs of LLMTS models, that is, embedding imperceptible signals into the generated time series data that remain detectable by specialized algorithms. We propose a novel post-hoc watermarking framework, Waltz, which is broadly compatible with existing LLMTS models. Waltz is inspired by the empirical observation that time series patch embeddings are rarely aligned with a specific set of LLM tokens, which we term ``cold tokens''. Leveraging this insight, Waltz embeds watermarks by rewiring the similarity statistics between patch embeddings and cold token embeddings, and detects watermarks using similarity z-scores. To minimize potential side effects, we introduce a similarity-based embedding position identification strategy and employ projected gradient descent to constrain the watermark noise within a defined boundary. Extensive experiments using two popular LLMTS models across seven benchmark datasets demonstrate that Waltz achieves high watermark detection accuracy with minimal impact on the quality of the generated time series.

In e-commerce recommender and search systems, tree-based models, such as LambdaMART, have set a strong baseline for Learning-to-Rank (LTR) tasks. Despite their effectiveness and widespread adoption in industry, the debate continues whether deep neural networks (DNNs) can outperform traditional tree-based models in this domain. To contribute to this discussion, we systematically benchmark DNNs against our production-grade LambdaMART model. We evaluate multiple DNN architectures and loss functions on a proprietary dataset from OTTO and validate our findings through an 8-week online A/B test. The results show that a simple DNN architecture outperforms a strong tree-based baseline in terms of total clicks and revenue, while achieving parity in total units sold.

Recent advances in recommender systems rely on external resources such as knowledge graphs or large language models to enhance recommendations, which limit applicability in real-world settings due to data dependency and computational overhead. Although knowledge-free models are able to bolster recommendations by direct edge operations as well, the absence of augmentation primitives drives them to fall short in bridging semantic and structural gaps as high-quality paradigm substitutes. Unlike existing diffusion-based works that remodel user-item interactions, this work proposes NodeDiffRec, a pioneering knowledge-free augmentation framework that enables fine-grained node-level graph generation for recommendations and expands the scope of restricted augmentation primitives via diffusion. By synthesizing pseudo-items and corresponding interactions that align with the underlying distribution for injection, and further refining user preferences through a denoising preference modeling process, NodeDiffRec dramatically enhances both semantic diversity and structural connectivity without external knowledge. Extensive experiments across diverse datasets and recommendation algorithms demonstrate the superiority of NodeDiffRec, achieving State-of-the-Art (SOTA) performance, with maximum average performance improvement 98.6% in Recall@5 and 84.0% in NDCG@5 over selected baselines.

We present ZSE-Cap (Zero-Shot Ensemble for Captioning), our 4th place system in Event-Enriched Image Analysis (EVENTA) shared task on article-grounded image retrieval and captioning. Our zero-shot approach requires no finetuning on the competition's data. For retrieval, we ensemble similarity scores from CLIP, SigLIP, and DINOv2. For captioning, we leverage a carefully engineered prompt to guide the Gemma 3 model, enabling it to link high-level events from the article to the visual content in the image. Our system achieved a final score of 0.42002, securing a top-4 position on the private test set, demonstrating the effectiveness of combining foundation models through ensembling and prompting. Our code is available at https://github.com/ductai05/ZSE-Cap.

Finding potential research collaborators is a challenging task, especially in today's fast-growing and interdisciplinary research landscape. While traditional methods often rely on observable relationships such as co-authorships and citations to construct the research network, in this work, we focus solely on publication content to build a topic-based research network using BERTopic with a fine-tuned SciBERT model that connects and recommends researchers across disciplines based on shared topical interests. A major challenge we address is publication imbalance, where some researchers publish much more than others, often across several topics. Without careful handling, their less frequent interests are hidden under dominant topics, limiting the network's ability to detect their full research scope. To tackle this, we introduce a cloning strategy that clusters a researcher's publications and treats each cluster as a separate node. This allows researchers to be part of multiple communities, improving the detection of interdisciplinary links. Evaluation on the proposed method shows that the cloned network structure leads to more meaningful communities and uncovers a broader set of collaboration opportunities.