Our Work

The Iranian–Turkish Shoeprint Database (ITSPDB-289) is an image dataset developed for research on automated footwear impression analysis, shoeprint recognition, image retrieval, and forensic pat…

The Iranian–Turkish Shoeprint Database (ITSPDB-289) is an image dataset developed for research on automated footwear impression analysis, shoeprint recognition, image retrieval, and forensic pattern analysis. The extended dataset comprises 289 physical shoes, with five ink-on-paper impressions per shoe, yielding 1,445 shoeprint images. It combines the original collection of 190 shoes and 950 impressions with an additional collection of 99 shoes and 495 impressions. All impressions were obtained from the participants’ right shoes. During acquisition, the shoe outsole was pressed onto an inkpad and subsequently onto an A3-sized paper sheet (297 × 420 mm). Five impressions were collected from each physical shoe. The resulting paper impressions were scanned in color at a resolution of 300 dots per inch (DPI). The dataset is distributed in two ZIP archives: Raw images.zip: Full-page scanned images containing the shoeprint impressions and the original recording forms. The forms include general participant attributes and footwear-related information, such as height, weight, gender, shoe size, shoe style, and footwear laterality. No names or direct personal identifiers were recorded. Cropped.zip: Cropped shoeprint images with the recording forms removed, prepared for image-processing and machine-learning applications. Each class represents an individual physical shoe rather than an image or footwear design. The five impressions belonging to a class were obtained from the same physical shoe. Researchers using this dataset should account for this structure when defining training, validation, calibration, and test partitions. The original collection is associated with the published ITSPDB research, while the extended collection was developed under the Karadeniz Technical University BAP project FHD-2025-16674. The dataset is intended to support reproducible research in shoeprint retrieval, physical-shoe recognition, feature extraction, partial-impression analysis, and the evaluation of automated footwear-analysis methods. Performance obtained from these controlled ink-on-paper impressions should not be interpreted as direct validation on naturally degraded crime-scene evidence.

Batch upload of files from directory 'output'.

V47 -  ILION v5 × OWASP Top 10 for Agentic Applications 2026 — Architectural Security Crosswalk provides an independent technical mapping between the public ILION v5 deterministic pre…

V47 -  ILION v5 × OWASP Top 10 for Agentic Applications 2026 — Architectural Security Crosswalk provides an independent technical mapping between the public ILION v5 deterministic pre-execution authorization architecture and the OWASP Top 10 for Agentic Applications 2026. ILION v5 is organized as a five-layer authorization stack: CAP → CGC → PCL → ILION Gate → SRM → EXECUTE The layers address distinct authorization dimensions: CAP — Context Attestation Pipeline: provenance, freshness, scope and request-binding of critical authorization context. CGC — Claim Graph Coherence: structural coherence of individually valid authorization claims for the exact requested action. PCL — Policy Constraint Layer: deterministic contextual policy enforcement. ILION Gate: semantic/vector-space role-action authorization before execution. SRM — Session Risk Memory: experimental temporal authorization across multi-step action sequences. The crosswalk maps these mechanisms against the ten OWASP Agentic Applications risk categories: ASI01 — Agent Goal Hijack ASI02 — Tool Misuse and Exploitation ASI03 — Identity and Privilege Abuse ASI04 — Agentic Supply Chain Vulnerabilities ASI05 — Unexpected Code Execution ASI06 — Memory & Context Poisoning ASI07 — Insecure Inter-Agent Communication ASI08 — Cascading Failures ASI09 — Human-Agent Trust Exploitation ASI10 — Rogue Agents Rather than claiming complete coverage, the document distinguishes between: direct / strong architectural overlap partial execution containment limited / adjacent controls areas outside the current ILION v5 scope The mapping intentionally preserves the boundaries of the public ILION v5 architecture. ILION is presented as a deterministic pre-execution authorization control plane, complementary to prompt-injection defenses, supply-chain security, secure inter-agent transport, sandboxing, runtime isolation, monitoring and human-governance mechanisms. The document also identifies important current limitations, including incomplete coverage of time-of-check/time-of-use changes, persistent-memory integrity, distributed multi-agent cascades, runtime code-execution vulnerabilities and human trust exploitation. The public ILION v5 architecture is associated with five Romanian patent applications: OSIM A/00052 — ILION Gate OSIM A/00170 — Policy Constraint Layer (PCL) OSIM A/00171 — Session Risk Memory (SRM) OSIM A/00342 — Context Attestation Pipeline (CAP) OSIM A/00595 — Claim Graph Coherence (CGC) This publication is an independent architectural crosswalk and is not affiliated with or endorsed by OWASP. Related public ILION material: ILION Gate: arXiv:2603.13247 SRM: arXiv:2603.22350 ILION v5 demonstrator: Zenodo Record 23017604 Project: ilion-project.org

This thesis reviews practical approaches to developing creative thinking in preschool children (about 3–6 years old). Drawing on early childhood guidance, two practitioner articles and…

This thesis reviews practical approaches to developing creative thinking in preschool children (about 3–6 years old). Drawing on early childhood guidance, two practitioner articles and a recent pilot study, it describes four approaches: guided pretend play, open-ended materials and process art, dialogic storytelling, and small inquiry projects. Each gives children a chance to propose varied ideas, adapt plans and explain choices through play. Based on these practices, a four-step classroom cycle is proposed: ask, explore, share, revise. Brief observations of talk, gestures, drawings and changes of plan can guide the next activity. The paper is a narrative review; it reports no original intervention or child-level data and makes no claim about measured effectiveness.

Artifact for paper submission: QKLEE: Quantitative Symbolic Execution with Pruning and Incremental Model Counting. 

The artifact contains the network traces and VV8 trace logs for the Penny for Your Prompts: A Measurement Study of Contextual Leakage in AI ChatPlatforms paper.