Humor generation is a challenging task in natural language generation that requires fluency, surprise, and cultural grounding. Although large language models (LLMs) show promise in English, little work addresses low-resource and culturally specific humor, such as in Korean. We propose **K-Humor**, a *dynamic multi-agent framework* for generating Korean jokes. Unlike static one-shot models, K-Humor integrates specialized agents for cultural awareness, audience simulation, editing, and punchline creation, which are coordinated through a dynamic iterative refinement loop. A human-data-grounded mechanism further guides the punchlines using authentic Korean jokes. Evaluation with automatic metrics, agent-internal diagnostics, LLM-as-a-judge, and human ratings shows that K-Humor consistently outperforms a single-prompt baseline and ablations in criteria such as fluency and cultural appropriateness. Although human-written jokes remain stronger overall, particularly on core creative dimensions, K-Humor demonstrates that dynamic, culturally grounded multi-agent systems can improve humor generation over standard LLM baselines in low-resource settings. Code is available at https://github.com/dxlabskku/k-humor.

As Large Language Models (LLMs) are increasingly used for content creation, detecting AI-generated text has become a critical challenge. Prior work has largely focused on English, leaving low-resource languages such as Korean underexplored. We propose an unsupervised detection framework that integrates two complementary signals: syntactic token cohesiveness (TOCSIN) and semantic regeneration similarity (SimLLM). To support evaluation, we construct a Korean pairwise dataset of 1,000 anchors with continuation- and regeneration-style generations and further assess performance across domains (news, research paper abstracts, essays) and model families (GPT-3.5 Turbo, GPT-4o, HyperCLOVA X, LLaMA-3-8B). Without any training, our ensemble achieves up to 0.963 F1 and 0.985 ROC-AUC, outperforming baselines. These results demonstrate that the combination of syntactic and semantic cues enables robust unsupervised detection in low-resource settings. Code available at https://github.com/dxlabskku/llm-detection-main.

CoT-style LLM recommenders generate a natural-language preference note as a chain-of-thought prefix to ranking, but the note is trusted without verification, and the ranker never consults behaviorally similar users. We propose **ReflectRec**, organized around a preference-to-history round-trip verifier whose recall score serves as an operational signal of preference-to-history recoverability, enabling conditional regeneration for low-recoverability preferences and selection among regeneration attempts. Around this verifier, we wire a semantic Stage-1 prompt, conditional regeneration, and Forward-Look retrieval that injects future watches of behaviorally similar users as a collaborative hint. On MovieLens-100K and MovieLens-1M, the full pipeline outperforms CoT4Rec-ICL by +16.25 / +4.46 pp Hit@10, with positive gains across all ValScore buckets and three LLM backbones (Qwen3-30B, Qwen2.5-72B, and Llama-3.3-70B). We also observe a similar verifier-guided regeneration pattern on Amazon Electronics.

Jujeop is a type of pun and a unique way for fans to express their love for the K-pop stars they follow using Korean. One of the unique characteristics of Jujeop is its use of exaggerated expressions to compliment K-pop stars, which contain or lead to humor. Based on this characteristic, Jujeop can be separated into four distinct types, with their own lexical collocations: (1) Fragmenting words to create a twist, (2) Homophones and homographs, (3) Repetition, and (4) Nonsense. Thus, the current study first defines the concept of Jujeop in Korean, manually labels 8.6K comments and annotates the comments to one of the four Jujeop types. With the given annotated corpus, this study presents distinctive characteristics of Jujeop comments compared to the other comments by classification task. Moreover, with the clustering approach, we proposed a structural dependency within each Jujeop type. We have made our dataset publicly available for future research of Jujeop expressions.