Retrieval-Augmented Generation (RAG) enhances Large Language Models (LLMs) by incorporating external knowledge, particularly for long-tail domains such as literary works. However, the critical step of document segmentation in RAG remains largely underexplored. Existing strategies typically either ignore semantics or overlook the complicated narrative structures of literary works, often resulting in chunks with fragmented plots and unclear references that hinder retrieval and generation performance. To address this, we propose LitSeg, a novel narrative-theory-guided segmentation framework. By employing multi-stage prompting, LitSeg explicitly extracts valid events, clarifies narrative structures, and locates turning points to inform segmentation. To alleviate the computational overhead of multi-stage inference with large-scale models, we further introduce LitSeg-Lite, a lightweight single-pass chunker fine-tuned on LitSeg-generated data via a two-stage training strategy, distilling the complex process into a single inference pass. Extensive experiments demonstrate that compared to baselines, our methods yield high-quality text chunks that are narratologically coherent and self-contained. This improved segmentation quality enhances retrieval accuracy and context relevance, and boosts downstream QA performance. Ablation studies validate the efficacy of narratological guidance and data distillation, and efficiency analysis shows that LitSeg-Lite matches the teacher at a substantially lower inference cost.

Computational value analysis has largely treated values as instances of predefined universal taxonomies, making it difficult to discover culturally specific value dimensions not fully represented in existing schemes. We formulate value analysis as an open-ended discovery problem grounded in the argumentative contexts in which values are expressed. To this end, we propose an LLM-assisted framework that discovers human value dimensions directly from text without relying on predefined taxonomies. It extracts Premise-Conclusion-Stance (P-C-S) argumentative edges under explicit argument tree constraints, expands each edge into structured value-semantic descriptions spanning goal, principle, tradeoff, and perspective dimensions, and induces coherent value clusters via co-association consensus clustering over the embedding representations. We evaluate the framework across ten languages using external label alignment, internal structural stability metrics, and human judgment. Results show that the framework is robust and transferable, with the induced value structures achieving low hallucination and strong alignment with human judgments while revealing culturally distinctive value orientations that both align with and meaningfully complement established psychological frameworks. By shifting computational value analysis from taxonomy-driven classification toward open-ended discovery, our framework offers a new approach to cross-cultural value research, with implications for culturally aware AI alignment.