Geonwoohong/modu-morph-encoded-ko

这是一个面向形态相关研究的韩语语料库，包含了来自八个不同来源（书面、口语、非出版、报纸、即时消息、方面情感、情感、对话）的清洗后的韩语数据。每个样本都被编码成结构化的形态素序列，用于表示语义和风格流。数据集使用Kiwi工具进行形态分析，并以数值张量形式存储在.npz文件中，方便模型直接加载。

This dataset is a morphologically encoded Korean corpus constructed for morph-related research. It integrates eight cleaned Korean sources — written, spoken, nonpublication, newspaper, messenger, aspect_emotion, emotion, and dialogue — and encodes each sample into structured morpheme sequences for semantic and stylistic streams. The dataset uses Kiwi for morphological analysis and is stored in .npz files in numerical tensor format for direct model loading.