用图聚类让无监督词发现更接近真实语言分布
Recovering the Zipfian Distribution in Unsupervised Term Discovery
- 用相似性构建段落嵌入图，再用莱登算法分割
- 在三种语言中词和音节级发现效果都优于K均值等方法
- 适合关注语言统计规律的自然语言处理研究者
无监督词发现旨在将未标注语音切分为类似词或音节的单元，并聚类形成候选词表。真实词表遵循齐普夫分布，但主流的中心点聚类方法（如K均值）因偏好球形簇，导致分布过于均匀。本文重新考察基于图的聚类作为自下而上的替代方案：将段落嵌入通过成对相似性连接，再使用莱登算法进行分割。实验表明，图聚类在三种语言的词级与音节级词表发现任务中均显著优于中心点方法（K均值、GMM、BIRCH），生成更接近齐普夫分布的结果。另一种自下而上方法——平均链接层次聚类也表现良好，但计算开销更大且难以控制最终分布。本工作质疑了中心点聚类在词发现中的主导地位，提倡图聚类作为有前景的替代方案。
原文摘要 · Abstract (English)
Unsupervised term discovery involves segmenting unlabelled speech into word- or syllable-like units and clustering these into a lexicon of candidate types. True lexicons follow a Zipfian distribution, yet the dominant centre-based clustering approach -- K-means -- produces a more uniform distribution due to an inductive bias toward spherical clusters. In this paper we revisit graph-based clustering as a bottom-up alternative, where segment embeddings are connected by pairwise similarity and partitioned using the Leiden algorithm. We show that graph clustering substantially outperforms centre-based approaches (K-means, GMM, BIRCH) in both word- and syllable-level lexicon discovery across three languages, producing more Zipf-like distributions. Another bottom-up approach, agglomerative clustering with average linkage, also performs well, although it is computationally less efficient and allows for less control over the resulting distribution. Our work calls into question the dominance of centre-based clustering for term discovery, and promotes graph clustering as an attractive alternative.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。