Contrastive learning with mutual information enhancement and negative sample augmentation combined with KAN for text clustering
Yuanmin Zhang, Hao Li, Chunzhi Xie, Yong Huang, Zhenyi Wu, Yanjun Li · 2026
Read Contrastive learning with mutual information enhancement and negative sample augmentation combined with KAN for text clustering by Yuanmin Zhang, Hao Li, Chunzhi Xie, Yong Huang, Zhenyi Wu, Yanjun Li (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0354718.txt.
Description
Yuanmin Zhang led a team of six authors on this 2026 computer science paper about grouping unlabeled text into topics. Clustering short documents is hard because word features are sparse and tangled. Their model learns by contrasting similar and different texts, sharpens the signal with mutual information, invents more negative examples, and uses a newer network type built for complex patterns.
Across eight standard data sets it took top accuracy on seven. Applied to Chinese social media posts about technology, it separated themes like artificial intelligence and consumer gadgets cleanly, showing real-world promise.
Catalog record
| CREATOR | Yuanmin Zhang, Hao Li, Chunzhi Xie, Yong Huang, Zhenyi Wu, Yanjun Li |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 113755 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |