SYNTOLOGY HomeExplorerAtlasCodeMethodologyAboutDevelopersFeedPricing
Paper · 2604.15998 · 2026

SCHK-HTC: SIBLING CONTRASTIVE LEARNING WITH HIERARCHICAL KNOWLEDGE-AWARE PROMPT TUNING FOR HIERARCHY TEXT CLASSIFICATION

Xuhong Zhang, Qian Wu, Yuke Li, Ke Xiong, Wangjie Gan

arXiv · PDF · Open in the Atlas

Code that ran

We lifted 10 functions out of this paper's own repositories and ran 7 of them in a sandbox. "Ran" means the function executed on a synthesized input and returned a value. It is not a reproduction of the paper's results.

RepositoryRoleRan
happywinder/SCHK-HTC canonical 7 of 10
FunctionStatusWhere it lives
flat_contrastive_loss_func Ran happywinder/SCHK-HTC/models/loss.py
pointer only (licence: NONE) · get_code("d0609b22f6776b04")
get_umls_neighbors Ran happywinder/SCHK-HTC/KG/add_neighbor.py
pointer only (licence: NONE) · get_code("b9730a1f561ca16e")
load_edge_index Ran happywinder/SCHK-HTC/KG/deepwalk.py
pointer only (licence: NONE) · get_code("0c201bdd1b74c3ac")
load_kg_data Ran happywinder/SCHK-HTC/KG/create_embedding.py
pointer only (licence: NONE) · get_code("c290744aae4d1f42")
load_umls_relations Ran happywinder/SCHK-HTC/KG/add_neighbor.py
pointer only (licence: NONE) · get_code("242fbb2b44aea26a")
query_conceptnet Ran happywinder/SCHK-HTC/query.py
pointer only (licence: NONE) · get_code("0cc82608dcded677")
sim Ran happywinder/SCHK-HTC/models/loss.py
pointer only (licence: NONE) · get_code("7f715d5aa1c04819")
constraint_multi_depth_loss_func_inverse Not yet run happywinder/SCHK-HTC/models/loss.py
pointer only (licence: NONE) · get_code("77936f8eb7dd305f")
get_hybrid_entity_info Not yet run happywinder/SCHK-HTC/KG/linker.py
pointer only (licence: NONE) · get_code("a7954e23a612ae67")
train_with_deepwalk Not yet run happywinder/SCHK-HTC/KG/deepwalk.py
pointer only (licence: NONE) · get_code("0c34c5f611850333")

Repositories linked to this paper

Some links come from the archived Papers with Code dataset (CC BY-SA 4.0): attribution and licence.

Abstract

Few-shot Hierarchical Text Classification (few-shot HTC) is a challenging task that involves mapping texts to a predefined treestructured label hierarchy under data-scarce conditions. While current approaches utilize structural constraints from the label hierarchy to maintain parent-child prediction consistency, they face a critical bottleneck, the difficulty in distinguishing semantically similar sibling classes due to insufficient domain knowledge. We introduce an innovative method named Sibling Contrastive Learning with Hierarchical Knowledge-aware Prompt Tuning for fewshot HTC tasks (SCHK-HTC). Our work enhances the model's perception of subtle differences between sibling classes at deeper levels, rather than just enforcing hierarchical rules. Specifically, we propose a novel framework featuring two core components: a hierarchical knowledge extraction module and a sibling contrastive learning mechanism. This design guides model to encode discriminative features at each hierarchy level, thus improving the separability of confusable classes. Our approach achieves superior performance across three benchmark datasets, surpassing existing state-of-the-art methods in most cases. Our code is available at https://github.com/happywinder/SCHK-HTC.

For agents

The same record, over MCP at https://syntology.ai/mcp:

get_harvested_code_for_paper("2604.15998")
get_code_for_paper("2604.15998")
have("2604.15998")

Connect an agent — have() is free.