arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 1908.06267cs.CL
收藏

用于文档理解的信息传递注意网络

Message Passing Attention Networks for Document Understanding

École Polytechnique(巴黎综合理工学院);Athens University of Economics and Business(雅典经济与商业大学)

Giannis Nikolentzos, Antoine J. -P. Tixier, Michalis Vazirgiannis

摘要:

图神经网络最近已经成为处理图结构数据的非常有效的框架。这些模型在许多任务中实现了最先进的性能。大多数图神经网络可以用消息传递、顶点更新和读出功能来描述。在本文中,我们表示文档的词共现网络,并提出了一个应用程序的消息传递框架NLP,消息传递注意网络文档理解(MPAD)。我们还提出了几个层次的MPAD的变体。在10个标准文本分类数据集上进行的实验表明,我们的体系结构与最先进的体系结构具有竞争力。消融研究揭示了不同组件对性能影响的进一步见解。代码可在https://github.com/giannisnik/mpad上公开获取。

英文摘要:

Graph neural networks have recently emerged as a very effective framework for processing graph-structured data. These models have achieved state-of-the-art performance in many tasks. Most graph neural networks can be described in terms of message passing, vertex update, and readout functions. In this paper, we represent documents as word co-occurrence networks and propose an application of the message passing framework to NLP, the Message Passing Attention network for Document understanding (MPAD). We also propose several hierarchical variants of MPAD. Experiments conducted on 10 standard text classification datasets show that our architectures are competitive with the state-of-the-art. Ablation studies reveal further insights about the impact of the different components on performance. Code is publicly available at: https://github.com/giannisnik/mpad .

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑