芬兰Doria和Theseus联盟机构知识库建设模式及其启示

联盟机构知识库是机构知识库建设与发展的趋势,文章介绍与研究了芬兰Doria和Theseus联盟机构知识库建设模式、运行机制及服务效果,提出了我国在行业/系统及区域联盟机构知识库建设与可持续发展中应借鉴的经验与受到的启示。

THESEUS1 positively modulates plant defense responses against Botrytis cinerea through GUANINE EXCHANGE FACTOR4 signaling

The plant cell wall is an important interface for sensing pathogen attack and activating signaling pathways that promote plant immune responses.THESEUS1(THE1) acts as a sensor of cell wall integrity that controls cell elongation during plant growth.However, no specific role for THE1 in plant defense responses has been reported. Here, we found that THE1 interacts with GUANINE EXCHANGE FACTOR4(GEF4)and that both proteins play regulatory roles in plant resistance to the necrotrophic fungus Botrytis cinerea.Genetic analysis showed that THE1 and GEF4 function in the same genetic pathway to mediate plant defense responses. In addition, using transcriptome analysis, we identified various genes(such as defense-related,secondary metabolite-related, and transcription factor genes) that are likely downstream targets in the THE1-GEF4 signaling pathway. Our results suggest that THE1 functions as an upstream regulator of GEF4 signaling to positively regulate defense responses against B. cinerea in Arabidopsis.

面向任务型对话机器人的多任务联合模型研究

在任务型对话机器人的搭建过程中,一般需要执行多个自然语言处理的子任务。目前传统的训练方式是将每个子任务独立训练后再进行整合,这样忽视了不同子任务之间的关联性,限制了模型的预测能力。现提出一种Joint-RoBERTa-WWM-of-Theseus压缩联合模型,一方面通过多任务联合学习训练的方式对意图识别、行业识别和语义槽填充3个子任务进行联合训练,并在多分类的子任务中引入Focal loss机制来解决数据分布不平衡的问题;另一方面,模型通过Theseus方法进行压缩,在略微损失精度的前提下,大幅提高模型预测速度,提高模型在生产环境中的实时性与实用性。