首页|Study Findings from Jiangnan University Provide New Insights into Artificial Int elligence (Improving Distantly Supervised Relation Extraction with Multi-Level N oise Reduction)
Study Findings from Jiangnan University Provide New Insights into Artificial Int elligence (Improving Distantly Supervised Relation Extraction with Multi-Level N oise Reduction)
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
NSTL
By a News Reporter-Staff News Editor at Robotics & Machine Learning Daily News Daily News – New study results on artificial intell igence have been published. According to news reporting from Wuxi, People’s Repu blic of China, by NewsRx journalists, research stated, “Distantly supervised rel ation extraction (DSRE) aims to identify semantic relations in large-scale texts automatically labeled via knowledge base alignment. It has garnered significant attention due to its high efficiency, but existing methods are plagued by noise at both the word and sentence level and fail to address these issues adequately .” The news correspondents obtained a quote from the research from Jiangnan Univers ity: “The former level of noise arises from the large proportion of irrelevant w ords within sentences, while noise at the latter level is caused by inaccurate r elation labels for various sentences. We propose a novel multi-level noise reduc tion neural network (MLNRNN) to tackle both issues by mitigating the impact of m ultilevel noise. We first build an iterative keyword semantic aggregator (IKSA) to remove noisy words, and capture distinctive features of sentences by aggrega ting the information of keywords. Next, we implement multi-objective multi-insta nce learning (MOMIL) to reduce the impact of incorrect labels in sentences by id entifying the cluster of correctly labeled instances. Meanwhile, we leverage mis labeled sentences with cross-level contrastive learning (CCL) to further enhance the classification capability of the extractor. Comprehensive experimental resu lts on two DSRE benchmark datasets demonstrated that the MLNRNN outperformed sta te-of-the-art methods for distantly supervised relation extraction in almost all cases.”
Jiangnan UniversityWuxiPeople’s Repu blic of ChinaAsiaArtificial Intelligence
NETL
NSTL
