内容

学术讲座: Text Matching Algorithms: the Past, Present and Some Thoughts about Future

阅读数:352    发布:2018-06-22 16:27    

主题: Text Matching Algorithms: the Past, Present and Some Thoughts about Future

主讲: 牛笛

时间:6月25日15:00-16:00
地点:科技楼1304会议室

Abstract:

I will describe our recent innovations on text matching algorithms based on deep models, including the emerging graph convolutional neural networks (GCN). I will describe various new techniques we have particularly developed not only for basic natural language sentence matching, but also for long document matching and short-long text matching. Our proposed algorithms can be used for paraphrase identification, question answering, information retrieval, locating the answer span in reading comprehension, as well as for event extraction and document clustering. I will describe the applications of our algorithms to several core NLP tasks at Tencent.

个人简介:牛笛博士于2012年毕业于加拿大多伦多大学,目前是加拿大阿尔伯塔大学电子与计算机工程系的副教授及终身教授,专攻机器学习,数据挖掘,大规模分布式优化算法和分布式系统等跨学科领域。他曾在世界顶级会议和期刊上发表过50多篇论文,包括WWW,SIGKDD,AAAI等,其中20多篇论文为CCF A类论文。他是2016-2017年度CCF-腾讯犀牛鸟基金卓越奖得主。他带领的团队,在犀牛鸟项目中,与腾讯浏览器团队合作,提出了故事森林系统(Story Forest)为热点新闻事件挖掘和事件脉络梳理奠定了基础。


深圳大学计算机与软件学院 2009-2016