学术前沿讲座——Which Solutions to Recommend? Evaluating Contents on Healthcare Forums Using Deep Learning Approaches

报告题目

Which Solutions to Recommend? Evaluating Contents on Healthcare Forums Using Deep Learning Approaches

报告人（单位）

张成龙 助理教授（香港中文大学（深圳））

主持人（单位）

郭怡婷 (泛亚电竞官方网站）

时间及会议ID

讲座时间：2023年9月11日（周一）下午4:00

讲座地点：泛亚电竞官方网站九龙湖校区经管楼B203会议室(科研活动中心)

(腾迅会议id 780-775-084

份线上形式时用！)

报告内容摘要

Online forums face the challenge of managing user-generated content while serving as custodians of the internet. The challenge is especially significant for online forums as on these forums millions of users with diverse health concerns seek advice, post opinions, and answer questions. With an overwhelming amount of daily generated information, it is crucial to moderate content and prioritize helpful answers for ongoing discussions. AI can be a necessary response to this challenge, but it remains unclear how to effectively automate UGC management. In this paper, we propose an efficient system of criteria for evaluating the answers on health forums and combine the criterion system with end-to-end deep learning models, which achieve both high accuracy and interpretability. When using the deep learning models, We adapt the existing models to the healthcare context for effectively learning some measures in the context.  This framework showcases how to leverage the capabilities of language models to eliminate the need for laborious and imperfect feature engineering when evaluating answers from online health forums.

报告人简介：

张成龙为香港中文大学（深圳）信息系统领域助理教授，2021年毕业于UT Dallas商学院。其研究领域为共享经济、应用人工智能和区块链。其以第一作者在Management Science、Information Systems Research等期刊发表论文。此外其担任MS, ISR, MISQ以及POM等期刊的审稿人，并参与国家自科重点项目一项。