Xin Xin(辛欣，北京理工大学)

I am currently an associate professor in Beijing Institute of Technology (Director: Prof. Heyan Huang). Before, I obtained the bachelor degree and the master degree (Supervisor: Prof. Juanzi Li) from Tsinghua University (Beijing) in 2006 and 2008, respectively, and obtained the Ph. D degree (Supervisor: Prof. Irwin King and Prof. Michael R. Lyu) from the Chinese University of Hong Kong (Hong Kong, China) in 2011. I visited Microsoft Asia under the "Star Track" program (Collaboration Mentor: Chin-Yew Lin) in 2013. Before the large language model (LLM) era, my research interests focused on the structure prediction problem, based on designing the architecture of probabilistic graphical models and investigating the information fusion technique for heterogeneous features. Recently, I try to apply these theories in the areas of multimodal information retrieval and inference based on LLMs.

1. Insterested Application: Multimodal Information Retrieval and Inference based on LLMs

• Semantic Parsing for Texts, Audios, and Images: To understand the semantic structure of these social media content.
• Automatic Knowledge Graph Construction: To convert to Web data into structured knowledge graph data.
• Fine-Grained Semantic Retrieval: To retrieval the information from the granularity of syntactic analysis.
• Reliable Fact Inference: To conduct explainable LLM inference based on logic theories.

2. Insterested Theory: Structure Prediction Approaches

2.1 Designing the Architecture of Probabilistic Graphical Models

• Recursive Semi-Markov model
  Xin Xin, Jinlong Li, and Zeqi Tan. N-ary Constituent Tree Parsing with Recursive Semi-Markov Model. In Proceedings of the ACL 2021 (CCF A).
• An unified model of semi-CRF and tree-CRF:
  Rui Wang, Xin Xin, Wei Chang, Kun Ming, Biao Li, and Xin Fan. Chinese NER with Height Limited Constituent Parsing. In Proceedings of the AAAI 2019 (CCF A). (Corresponding)
• Multi-layer continuous CRF:
  Xin Xin, Irwin King, Hongbo Deng and Michael R. Lyu. A Social Recommendation Framework Based on Multi-scale Continuous Conditional Random Fields. In Proceedings of the CIKM 2009 (CCF B).
• Constrained tree-structure CRF:
  Xin Xin, Juanzi Li, Jie Tang and Qiong Luo. Academic Conference Homepage Understanding Using Constrained Hierarchical Conditional Random Fields. In Proceedings of the CIKM 2008 (CCF B).

2.2 Investigating the Information Fusion Technique for Heterogeneous Features

• Information fusion by transfer learning:
  Xin Xin, Zhirun Liu, Chin-Yew Lin, Heyan Huang, Xiaochi Wei, Ping Guo. Cross-Domain Collaborative Filtering with Review Text. In Proceedings of the IJCAI 2015 (CCF A).
• Information fusion by multi-view learning:
  Xin Xin, Chunwei Lu, Yashen Wang, and Heyan Huang. Forecasting Collector Road Speeds Under High Percentage of Missing Data. In Proceedings of the AAAI 2015 (CCF A).
• Information fusion by space projection:
  Xin Xin, Xiaochi Wei, Chin-Yew Lin, Heyan Huang. When Factorization Meets Heterogenous Latent Topics: An Interpretable Cross-Site Recommendation Framework. JCST 2015 (CCF B).
• Information fusion by network diffusion:
  Xin Xin, Michael R. Lyu, and Irwin King. CMAP: Effective Fusion of Quality and Relevance for Multi-Criteria Recommendation. In Proceedings of the WSDM 2011 (CCF B).

3. Current Group Members

• Jinlong Li (李金龙), jointlee@qq.com
• Sijia Zhao (赵思佳), sijiazhao@bit.edu.cn
• Zhengfei Xu (徐正斐), xuzhengfei-email@qq.com
• Jiaxuan Dong (董家暄), 867055628@qq.com
• Hongyu Chen (陈泓宇), 786761046@qq.com
• Haoze Wang (王皓泽), w_haoze@bit.edu.cn
• Chaojie Jiang (姜超杰), chaojie_jiang@qq.com
• Weihan Li (李纬涵), lwh200218@gmail.com
• Yansong Wang (王岩松), 1732085139@qq.com

4. Previous Group Members

• 王宇杰(2024)，成都飞机工业集团
• 徐进(2024)，快手
• 葛文雪(2024)，中国建设银行
• 苗昊雲(2023)，北方自动控制技术研究所
• 田雄辉(2023)，招商银行
• 李丽丽(2022)，腾讯
• 袁嘉乐(2022)，阿里巴巴
• 赖一多(2022)，阿里巴巴
• 王艳平(2022)，航天院
• 王艳(2021)，字节跳动
• 史秀杰(2021)，国家电网
• 刘明磊，中国工商银行
• 朱宗伟，Shopee
• 潘强，搜狗
• 王亚，中国交通银行
• 王睿，阿里巴巴（校级优秀毕业生）
• 张璐璐，网联清算有限公司
• 常薇，民生银行
• 仇金娟，华为
• 张快，凤凰网
• 马翠花，中国移动研究院
• 季鹏，中信银行
• 罗佩，航天院
• 胡健，中科院自动化所
• 张杰，美团
• 袁鹏，中信银行