"NLP as well as Deep Learning could not only help machines better understand human beings, but also help us to know ourselves better."
My name is S. Xu (许士亭), also known as Will Xu. I am a lecturer at the Department of Cyberspace Security in Shandong University of Political Science and Law.
My research interests are Machine Learning, NLP, Deep Learning and Data Mining. I am very interested in erecting novel models and admiring effects they bring to the world. I have also worked on Information Security related to malicious software classification.
News
- 2022-09-01 I am a lecturer at Department of Cyberspace Security in Shandong University of Political Science and Law.
Publications
Journal Papers
- 2026 S. Xu VIMAR: vision-language informed malware analysis and reasoning model. Cybersecurity, 9, 49, 2026. journal
- 2025 S. Xu DEEP-CWS: Distilling Efficient pre-trained models with Early exit and Pruning for scalable Chinese Word Segmentation. Information Sciences, 719, 122470, 2025. journal
Conference Papers
- 2024 S. Xu. BED: Chinese Word Segmentation Model Based on Boundary-Enhanced Decoder. CACML, 2024. conference
- 2021 S. Xu, et al. Automatic Task Requirements Writing Evaluation via Machine Reading Comprehension. AIED, 2021. [pdf] conference
- 2020 S. Xu, W. Ding, and Z. Liu. Automatic Dialogic Instruction Detection for K-12 Online One-on-One Classes. AIED, 2020. [pdf] conference
- 2017 S. Xu, X. Ma, Y. Liu, and Q. Sheng. Malicious Application Dynamic Detection in Real-Time API Analysis. IEEE iThings-GreenCom-CPSCom-SmartData, pp. 788-794, 2016. [pdf] conference
Projects
BUPT_TAOBAO
Customer classification on TaoBao user behavior data using machine learning (Adaboost, K-means). Achieved 34% precision rate in multi-class classification.
Experience
Du Xiao Man Beijing, China
Responsible for NLP infrastructure. Focused on Chinese Word Segmentation Task.
Tomorrow Advanced Life Beijing, China
Working on Chinese writing judgement system. Developed Chinese Word Correction model based on pre-trained language model.
Working on English writing evaluation, responsible for the whole system. Focused on English Grammar Correction task based on Transformer architecture. Also built a prompt writing task evaluation model based on MRC technology.
Pachira Information Technology Beijing, China
Improved Role accuracy of speech translation model with seq2seq model based on semantic information.
Participated in building a system based on Question-Answer model to extract user information from conversations.
Kaspersky Lab Beijing, China Internship
Designed a malicious software family classification model based on CNN. [Details]
Implemented a CS system (based on tornado) to help analysts train and invoke the model.
Education
Master of Science
School of Cyberspace Security (Former School of Computer Science), Beijing University of Posts and Telecommunications
Bachelor of Engineering
Computer Science Department, Shandong University of Technology
Awards
- 2023 Annual Performance Evaluation Excellence
- 2025 Annual Performance Evaluation Excellence
- 2014.9 – 2017.3 The First Honor Graduate Scholarship for 3 consecutive years
Recent Posts
- 13 May 2026 Harness Engineering Agent LLM
- 19 May 2019 Malicious Software Classification Model Based On Cnn Machine Learning Virus Analysis Information Security Deep Learning
- 19 May 2019 2015 Taobao Challenge Of User Classification Machine Learning Data Mining Competition