Leave a little heart!
Feiyang Ying's bear avatar

Hi,

I'm Feiyang.

保持变化,保持探索,保持思考。

I am an undergraduate in Data Science and Big Data Technology at Chongqing University, originally from Danyang, Jiangsu.

My interests currently sit around Natural Language Processing, Large Language Models, Model Behavior & Evaluation, and Reliable AI Systems.

I want to pair research judgment with the ability to build: choosing worthwhile questions, then testing them carefully in research and code. I am always happy to connect.

News! ✨

  1. 📝 Revising our PTOC paper after a minor-revision decision from IPM.

  2. 💻 Won Provincial First Prize in the Chinese Collegiate Computing Competition (4C)!

  3. 🔍 Our cross-platform ranking paper is under review at AEI.

  4. 🏅 Received the National Scholarship!

  5. 🏆 Won First Prize in the National English Competition for College Students!

feiyang — PowerShell

visitor@feiyang.dev:~$ whoami

Name:Feiyang Ying

Role:Undergraduate Researcher

Study:Chongqing University

Code:Python / C++ / TypeScript

Focus:NLP × Reliable Systems

Work:02 papers / 01 project

01 /

Publications & research

02 first-author manuscripts · 64,041 reviews · National project

R—01 / FIG. 02 Five-module PTOC consumer decision-support framework

Prospect trade-off comparison framework: Mining online reviews for consumer decision support

Feiyang Ying1, Honggang Peng2,*, Jianqiang Wang3, Xiaokang Wang4

Information Processing & Management IPM · Minor revision

Extracts aspect–opinion–sentiment structures with a QLoRA-adapted Qwen model, then models compensatory consumer trade-offs for decision support.

R—02 / FIG. 01 Five-stage cross-platform product ranking framework

Mining cross-platform online reviews for consumer decision support: A representation-debiased SAGE-TODIM framework

Feiyang Ying1, Bo Deng1, Jiahui Liu2, Xianghui Lyu2, Honggang Peng2,*, Jianqiang Wang3, Xiaokang Wang4

Advanced Engineering Informatics AEI · Under review

Probes platform signals in review representations, removes their linear directions with LEACE, and aligns cross-platform semantics before ranking.