Scholar

Kaiqiang Song (宋凯强）

Google Scholar ID: PHoJwakAAAAJ

AI Scientist @ Zoom | Ex-Tencent

LLMNLPNLGSummarization

Citations & Impact

All-time

Citations

1,318

H-index

i10-index

Publications

Co-authors

list available

Contact

Publications

20 items

Browse publications on Google Scholar (top-right) ↗

Resume (English only)

Academic Achievements

Won 1st place in automatic evaluation and 2nd place in human evaluation in the TREC2020 podcast summarization challenge. Published papers: 'Bridging Continuous and Discrete Spaces: Interpretable Sentence Representation Learning via Compositional Operations' at EMNLP 2023, 'DecipherPref: Analyzing Influential Factors in Human Preference Judgments via GPT-4' at EMNLP 2023, 'PIVOINE: Instruction Tuning for Open-world Information Extraction' at EMNLP 2023 (Findings), 'MMC: Advancing Multimodal Chart Understanding with Large-scale Instruction Tuning' on ArXiv.org, 'Unsupervised Multi-document Summarization with Holistic Inference' at IJCNLP-AACL 2023.

Research Experience

Currently a Senior Research Scientist at Tencent AI Lab, developing fundamental and impactful technologies for ML, NLP, and LLM. Leads summarization research and provides related application support (e.g., VooV Meeting, QQ Web-Browser, Effidit, and Sougou Baike). Participates in developing Tencent Foundation Model HunyuanAide (a ChatGPT-like AI-Bot). Manages computation and storage resources for Tencent Seattle AI Lab. Communicates with Infrastructure and Platform Teams for better support of daily research.

Education

PhD in Computer Science from the University of Central Florida in 2021; BS in Computer Science from Fudan University in 2016.

Background

Research interests include machine learning (ML), natural language processing (NLP), and large language models (LLM). He is dedicated to enhancing AI performance by optimizing model architectures and training strategies, with practical applications such as text summarization and text generation.

Miscellany