Yiming Qiao

Yiming Qiao

Phd student in Institute for Interdisciplinary Information Sciences, Tsinghua University

Yiming Qiao

I am a Ph.D. student in Institute for Interdisciplinary Information Sciences (IIIS) at Tsinghua University, where I am advised by Prof. Huanchen Zhang. My research focuses on database management systems, with particular interests in vectorized execution, query optimization, and data compression. I obtained my Bachelor’s degree from Nanjing University of Posts and Telecommunications.

Recently, I have been focusing on two major projects:

  1. Parallelism-friendly Query Optimization: This project aims to enhance the parallelism of query plans, improving the overall efficiency and scalability of query execution.

  2. Multi-modal Query Execution: This project supports complex queries that involve both structured and unstructured data. An example of such a query is:

SELECT * 
FROM Movies AS M 
WHERE M.release_date = 2012 
  AND M.summary IS ABOUT 'A Love Story in World War II'; -- Uses LLM to filter summaries


[CV] [中文简历]

Education

Tsinghua University
Ph.D. in Computer Science, Sept. 2021 - Jun. 2026

Nanjing University of Posts and Telecommunications
B.Eng. in Computer Science, Sept. 2017 - Jun. 2021

Experience

eBay - Software Engineering Intern
Shanghai, China | Sept. 2020 - Nov. 2020

Oracle - Software Engineering Intern
Nanjing, China | Nov. 2019 - May 2020

Nanjing University - Exchange Student
Sept. 2018 - Jun. 2019

Publications

  1. Yiming Qiao, Huanchen Zhang, “Data Chunk Compaction in Vectorized Execution,” Proceedings of the ACM on Management of Data (SIGMOD’25), Accepted. [Paper] [Code]

    In this paper, we highlight the “small chunk problem” in vectorized query execution, identifying the filter and hash join as the most problematic operators. We propose two solutions: Learning Compaction for filters and Logical Compaction for hash joins. The Logical Compaction has been used by DuckDB (#14956).

  2. Yiming Qiao, Yihan Gao, Huanchen Zhang, “Blitzcrank: Fast Semantic Compression for In-memory Online Transaction Processing,” Proceedings of the VLDB Endowment (VLDB’24) 17, no. 10, pp. 2528 - 2540. [Paper] [Code] [Poster] [Slides]

    We propose Delayed Coding, a new entropy coding algorithm (e.g., Arithmetic Coding, Asymmetric Numeral System). Based on it, we design a compressor for row-store OLTP databases. It has high compression factors, and fast random access (to a single tuple), about 1 µs/tuple – slower than memory (200 ns) but much faster than SSD (40 µs).

  3. Hu Zhu*, Yiming Qiao*, Guoxia Xu, Lizhen Deng, and Yu-Feng Yu. “DSPNet: A Lightweight Dilated Convolution Neural Networks for Spectral Deconvolution with Self-paced Learning,” IEEE Transactions on Industrial Informatics (TII) 16, no. 12 (2019): 7392-7401. (*Equal Contribution)

  4. Huihui Wang, Shunmei Meng, Yiming Qiao, and Jing Zhang. “Fast Classification Algorithms via Distributed Accelerated Alternating Direction Method of Multipliers,” Proceedings of 2019 IEEE International Conference on Data Mining (ICDM’19), Nov. 2019, pp. 1354 - 1359.

Patents

  1. Xingguo Chen, Yiming Qiao, Wei Liu, Jie Zhu, “A User-Oriented Method for Enhancing Custom Sports Commentary,” Patent CN202010284204.8, China, Filed Apr. 2020, Granted Sept. 2023.

  2. Zhiqiang Zou, Linrui Li, Shuyu Chang, Yiming Qiao, “A Classification Method for Outlier Celestial Objects Based on Astronomical Spectral Data,” Patent CN202010983397.6, China, Filed Sept. 2020.

Awards

Teaching

  • Teaching Assistant - Quantitative Investment (Tsinghua 80470273) - Fall 2023
  • Teaching Assistant - Data Mining (Tsinghua 40470333) - Fall 2021