Mei, Li have papers in premier data mining conference
UMSI associate professor Qiaozhu Mei and UMSI doctoral student Cheng Li are co-authors of papers being presented at the 21st ACM SIGKDD Conference on Knowledge Discovery and Data Mining, held August 10-13 in Sydney, Australia.
Cheng Li will present her paper, “Click-through Prediction for Advertising in Twitter Timeline,” at the Industry and Government Session: Business, Sales, Marketing, Advertising on Tuesday, August 11, 3:20 pm. Her paper was co-authored with Yue Lu, Twitter Inc; Qiaozhu Mei, University of Michigan; Dong Wang, Twitter Inc; and Sandeep Pandey, Twitter Inc.
Li’s research was conducted during an internship at Twitter. Her paper examines the challenges of placing ads in a Tweet stream, given the changing nature of the data, and proposes an improved method that results in a better click-through rate.
Qiaozhu Mei also co-authored the paper “PTE: Predictive Text Embedding through Large-scale Heterogeneous Text Networks,” to be presented during the Research Session: Clustering and Text, on Wednesday, August 12, 1 p.m. The paper is authored by Jian Tang, Microsoft Research Asia and Meng Qu, Peking University.
The paper proposes practical guidelines to follow when choosing between convolutional neural networks and predictive text embedding in various scenarios. Predictive text embedding utilizes both labeled and unlabeled data to learn the embedding of text.
“KDD is the very top of venue of data mining research,” said Mei. “Given the extremely harsh competition for papers, acceptance is a great achievement.”
SIGKDD is an international conference that brings together researchers and practitioners from data mining, knowledge discovery, data analytics and big data. KDD is a special interest group of the Association for Computing Machinery, the premier membership organization for computing professionals.