Compact Web browsing profiles for click-through rate prediction

Bjarne Ørum Fruergaard; Lars Kai Hansen

doi:10.1109/MLSP.2014.6958852

Compact Web browsing profiles for click-through rate prediction

Bjarne Ørum Fruergaard, Lars Kai Hansen

Adform ApS

Research output: Chapter in Book/Report/Conference proceeding › Article in proceedings › Research › peer-review

Abstract

In real time advertising we are interested in finding features that improve click-through rate prediction. One source of available information is the bipartite graph of websites previously engaged by identifiable users. In this work, we investigate three different decompositions of such a graph with varying degrees of sparsity in the representations. The decompositions that we consider are SVD, NMF, and IRM. To quantify the utility, we measure the performances of these representations when used as features in a sparse logistic regression model for click-through rate prediction. We recommend the IRM bipartite clustering features as they provide the most compact representation of browsing patterns and yield the best performance.

Original language	English
Title of host publication	Proceedings of IEEE 24th International Workshop on Machine Learning for Signal Processing (MLSP)
Number of pages	6
Publisher	IEEE
Publication date	2014
Pages	1-6
ISBN (Electronic)	9781479936946
DOIs	https://doi.org/10.1109/MLSP.2014.6958852
Publication status	Published - 2014
Event	2014 IEEE International Workshop on Machine Learning for Signal Processing - Reims Centre des Congrès, Reims, France Duration: 21 Sept 2014 → 24 Sept 2014 Conference number: 24 https://ieeexplore.ieee.org/xpl/conhome/6945945/proceeding

Conference

Conference	2014 IEEE International Workshop on Machine Learning for Signal Processing
Number	24
Location	Reims Centre des Congrès
Country/Territory	France
City	Reims
Period	21/09/2014 → 24/09/2014
Internet address	https://ieeexplore.ieee.org/xpl/conhome/6945945/proceeding

Keywords

Bioengineering
Communication, Networking and Broadcast Technologies
Computing and Processing
Engineering Profession
Signal Processing and Analysis
Computational modeling
Data models
Logistics
Matrix decomposition
Predictive models
Uniform resource locators
Vectors

Access to Document

10.1109/MLSP.2014.6958852

OpenUrl availability

Full text

Cite this

@inproceedings{f3f62c5012fd409a8cf22b02a95b8cb8,

title = "Compact Web browsing profiles for click-through rate prediction",

abstract = "In real time advertising we are interested in finding features that improve click-through rate prediction. One source of available information is the bipartite graph of websites previously engaged by identifiable users. In this work, we investigate three different decompositions of such a graph with varying degrees of sparsity in the representations. The decompositions that we consider are SVD, NMF, and IRM. To quantify the utility, we measure the performances of these representations when used as features in a sparse logistic regression model for click-through rate prediction. We recommend the IRM bipartite clustering features as they provide the most compact representation of browsing patterns and yield the best performance.",

keywords = "Bioengineering, Communication, Networking and Broadcast Technologies, Computing and Processing, Engineering Profession, Signal Processing and Analysis, Computational modeling, Data models, Logistics, Matrix decomposition, Predictive models, Uniform resource locators, Vectors",

author = "Fruergaard, {Bjarne {\O}rum} and Hansen, {Lars Kai}",

year = "2014",

doi = "10.1109/MLSP.2014.6958852",

language = "English",

pages = "1--6",

booktitle = "Proceedings of IEEE 24th International Workshop on Machine Learning for Signal Processing (MLSP)",

publisher = "IEEE",

address = "United States",

note = "2014 IEEE International Workshop on Machine Learning for Signal Processing, MLSP 2014 ; Conference date: 21-09-2014 Through 24-09-2014",

url = "https://ieeexplore.ieee.org/xpl/conhome/6945945/proceeding",

}

TY - GEN

T1 - Compact Web browsing profiles for click-through rate prediction

AU - Fruergaard, Bjarne Ørum

AU - Hansen, Lars Kai

N1 - Conference code: 24

PY - 2014

Y1 - 2014

N2 - In real time advertising we are interested in finding features that improve click-through rate prediction. One source of available information is the bipartite graph of websites previously engaged by identifiable users. In this work, we investigate three different decompositions of such a graph with varying degrees of sparsity in the representations. The decompositions that we consider are SVD, NMF, and IRM. To quantify the utility, we measure the performances of these representations when used as features in a sparse logistic regression model for click-through rate prediction. We recommend the IRM bipartite clustering features as they provide the most compact representation of browsing patterns and yield the best performance.

AB - In real time advertising we are interested in finding features that improve click-through rate prediction. One source of available information is the bipartite graph of websites previously engaged by identifiable users. In this work, we investigate three different decompositions of such a graph with varying degrees of sparsity in the representations. The decompositions that we consider are SVD, NMF, and IRM. To quantify the utility, we measure the performances of these representations when used as features in a sparse logistic regression model for click-through rate prediction. We recommend the IRM bipartite clustering features as they provide the most compact representation of browsing patterns and yield the best performance.

KW - Bioengineering

KW - Communication, Networking and Broadcast Technologies

KW - Computing and Processing

KW - Engineering Profession

KW - Signal Processing and Analysis

KW - Computational modeling

KW - Data models

KW - Logistics

KW - Matrix decomposition

KW - Predictive models

KW - Uniform resource locators

KW - Vectors

U2 - 10.1109/MLSP.2014.6958852

DO - 10.1109/MLSP.2014.6958852

M3 - Article in proceedings

SP - 1

EP - 6

BT - Proceedings of IEEE 24th International Workshop on Machine Learning for Signal Processing (MLSP)

PB - IEEE

T2 - 2014 IEEE International Workshop on Machine Learning for Signal Processing

Y2 - 21 September 2014 through 24 September 2014

ER -

Compact Web browsing profiles for click-through rate prediction

Abstract

Conference

Keywords

Access to Document

OpenUrl availability

Fingerprint

Cite this