Dr. Jun Wang

Are you looking for a challenging role within a dynamic and ambitious team, who are rapidly becoming the leaders in the home energy management sector? Do you have the ability and confidence to manage a strategic and technical project, crucial to support increased company growth?

The successful candidate will deliver a Knowledge Transfer Project (KTP) in partnership with PassivSystems Ltd, based in Newbury. PassivSystems Ltd (http://www.passivsystems.com/) develops and markets energy management and monitoring systems to home owners.

Under the co-supervision of Dr Jun Wang at UCL and Dr Edwin Carter, Senior Engineer at PassivSystems Ltd, the postholder will develop intelligent energy advisory and feedback data systems to improve home energy advice and feedback, using consumer profiles and energy consumption data to automatically recommend and optimise services and influence new product development.

This challenging position is based at PassivSystems in Newbury and will provide an ambitious and committed candidate with invaluable personal and career development. The post is funded for 24 months initially from October 2012.

For more information, please check here.

Online advertising has become a key source of revenue for both web search engines and online publishers. For them, the ability of allocating right ads to right webpages is critical because any mismatched ads would not only harm web users’ satisfactions but also lower the ad income. In this paper, we study how online publishers could optimally select ads to maximize their ad incomes over time. The conventional offline, content-based matching between webpages and ads is a fine start but cannot solve the problem completely because good matching does not necessarily lead to good payoff. Moreover, with the limited display impressions, we need to balance the need of selecting ads to learn true ad payoffs (exploration) with that of allocating ads to generate high immediate payoffs based on the current belief (exploitation). In this paper, we address the problem by employing Partially observable Markov decision processes (POMDPs) and discuss how to utilize the correlations of ads to improve the efficiency of the exploration and increase ad incomes in a long run. Our mathematical derivation shows that the belief states of correlated ads can be naturally updated using a formula similar to collaborative filtering. To test our model, a real world ad dataset from a major search engine is collected and categorized. Experimenting over the data, we provide an analyse of the effect of the underlying parameters, and demonstrate that our algorithms significantly outperform other strong baselines.

UCL invites expressions of interest in fully funded PhDs in Computer Science, Statistics and Psychology. The PhDs will involve working with a world leading company which has purchasing data from over 350 million people in 28 countries. They process and analyse over a billion rows of data each week and have customer data spanning 5 years.

The vast amount of loyalty card transaction data and market research attitudinal data allows a great depth of understanding in terms of what customers are doing and what they are thinking. These insights can be further validated through real‐life experiments, both online and in‐store.

A typical supermarket sells 50,000 items across 100 categories. When deciding on which product to buy factors such as: price, quality, brand, recommendation, offers etc. all play their part in the decision‐making process for a customer. By understanding the complex interaction of these factors customers can be treated as an individual and their shopping experience improved and made more relevant to them.

This is a fantastic chance to explore the cutting‐edge of extracting real‐life insight into customer behaviour from the increasingly massive and complex data available.

The various players of Internet advertising and the trading process

This paper is under submission.

• S. Yuan, A. Z. Abidin, M. Sloan, and J. Wang, "Internet Advertising: An Interplay among Advertisers, Online Publishers, Exchanges and Web Users," Under Submission, ArXiv e-prints http://arxiv.org/pdf/1206.1754v1.pdf, 2012. bibtex
  
  @ARTICLE{ipm2012,
  author = {Shuai Yuan and Ahmad Zainal Abidin and Marc Sloan and Jun Wang},
  title = {Internet Advertising: An Interplay among Advertisers, Online Publishers, Exchanges and Web Users},
  journal = {Under Submission, ArXiv e-prints http://arxiv.org/pdf/1206.1754v1.pdf},
  year = {2012},
  pdf={http://arxiv.org/pdf/1206.1754v1.pdf},
  }

The top-k retrieval problem aims to find the optimal set of k documents from a number of relevant documents given the user’s query.
The key issue is to balance the relevance and diversity of the top-k search results. In this paper, we address this problem using Facility Location
Analysis taken from Operations Research, where the locations of facilities are optimally chosen according to some criteria. We show how this
analysis technique is a generalization of state-of-the-art retrieval models for diversification (such as the Modern Portfolio Theory for Information
Retrieval), which treat the top-k search results like “obnoxious facilities” that should be dispersed as far as possible from each other. However,
Facility Location Analysis suggests that the top-k search results could be treated like “desirable facilities” to be placed as close as possible to their
customers. This leads to a new top-k retrieval model where the best representatives of the relevant documents are selected. In a series of
experiments conducted on two TREC diversity collections, we show that significant improvements can be made over the current state-of-the-art
through this alternative treatment of the top-k retrieval problem.

• G. Zuccon, L. Azzopardi, D. Zhang, and J. Wang, "Top-k Retrieval using Facility Location Analysis," in ECIR Best Paper, 2012. bibtex
  
  @inproceedings{ecir2012,
  author = {Guido Zuccon and Leif Azzopardi and Dell Zhang and Jun Wang},
  title = {Top-k Retrieval using Facility Location Analysis},
  booktitle = {ECIR Best Paper},
  year = {2012},
  url={http://www.dcs.bbk.ac.uk/~dell/publications/dellzhang_ecir2012.pdf}
}

Ad Prices in a Spot Market

Many online advertising slots are sold through bidding mechanisms by publishers and search engines. Highly affected by the dual force of supply and demand, the prices of advertising slots vary significantly over time. This then influences the businesses whose major revenues are driven by online advertising, particularly for publishers and search engines. To address the problem, we propose to sell the future ad- vertising slots via option contracts (also called ad options). The ad option can give its buyer the right to buy the future advertising slots at a prefixed price. The pricing model of ad options is developed in order to reduce the volatility of the income of publishers or search engines. Our experimental results confirm the validity of ad options and the embedded risk management mechanisms.

• J. Wang and B. Chen, "Selling Futures Online Advertising Slots via Option Contracts," in WWW Poster, 2012. bibtex
  
  @inproceedings{www2012-adoptions,
  author = {J. Wang and B. Chen},
  title = {Selling Futures Online Advertising Slots via Option Contracts},
  booktitle = {WWW Poster},
  year = {2012},
  pdf={http://web4.cs.ucl.ac.uk/staff/jun.wang/papers/2012-www-adoptions.pdf},
  }

The aim of a web-based recommender system is to provide highly accurate and up-to-date recommendations to its users; in

A mass attached to a spring and damper

practice, it will hope to retain its users over time. However, this raises unique challenges. To achieve complex goals such as keeping the recommender model up- to-date over time, we need to consider a number of external requirements. Generally, these requirements arise from the physical nature of

A controlled recommender system

the system, for instance the available computational resources. Ideally, we would like to design a system that does not deviate from the required outcome. Modeling such a system over time requires to describe the internal dynamics as a combination of the underlying recommender model and the its users’ behavior. We propose to solve this problem by applying the principles of modern con- trol theory—a powerful set of tools to deal with dynamical systems—to construct and maintain a stable and robust recommender system for dynamically evolving environments. In particular, we introduce a design principle by focusing on the dynamic relationship between the recommender sys- tem’s performance and the number of new training samples the system requires. This enables us to automate the control other external factors such as the system’s update frequency. We show that, by using a Proportional-Integral-Derivative controller, a recommender system is able to automatically and accurately estimate the required input to keep the out- put close to a pre-defined requirements. Our experiments on a standard rating dataset show that, by using a feedback loop between system performance and training, the trade- off between the effectiveness and efficiency of the system can be well maintained. We close by discussing the widespread applicability of our approach to a variety of scenarios that recommender systems face.

• T. Jambor, J. Wang, and N. Lathia, "Using Control Theory for Stable and Efficient Recommender Systems," in WWW, 2012. bibtex
  
  @inproceedings{www2012,
  author = {T. Jambor and J. Wang and N. Lathia},
  title = {Using Control Theory for Stable and Efficient Recommender Systems},
  booktitle = {WWW},
  year = {2012},
  pdf={http://web4.cs.ucl.ac.uk/staff/jun.wang/papers/2012-www12-control-cf.pdf},
  url={http://web4.cs.ucl.ac.uk/staff/jun.wang/blog/2012/02/16/control-theory-recommender/},
  }