Computational Advertising: Overview and Key Techniques from the Second Edition

The second edition of Computational Advertising is presented as a comprehensive guide that bridges large‑scale search, text analysis, statistical modeling, machine learning, optimization, and micro‑economics to explain the product, problems, systems, and algorithms behind internet ad monetization.

The book has gained unexpected industry traction, being adopted by many internet companies for commercial training and even used as a prize in corporate events; a promotional giveaway invites readers to comment for a chance to win a copy.

Excerpted chapters detail core bidding‑price algorithms: CPC‑based pricing is illustrated with concrete code‑like descriptions, and the eCPM estimation formula r(a,u,c)=µ(a,u,c)·ν(a,u) emphasizes the need for accurate click‑through‑rate (CTR) prediction and click‑value estimation.

Query expansion techniques are surveyed, including recommendation‑based collaborative filtering on session‑query matrices, model‑based dimensionality reduction, and history‑effect methods that leverage past eCPM performance to suggest high‑value queries.

Ad placement is framed as a constrained optimization problem that balances the number of ads shown in the “north” and “east” slots while maximizing overall revenue, with personalization adjustments based on user tolerance for ads.

Author biographies are provided: Liu Peng, Vice President of iFlytek and head of its Big Data Research Institute, with a background in AI research at Microsoft and Yahoo; and Wang Chao, former ad engineer at Weibo and Autohome, now a senior R&D architect at Baidu focusing on personalized recommendation.

The article concludes with references to additional study notes, a link to further reading, and a brief promotion of the DataFun platform for big‑data and AI knowledge sharing.