Chen-Kun Tsung

Correcting vindictive bidding behaviors in sponsored search auctions

In this study, we aim to develop a pricing mechanism that reduces the effects resulted by vindictive advertisers who bid on sponsored search auctions run by search engine providers. In particular, we aim to ensure payment fairness and price stability in these auctions. With the generalized second price principle, advertisers pay the next-ranked bid value rather than the price that they bid. Vindictive bidders take advantage of this principle to manipulate the payment of a previously-ranked advertiser. Vindictive bidding results in unfair outcomes and eliminates equilibria. However, it is difficult to compute rational payments for all advertisers as advertisers’ valuations are private. Our proposed mechanism decreases the payment to make up for the utility loss that is induced by vindictive bidding. The vindictive advertiser is simultaneously punished with an additional payment. According to our theoretical analyses and simulations, the proposed mechanism efficiently decreases the effects that result from vindictive bidding, and guarantees equilibrium outcomes.

Analysis of Stable Prices in Non-Decreasing Sponsored Search Auction

Most critical challenge of applying generalized second price (GSP) idea in multi-round sponsored search auction (SSA) is to prevent revenue loss for search engine provider (SEP). In this paper, we propose non-decreasing Sponsored Search Auction (NDSSA) to guarantee SEP’s revenue. Each advertiser’s bid increment is restricted by minimum increase price (MIP) in NDSSA. The MIP determination strategy influences bid convergence speed and SEP’s revenue. Fixed MIP strategy and Additive-Increase/Multiplicative-Decrease (AIMD) principle are applied to determine MIP values, and they are evaluated in this paper. For the convergence speed analysis, fixed MIP strategy converges faster than AIME in most instances. For SEP’s revenue, AIMD assists SEP to gain more revenue than fixed MIP strategy by experiments. Simultaneously, SEP’s revenue in Vickrey-Clarke-Groves auction (VCG) is the lower bound of that in AIMD.

Repeated sponsored search auction with non-decreasing bid values

We apply ascending biding behavior of English auction to solve the revenue loss problem of generalized second price auction (GSP) in repeated sponsored search auction (SSA) for the search engine provider (SEP). The mechanism proposed in this paper is denoted by Non-decreasing Sponsored Search Auction (NDSSA). We focus on analyzing SEPs revenue lower bound and auction efficiency after converging. We proof that SEPs revenue in NDSSA is at least equal to α times of that in Vickrey-Clarke-Groves auction (VCG), while α is the reciprocal of the first click-through rate (CTR). Since any CTR is normalized to a decimal, SEP in NDSSA will obtain more revenue than in VCG. Moreover, NDSSA is efficient, and advertisers with higher valuations will be ranked in better slots.

A Spectral Clustering Approach Based on Modularity Maximization for Community Detection Problem

Modularity is widely-used objective function to detect communities and there are lots of algorithms based on modularity maximization. The leading eigenvector method is one of them where modularity is maximized by choosing the first eigenvector as partition result. To analyze in depth the information provided by other eigenvectors, modularity maximization could be transformed to vector partitioning problem. This paper proposes a method to find non-overlapping vertex vector sets so as to maximize the quadratic sum of norms of community vectors. We observe spatial distribution of the vertex vectors of networks and then discover two phenomenons. First, the vertex vectors belong to different communities are separated by an angle. Second, the node with a larger degree would correspond to a vertex vector with a larger norm. Based on these two phenomena, we design a heuristic community detection algorithm. When a network with weaker community structure, the over-partition problem is considered. The experiment results show that the proposed solution provides higher accuracy than other solutions.

An Evaluation for Different Pricing Mechanisms under the Sponsored Search with Various Bidding Processes

In this paper, we build a sponsored search auction platform to simulate and analysis the auction result for auctioneers. We propose to use geometrically decreasing sequence to capture the click behavior of internet users. In our simulations, the results show that theoretical properties of the charging mechanisms are captured, and the proposed platform reflects the charging mechanisms property. Therefore, the auctioneer can use the proposed platform to simulate the auction and estimate the result in expectations.

A Game Theoretical Approach for Solving Winner Determination Problems

Determining the winners in combinatorial auctions to maximize the auctioneers revenue is an NP-complete problem. Computing an optimal solution requires huge computation time in some instances. In this paper, we apply three concepts of the game theory to design an approximation algorithm: the stability of the Nash equilibrium, the self-learning of the evolutionary game, and the mistake making of the trembling hand assumption. According to our simulation results, the proposed algorithm produces near-optimal solutions in terms of the auctioneers revenue. Moreover, reasonable computation time is another advantage of applying the proposed algorithm to the real-world services.

Strategic Bidding Behaviors in Nondecreasing Sponsored Search Auctions

To realize the specific results in the sponsored search auctions, most advertisers submit particular bid prices. The bidding behaviors with specific purposes are called as the strategic bidding. However, some strategic bidding behaviors will result in negative effects, such as the elimination of the equilibrium and the payment increase for some advertisers. The bidding behaviors with negative results are termed as the vindictive bidding. We survey four strategic bidding behaviors which include a rational bidding and three vindictive bidding strategies. In this paper, we study the relationship between the effects resulted by the vindictive bidding and the valuations of the vindictive advertisers. In our experiments, the search engine provider (SEP) is benefited by all vindictive bidding behaviors, and the increment of the SEPs revenue is proportional to the degree of the vindictiveness. Bidding vindictively without sacrificing the own utility improves the advertisers utility with high probability. Moreover, we observe that the SEPs revenue is improved by the following situations. First, the vindictive advertiser with low valuation in the keywords with high market value results in more SEPs revenue than that in the keywords with low market value. The second case is to raise the bidding competition between advertisers.

Sponsored search auction without bias

In this paper, we focus on finding out the parameter setting of the non-decreasing sponsored search auction (NDSSA) such that the produced outcome is satisfied by both the search engine provider (SEP) and advertisers. Six satisfaction factors are included in our problem formulation. The outcome satisfactions of the SEP and advertisers are modeled as the multi-objective optimization problem based on the satisfaction factors. The weighted versions of L2-norm, cosine similarity, and Euclidean distance are proposed to solve the multi-objective problem. We observe the outcome without bias for both participants in our simulation. Both cosine similarity and Euclidean distance show that SEP and advertisers prefer fixed-MIP to AIMD. Moreover, the initial MIP value should be set as small enough such that each advertiser has at least one opportunity to update his/her bid price.