1、1 外文翻译 Warrington College of Business Material Source:IEEE presss,2007 Author:Juan Feng The practice of sponsored search advertising where advertisers pay a fee to appear alorgside particular Web search results is now one of the largest and fastest growing source of revenue for Web search engines. W
2、e model and compare several mechanisms for allocating sponsored slots, including stylized versions of those used by Overture and Google, the two biggest brokers of sponsored search. The performance of these mechanisms depends on the degree of correlation between providers willingness to pay and thei
3、r relevance to the search term. Ranking providers based on the product of relevance and bid price performs well and is robust across varying degrees of correlation. Ranking purely based on bid price fares nearly as well when bids and relevance are positively correlated (the expected regime), and is
4、further enhanced by adding an editorial filter. Regardless of the allocation mechanism, sponsored search revenues are lower when users attention decays quickly at lower ranks, emphasizing the need to develop better user interfaces and control features. The search engine can address initial inscience
5、 of relevance scores by modifying rank allocations over time as it observes clickthroughs at each rank. We propose a rank-revision strategy that weights clicks on lower ranked items more than clicks on higher ranked items. This method is shown to converge to the optimal (maximum revenue) ordering fa
6、ster and more consistently than other methods. Internet search engines index billions of Web pages and employ information retrieval algorithms to display, in response to a users query, links to Web pages deemed relevant to the query. These pages might represent commercial firms selling goods or serv
7、ices, information sites, government entities, etc. Lawrence and Giles (1999) estimated the publicly indexable Web at 800 million pages, containing 6 terabytes of text data on 2.8 million servers; today, Internet statistics indicate that the publicly indexable Web contains billions of pages, served b
8、y over 40 million Web servers. Due to this vast amount of information, search engines act as an 2 information gateway to many search and decision-making tasks. More than 50% of Web users visit a search engine every few days, the leading search engine (Google) gets over250 million search requests eac
9、h day, over 13% oftraffic to commercial sites was generated by searchengines, and over 40% of product searches on the Web were initiated via search engines. For us, the term search engine encompasses various applications of these indexing-retrieval technologies, including pure Web search engines (e.
10、g., Google), information portals with search functionality (e.g., Yahoo!), metasearch engines (e.g., Metacrawler), niche search engines (e.g., CiteSeer), and comparison shopping engines (e.g., mySiinon, S). Due to the critical influence of search engines on Web users actions, many commercial firms h
11、ave realized the importance of gaining a high position on the search results for specific queries. Entire niche industries exist touting services to boost a Web pages ranking on the popular search engines, in part by reverse engineering the search engines information retrieval algorithms. The expect
12、ation of increased traffic from good placement on a search page has led to the creation of a market for sponsored search (or paid placement we use the termsinterchangeably, as dictated by context) where search engines can charge a fee for prominent positioning witMn a “sponsored“ section in the resu
13、lts page. For example, a digital camera retailer may pay a search engine to appear among the sponsored results when users search for “digital cameras.“ Usually, sponsored results are shown on top of, or to the side of, the standard unpaid search results (also called algorithmic results). In general,
14、 sponsored listings are explicitly marked as sponsored results or advertising, and the FTC has advised search engines to disclose paid links appropriately, see Tantono et al. (2002). Figure 1 displays a screen shot with sponsored and algorithmic search results in an Internet search engine. advertisi
15、ng on the Internet, and acts as a broker between advertisers and information gatekeepers like Yahoo! and MSN. Overtures success has prompted several other companies to adopt similar business models involving paid placement, most prominently Google, the leading Internet search engine. Other sponsored
16、 search providers include LookSmart, FindWhat, and eSpotting. In 4, we report results of computational simulations of the equilibrium performance of these four mechanisms. Ranking by the product of bid price times clickthrough weakly dominates other mechanisms in all tested regimes, though the mecha
17、nism is statistically equal to ranking by bid price in the expected region of positive correlation between willingness to pay and relevance. Editorial filtering 3 helps the basic rank-by-bid mechanism significantly. Section 5 studies two alternative approaches for the dynamic ranking problem, where
18、the search engine periodically revises the ranking of listings within the paid slots based on its observations about user clickthroughs. Section 6 provides concluding remarks. The study of sponsored search is crucial to understanding the future design, quality levels, and market structure of Web sea
19、rch engines. We elaborate on the research focus of this paper and summarize other management problems and research related to paid placement. We begin by discussing essential characteristics of sponsored sea . The study of sponsored search is crucial to understanding the future design, quality level
20、s, and market structure of Web search engines. We elaborate on the research focus of this paper and summarize other management problems and research related to paid placement. We begin by discussing essential characteristics of sponsored sea banner advertising on the Web) is typically priced via an
21、informal process of estimation and negotiation. The Internet, however, supports more efficient and mechanized pricing via real-time auctions that can capture the advertisers true uillingness to pay, and track it over time. The sponsor-search industry typically runs separate auctions for different se
22、arch terms: For example, the search queries “plasma television“ and “investment advice“ are associated with two distinct auctions. Whats being sold in each auction is the right to appear alongside the results for that search query (actually Uhats being sold is the right to the users proceeding dicks
23、 if any). In practice, hundreds of thousands of advertisers compete for positions alongside several million search queries every day. So, for example, travel vendors like Expedia and Orbitz may compete in an auction for the right to appear alongside the result of a users search on “Las Vegas travel.
24、“ Generally auctions are dynamic, meaning that advertisers can change their bids at any time, and a new auction clears every time a user enters the search query. In this way, advertisers can adapt to changing environments, for example boosting their bids for “Harry Potter“ around the release date of
25、 the latest book in the series. 4 译文 Warrington College of Business 资料来源： IEEE presss,2007 作者： Juan Feng 实践的付费搜索广告，即客户支付一定的费用才显示搜索结果的广告，是现在网络搜索引擎收入的最大和增长最快的来源之一。我们模拟并比较分配了一些赞助机制，包括用于版本的程式化的若干 机制。 由序曲和谷歌两个最大的赞助商搜索经纪人。这些机制的性能取决于供应商之间的支付意愿及其相关的搜索词关联度。在不同强劲的关联度上，网上的排名和相关产品的供应商投标价格表现良好。排名纯粹以买入价以及票价近时的出价和
26、相关的是正相关（预计政权），并进一步通过添加一个编辑过滤器增强。不管分配机制，付费搜索收入较低时，用户的注意力迅速衰减职级较低，强调需要开发更好的用户界面和控制功能。搜索引擎可以通过修改时间的推移地址分配的相关性排名得分，因为它最初需要观察每个职级的点击率。我们提出了一个等级，调整策略，以较低的 权重比排名点击，点击多个项目上排名较高的项目。这种方法被证明收益达到最优（最大的收益）订购更快，更持续高于其他方法。 互联网搜索引擎的索引数十亿网页和运用信息检索算法，以显示响应用户的查询，以网页的链接被视为有关的查询。这些网页可能代表销售商品或服务，信息网站，政府机构等商业公司劳伦斯和 Giles（
27、 1999）估计在 800万页的公开可转载网页，载有 280万的服务器 6万亿字节的文本数据，今天，互联网统计数字可转位表明，公开网页中包含数十亿个网页，超过 4000 万的 Web 服务器服务。由于这种大量的信息，搜索引擎作为 一个信息门户，担任许多搜索和决策的任务。超过 50的网民每隔几天要访问一个搜索引擎，领先的搜索引擎（谷歌）搜索请求得超过 250万，每天以 13以上是用于产生商业领域，以及超过40的产品搜索启动了网络搜索引擎。对我们来说，搜索引擎这个术语包括各种应用程序，检索技术，包括纯粹的 Web 搜索引擎（例如，谷歌），与搜索功能（例如，雅虎），元搜索引擎（例如， Metacra
28、wler），利基搜索信息门户，引擎（例如， CiteSeer），比较购物引擎（例如， mySiinon， S）。 由于 Web用户行为 的搜索引擎关键性的影响，很多商业公司都清楚了对特定查询获得的搜索结果中的高地位的重要性。整个行业存在利基招徕服务，加大了对流行的搜索引擎网页的排名部分，逆向工程的搜索引擎的信息检索算法。流量的增加，从一个搜索页面上放置好，付费搜索市场是建立在搜索引擎可以收取一定的费用的基础上，突出一个定位 “赞助”一节中的结果页面。例如，5 数码相机零售商将付出巨大的搜索引擎中出现赞助商的结果，当用户搜索“数码相机”。通常情况下，赞助商的结果显示在顶部，或向一侧，标准未付费的
29、搜索结果（也称为演算结果）。在一般情况下，赞助商列表 被明确标记结果为赞助商或广告，和联邦贸易委员会已建议搜索引擎付费链接适当披露，看到 Tantono等。我们可以通过计算机模拟结果报告这四个机制的平衡性能。由投标价格乘以点击排行弱主导产品在所有测试制度等机制，但其机制是统计上平等地竞标价格居的地区之间的正相关预期的支付意愿和相关性。编辑的基本职级过滤有助于按投标机制显着。第 5节研究两个问题的动态排名，在搜索引擎内定期修正了对有关用户点击率的意见的基础上支付的各种途径插槽列表的排名。 该研究寻找赞助理解是至关重要的未来设计，质量水平和市场的网络搜索引擎的结构。我 们阐述了本文研究的重点和总结
30、其他管理问题，并研究相关的付费排名。我们首先讨论赞助海本质特征。 该研究寻找赞助理解是至关重要的未来设计，质量水平和市场的网络搜索引擎的结构。我们阐述了本文研究的重点和总结其他管理问题，并研究相关的付费排名。我们首先讨论在网络上发起海横幅广告本质特征）通常是通过非正式的估计和谈判过程中的价格。在网上，但是，支持更高效，机械化通过实时拍卖定价，可以捕捉到广告客户的真实支付，并随着时间的推移它的轨道。赞助商搜索行业通常运行不同的搜索条件单独拍卖：例如，搜索查询“等离子电视”和“投资咨 询”是两个不同的拍卖有关。什么是在每次拍卖出售的权利出现并列，搜索查询的结果（实际上 Uhat 的出售是对用户的权利，如果诉讼迪克斯的话）。在实践中，为成千上万的广告客户一起竞争职位数百万的搜索查询的每一天。因此，举例来说，像 Expedia 和 Orbitz 的旅游供应商可以在竞争拍卖的权利，同时出现了一个用户对搜索结果“拉斯维加斯旅行。”拍卖一般是动态的，也就是说，广告客户可以随时更改自己的出价，和一个新的拍卖清除每次用户输入搜索查询。通过这种方式，广告客户可以适应不断变化的，例如环境，提高周围的系列中的最新著作 发行日期为“哈利波特”他们的出价。
