Interpreting Advertiser Intent in Sponsored Search

Abstract

Search engines derive revenue by displaying sponsored results along with organic results in response to user queries. In general, search engines run a per-query, on-line auction amongst interested advertisers to select sponsored results to display. In doing so, they must carefully balance the revenue derived from sponsored results against potential degradation in user experience due to less-relevant results. Hence, major search engines attempt to analyze the relevance of potential sponsored results to the user's query using supervised learning algorithms. Past work has employed a bag-of-words approach using features extracted from both the query and potential sponsored result to train the ranker.We show that using features that capture the advertiser's intent can significantly improve the performance of relevance ranking. In particular, we consider the ad keyword the advertiser submits as part of the auction process as a direct expression of intent. We leverage the search engine itself to interpret the ad keyword by submitting the ad keyword as an independent query and incorporating the results as features when determining the relevance of the advertiser's sponsored result to the user's original query. We achieve 43.2% improvement in precision-recall AUC over the best previously published baseline and 2.7% improvement in the production system of a large search engine.

Social media intention mining for sustainable information systems: categories, taxonomy, datasets and challenges

“…In [10], Statistical techniques based collaborative intent nowcasting model was developed to extract the complex relation between intent and context. In [91][92][93][94][95][96][97][98][99][100][101][102][103][104][105], multiple statistical techniques were used to proposed intention detection frameworks such as the S-O-R model base framework, decomposed theory of planned behavior, self-determination theory, TPB, TAM, CTO, regression correlation, composite reliability, variance extracted, AVE, t value, social cognitive theory and many more.…”

“…Izquierdo et al [98] addressed the issue of relating the advertisement to the user search query using the feature extraction mechanism as well as automate the TV ad schedule by mathematical programming. Social communities share information, ideas, opinions, and attitudes to make a suggestion or propose a solution to anyone's problem [99][100][101][102][103][104][105][106][107]. The statistical structured model used to detect tourist intentions, intention to use transportation, and intention to participate in online travel companies is addressed [45,84,108].…”

Where Can I Buy a Boulder?

“…They provide an estimate of the frequency of shoppingrelated queries, but their analysis does not investigate the distribution across product categories that we present in this article, nor do they focus on the important challenge of identifying locations from which the products could be purchased offline. Given our focus on products, prior work in areas such as sponsored search [6,13,31,37], local recommendations [27,21,22], and product search (primarily e-commerce related) (e.g., [25,36,38]) is also relevant to our research.…”

“…Web search engines are often used as a starting point for product purchases, including the purchase of services [5,34,39]. This has inspired research in sponsored search [6,13,31,37,43], local search [27,21,22], and product search [25,36,38] directly. Product purchases can happen online (as in e-commerce) or offline (in person).…”