The critical task of predicting clicks on search advertisements is typically addressed by learning from historical click data. When enough history is observed for a given query-ad pair, future clicks can be accurately modeled. However, based on the empirical distribution of queries, sufficient historical information is unavailable for many query-ad pairs. The sparsity of data for new and rare queries makes it difficult to accurately estimate clicks for a significant portion of typical search engine traffic. In this paper we provide analysis to motivate modeling approaches that can reduce the sparsity of the large space of user search queries. We then propose methods to improve click and relevance models for sponsored search by mining click behavior for partial user queries. We aggregate click history for individual query words, as well as for phrases extracted with a CRF model. The new models show significant improvement in clicks and revenue compared to state-of-the-art baselines trained on several months of query logs. Results are reported on live traffic of a commercial search engine, in addition to results from offline evaluation.

We describe a machine learning approach for predicting sponsored search ad relevance. Our baseline model incorporates basic features of text overlap and we then extend the model to learn from past user clicks on advertisements. We present a novel approach using translation models to learn user click propensity from sparse click logs. Our relevance predictions are then applied to multiple sponsored search applications in both offline editorial evaluations and live online user tests. The predicted relevance score is used to improve the quality of the search page in three areas: filtering low quality ads, more accurate ranking for ads, and optimized page placement of ads to reduce prominent placement of low relevance ads. We show significant gains across all three tasks.

Extraction of entities from ad creatives is an important problem that can benefit many computational advertising tasks. Supervised and semi-supervised solutions rely on labeled data which is expensive, time consuming, and difficult to procure for ad creatives. A small set of manually derived constraints on feature expectations over unlabeled data can be used to partially and probabilistically label large amounts of data. Utilizing recent work in constraint-based semi-supervised learning, this paper injects light weight supervision specified as these "constraints" into a semi-Markov conditional random field model of entity extraction in ad creatives. Relying solely on the constraints, the model is trained on a set of unlabeled ads using an online learning algorithm. We demonstrate significant accuracy improvements on a manually labeled test set as compared to a baseline dictionary approach. We also achieve accuracy that approaches a fully supervised classifier.

We present a probabilistic model of a user's search history and a target query reformulation. We derive a simple transitive similarity algorithm for disambiguating queries and improving history-based query reformulation accuracy. We compare the merits of this approach to other methods and present results on both examples assessed by human editors and on automatically-labeled click data.

We present a document expansion approach that uses Conditional Random Field (CRF) segmentation to automatically extract salient phrases from ad titles. We then supplement the ad document with query segments that are probable translations of the document phrases, as learned from a large commercial search engine's click logs. Our approach provides a significant improvement in DCG and interpolated precision and recall on a large set of human labeled query-ad pairs.

Discovering users' specific and implicit geographic intention in web search can greatly help satisfy users' information needs. We build a geo intent analysis system that uses minimal supervision to learn a model from large amounts of web-search logs for this discovery. We build a city language model, which is a probabilistic representation of the language surrounding the mention of a city in web queries. We use several features derived from these language models to: (1) identify users' implicit geo intent and pinpoint the city corresponding to this intent, (2) determine whether the geo-intent is localized around the users' current geographic location, (3) predict cities for queries that have a mention of an entity that is located in a specific place. Experimental results demonstrate the effectiveness of using features derived from the city language model. We find that (1) the system has over 90% precision and more than 74% accuracy for the task of detecting users' implicit city level geo intent (2) the system achieves more than 96% accuracy in determining whether implicit geo queries are local geo queries, neighbor region geo queries or none-of-these (3) the city language model can effectively retrieve cities in location-specific queries with high precision (88%) and recall (74%); human evaluation shows that the language model predicts city labels for location-specific queries with high accuracy (84.5%).