Learning to Advertise: How Many Ads Are Enough?

Wang, Bo; Li, Zhaonan; Tang, Jie; Zhang, Kuo; Chen, Songcan; Ru, Liyun

doi:10.1007/978-3-642-20847-8_42

Cited by 13 publications

(8 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This motivated research that investigated whether any ads should be displayed at all [3]. It is also easy to see that a similar and related problem formulation is to determine how many ads should be displayed to the users [16]. Moreover, determining the optimal number of ranked results is also important in a number of other IR applications such as legal e-discovery [14], where there is an significant financial or labor cost associated with reviewing results.…”

Section: Related Workmentioning

confidence: 99%

“…While much of the work in information retrieval has been centered around ranking, there is growing interest in methods for ranked list truncation -the problem of determining the appropriate cutoff of candidate results [1,9]. This problem has garnered attention in fields like legal search [14] and sponsored search [3,16], where there could be a monetary cost for users looking into an irrelevant tail of documents or where showing too many irrelevant ads could result in ad blindness. The fundamental importance of this problem has led to development of methods that are automatically able to learn in a data-driven fashion [9].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Choppy: Cut Transformer for Ranked List Truncation

Bahri

Tay

Zheng

et al. 2020

Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

Work in information retrieval has traditionally focused on ranking and relevance: given a query, return some number of results ordered by relevance to the user. However, the problem of determining how many results to return, i.e. how to optimally truncate the ranked result list, has received less attention despite being of critical importance in a range of applications. Such truncation is a balancing act between the overall relevance, or usefulness of the results, with the user cost of processing more results. In this work, we propose Choppy, an assumption-free model based on the widely successful Transformer architecture, to the ranked list truncation problem. Needing nothing more than the relevance scores of the results, the model uses a powerful multi-head attention mechanism to directly optimize any user-defined IR metric. We show Choppy improves upon recent state-of-the-art methods. CCS CONCEPTS • Information systems → Presentation of retrieval results; Retrieval effectiveness.

show abstract

Section: Related Workmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Choppy: Cut Transformer for Ranked List Truncation

Bahri

Tay

Zheng

et al. 2020

Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval

View full text Add to dashboard Cite

show abstract

“…As a result, there has been research that investigated whether it is possible to determine whether any ads should be returned to the user or not (Broder et al, 2008). A similar problem formulation investigated how many ads should be returned to the users (Wang et al, 2011a). Determining the optimal number of results to return is also important in a number of other search tasks, including legal e-discovery (Tomlinson et al, 2007), where there is an immense cost associated with reviewing results.…”

Section: Modeling Relevance Scoresmentioning

confidence: 99%

Surprise: Result List Truncation via Extreme Value Theory

Bahri¹,

Zheng²,

Tay³

et al. 2020

Preprint

View full text Add to dashboard Cite

Work in information retrieval has largely been centered around ranking and relevance: given a query, return some number of results ordered by relevance to the user. The problem of result list truncation, or where to truncate the ranked list of results, however, has received less attention despite being crucial in a variety of applications. Such truncation is a balancing act between the overall relevance, or usefulness of the results, with the user cost of processing more results. Result list truncation can be challenging because relevance scores are often not wellcalibrated. This is particularly true in large-scale IR systems where documents and queries are embedded in the same metric space and a query's nearest document neighbors are returned during inference. Here, relevance is inversely proportional to the distance between the query and candidate document, but what distance constitutes relevance varies from query to query and changes dynamically as more documents are added to the index. In this work, we propose Surprise scoring, a statistical method that leverages the Generalized Pareto Distribution that arises in Extreme Value Theory to produce interpretable and calibrated relevance scores at query time using nothing more than the ranked scores. We demonstrate its effectiveness on the result list truncation task across image, text, and IR datasets and compare it to both classical and recent baselines. We draw connections to hypothesis testing and p-values.

show abstract

“…Moreover, Dembczyński et al demonstrated how their suggested algorithms can be used to provide recommendations in order to improve the ads' quality. Later in 2011, Wang et al [33] attempted to predict the ideal number of ads that should be displayed for a given searchengine query.…”

Section: Ad Success Predictionmentioning

confidence: 99%

Exploring Online Ad Images Using a Deep Convolutional Neural Network Approach

Fire

Schler

2017

2017 IEEE International Conference on Internet of Things (iThings) and IEEE Green Computing and Communications (GreenCom) and I

View full text Add to dashboard Cite

Online advertising is a huge, rapidly growing advertising market in today's world. One common form of online advertising is using image ads. A decision is made (often in real time) every time a user sees an ad, and the advertiser is eager to determine the best ad to display. Consequently, many algorithms have been developed that calculate the optimal ad to show to the current user at the present time. Typically, these algorithms focus on variations of the ad, optimizing among different properties such as background color, image size, or set of images. However, there is a more fundamental layer. Our study looks at new qualities of ads that can be determined before an ad is shown (rather than online optimization) and defines which ads are most likely to be successful.We present a set of novel algorithms that utilize deep-learning image processing, machine learning, and graph theory to investigate online advertising and to construct prediction models which can foresee an image ad's success. We evaluated our algorithms on a dataset with over 260,000 ad images, as well as a smaller dataset specifically related to the automotive industry, and we succeeded in constructing regression models for ad image click rate prediction. The obtained results emphasize the great potential of using deep-learning algorithms to effectively and efficiently analyze image ads and to create better and more innovative online ads. Moreover, the algorithms presented in this paper can help predict ad success and can be applied to analyze other large-scale image corpora. 7 Each image in our dataset has its own unique MD5 hash. However, it is worth noticing that the same ad can still appear multiple times with minor changes, such as same images with different resolutions, different file formats, different colors, or even very minor changes in the ads' texts.8 Brand can be a company, or a specific product of a company. 9 The image ads in the auto-ads dataset were created in a separate process than the ad images in the all-ads dataset. Nevertheless, both datasets share 25,611 ad images with unique MD5 hash.

show abstract

Learning to Advertise: How Many Ads Are Enough?

Cited by 13 publications

References 20 publications

Choppy: Cut Transformer for Ranked List Truncation

Choppy: Cut Transformer for Ranked List Truncation

Surprise: Result List Truncation via Extreme Value Theory

Exploring Online Ad Images Using a Deep Convolutional Neural Network Approach

Contact Info

Product

Resources

About