SalAd: A Multimodal Approach for Contextual Video Advertising

Chen, Xiang; Nguyên, Tam; Kankanhalli, Mohan

doi:10.1109/ism.2015.75

Cited by 15 publications

(12 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…None was concerned with advertising inventory prediction, which is one main issue in the advertising market. Xiang et al [29] conducted a study that proposed an advertising system using textual information and visual content to connect and provide advertisements most appropriate for online video. However, this study relied on an experimental method and did not use actual user data and predicted advertisements based on digital content [29].…”

Section: B Literature Reviewmentioning

confidence: 99%

Prediction of Online Video Advertising Inventory Based on TV Programs: A Deep Learning Approach

Lee

Yoon²,

Kim

2021

IEEE Access

View full text Add to dashboard Cite

With the recent spread of digital content, patterns of media viewing have changed. This is especially true for programs formerly watched on TV but are now increasingly viewed through online videos. As more and more people watch online videos, the market for online video advertising is increasing. Including online video advertising, online advertising can be effective if advertisers and online service providers attract as many viewers as possible. In particular, service providers try to maximize their profits by efficiently selling advertising inventory, which indicates the volume of space available for advertisements. However, most of today's service providers use simple statistical applications to predict advertising inventory that leads to relatively inaccurate predictions. Therefore, this study aims to develop a model capable of accurately predicting advertising inventory and then validate the model. This study in predicting online video advertising inventory is based on using deep learning to analyze the raw data of online video channels and then comparing the results of these predictions with actual inventory, other results of machine learning techniques, and work-site method results. Using these techniques and approaches, future advertising inventory can be more accurately predicted. In addition, detailed strategies for the practice of online video advertising are suggested. INDEX TERMS advertising inventory, online video advertising, TV programs, deep learning, predictionThis work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.

show abstract

Section: B Literature Reviewmentioning

confidence: 99%

Prediction of Online Video Advertising Inventory Based on TV Programs: A Deep Learning Approach

Lee

Yoon²,

Kim

2021

IEEE Access

View full text Add to dashboard Cite

show abstract

“…Finally select the most salient advertisement among the relevant advertisements as the most appropriate one. [7] 3. METHODOLOGY…”

Section: Related Workmentioning

confidence: 99%

“…For every term in the current document, and every document in the set, compute the Tf-Idf score. 7. Select the document with highest Tf-Idf Score.…”

Section: Related Workmentioning

confidence: 99%

Advertisement and Document Recommendation based on Content in the Image

Chandak¹,

S.²

2018

IJCA

View full text Add to dashboard Cite

In recent years, more and more images have been uploaded and published on the web. Along with text web pages, images have become an important media for various social media platforms to place relevant advertisements. However, conventional image advertising primarily uses text content rather than image content to match relevant advertisements. There is no existing system to automatically monetize the opportunities brought by individual image. As a result, the advertisements are only generally relevant to the entire web page rather than specific to images it contained. To overcome this, advertisements in the proposed system are recommended based on images. The objects are detected from the image using TensorFlow API Model and based on those objects (keywords) advertisements are recommended. An additional application is provided, were based on the detected objects (keywords) relevant documents are recommended using Term Frequency-Inverse Document Frequency algorithm. From the experimental results, it is seen that system could recognize over 90 percent of objects and could recommend relevant advertisement with mean average precision of 0.66.

show abstract

“…However, manually selecting an ad to match a provided multimedia content is time-consuming and labor-intensive. Thus some automatic advertising techniques are developed [4,7,22,41], such as the contextual advertising, which aims to find the most relevant ad to a provided content without affecting customers' watching experience (as shown in Figure 1) [21,27,35]. Therefore, it is necessary to understand the ad content thoroughly.…”

Section: Introductionmentioning

confidence: 99%

“…For example, [21] proposed a framework to insert ads into videos based on global textual relevance gathered from video metadata and local visual-aural relevance gathered from low-level image and audio features. A similar system is presented in [35], but a more advanced deep learning technique is introduced to analyze both video content and ad images. Different from these topic-only analysis work, in [33], the authors developed an ads recommendation system based on the sentiment of multimedia contents, and a unified framework to understand both topic and sentiment of the multimedia contents is developed in [19].…”

Section: Introductionmentioning

confidence: 99%

Look, Read and Feel: Benchmarking Ads Understanding with Multimodal Multitask Learning

Zhang

Luo

et al. 2020

Proceedings of the 28th ACM International Conference on Multimedia

View full text Add to dashboard Cite

Given the massive market of advertising and the sharply increasing online multimedia content (such as videos), it is now fashionable to promote advertisements (ads) together with the multimedia content. However, manually finding relevant ads to match the provided content is labor-intensive, and hence some automatic advertising techniques are developed. Since ads are usually hard to understand only according to its visual appearance due to the contained visual metaphor, some other modalities, such as the contained texts, should be exploited for understanding. To further improve user experience, it is necessary to understand both the ads' topic and sentiment. This motivates us to develop a novel deep multimodal multitask framework that integrates multiple modalities to achieve effective topic and sentiment prediction simultaneously for ads understanding. In particular, in our framework termed Deep 2 Ad, we first extract multimodal information from ads and learn high-level and comparable representations. The visual metaphor of the ad is decoded in an unsupervised manner. The obtained representations are then fed into the proposed hierarchical multimodal attention modules to learn task-specific representations for final prediction. A multitask loss function is also designed to jointly train both the topic and sentiment prediction models in an end-to-end manner, where bottom-layer parameters are shared to alleviate over-fitting. We conduct extensive experiments on a large-scale advertisement dataset and achieve state-of-the-art performance for both prediction tasks. The obtained results could be utilized as a benchmark for ads understanding.

show abstract

SalAd: A Multimodal Approach for Contextual Video Advertising

Cited by 15 publications

References 16 publications

Prediction of Online Video Advertising Inventory Based on TV Programs: A Deep Learning Approach

Prediction of Online Video Advertising Inventory Based on TV Programs: A Deep Learning Approach

Advertisement and Document Recommendation based on Content in the Image

Look, Read and Feel: Benchmarking Ads Understanding with Multimodal Multitask Learning

Contact Info

Product

Resources

About