An advertisement is a notice or announcement in a public medium promoting a product. Advertising plays significant role in the introduction of a new product in the market. It stimulates the people to purchase the product. In this paper, we propose a novel personalized Online Video Advertising System which is presented to recommend product ads from ecommerce sites to users of online video hosting. First we have to find a user preference, so we have to analyze each user’s behavior of ecommerce sites. If the user wants to buy a product, user will spend more time to view the specification at the same time, user clicks that product more number of times to view the specification, price etc. These techniques are used to find the user preference. After collecting each user preference, we have to identify the semantic association between videos and products and construct the association between the key frames and products. A multi view deep learning approach is brought to view item features in different domains. When the User plays the video the user preferred advertisements are shown in the video in proper timestamps. Thus the advertisements are displayed only based on the user’s preference and the user’s will to purchase the desired product. This has been the key feature of our project. Maintaining privacy is one of the major problems for sharing personal information through social sites. Sharing video in social media may lead to unwanted problem and less privacy. As a result we need some tools for secured transmission. For satisfying this need, we propose a system called Adaptive Privacy Policy Prediction (A3P) to enhance privacy settings for user’s data. This system provides a two-level framework for securing the data based on users browsing history on shopping site. It determines the best available privacy policy for users. Our solution relies on image classification which is associated with policies to upload images and also to user's social features. Hence by using this adaptive privacy policy our sharing of shopping and recommended videos can be secured to only desire people.