Video Content-Based Advertisement Recommendation System using Classification Technique of Machine Learning

Abstract

Abstract Content-based advertising is a method by which we advertise on a video media based on a relevant topic assigned to the video. In digital advertising, the advertisements shown to a user is based on the user’s behaviour on the internet. Streaming platforms are then used to target audience based on parameters like user’s geo-location, interests, watch history, age, etc. In most cases, the advertisements shown are not relevant; an undesired impact is created. Content-based advertising helps to convey the message with increased efficiency and simultaneously optimizes its conversion rate. In this proposed system we take the video metadata as input and apply the NLP techniques for text classification which categories the video and assigns a relevant advertisement to it. The second module takes the video as an input. Thereafter the video is converted into N individual frames to tackle the video classification as an image classification problem. In this proposed system we train a Convolutional Neural Network to identify the topic of the video on an image dataset and compare its performance with a pre-trained model. We create the image dataset by downloading images from the internet. We also create a video advertisement’s dataset by web scrapping. This proposed system makes sure that the user is shown the advertisement in reference to the video. This increases probability of the user visiting the client’s website.