In a yearlong effort, a team of our research scientists and engineers who specialize in machine learning created an advanced query-to-ad matching model for sponsored search. Today, search2vec accounts for more than 30% of all broad match impressions and revenue on Yahoo Search! Their new blog post provides highlights of their SIGIR 2016 paper describing search2vec and presents a new dataset available to researchers.
The EURO 2016 football tournament begins this week and we thought it would be fun to try to predict the results through a combination of big data, science, and social media. We used our expertise and unique access to data from Tumblr – as well as Yahoo Sports – to draw conclusions on who will be crowned the champions.
How do users perceive native ads and why? In her new post, Research Director Mounia Lalmas sheds light on some important insights based on her team's WWW2016 paper, "Predicting Pre-click Quality for Native Advertisements."
What is perhaps the best way to measure user attention for online news reading? In a new post from Director of Research Mounia Lalmas, she argues for viewport time.
Who spends the most online? When do people shop most online? How do people spend their money online? Our research scientists explored these shopping behavior questions in their paper, "Portrait of an Online Shopper: Understanding and Predicting Consumer Behavior," published this week at WSDM in San Francisco.
Recently we announced our efforts to make Yahoo a more focused company. This focus will let us accelerate the pace of innovation to make our products even better. We saw these changes as an opportunity to better align our research efforts, while preserving Yahoo’s culture of exploration and inquiry. As a result, we are reorganizing Yahoo Labs and moving forward with a new approach to research at Yahoo.
The Yahoo Flickr Creative Commons 100 Million (YFCC100M) dataset and the Multimedia Commons initiative are featured in the February edition of the Communications of the ACM magazine. The dataset and initiative aim to open and share community-contributed dataset features and ground truth annotations.