Advertising Science experts at Yahoo Research work closely with Yahoo's advertising product and engineering teams to monetize Yahoo's and publishers’ content by connecting advertisers and users in the right context. We use large-scale statistical modeling, data mining, and optimization techniques to develop algorithms for serving the most relevant ads to users and provide value to advertisers. We also understand marketplaces and mechanism design, auctions, and other important related economic phenomena.
Computer Vision at Yahoo Research tackles some of the most interesting real-world problems in image and video understanding. Powered by Yahoo’s massive repository of multimedia content, we conduct research and development in a variety of areas including visual recognition, visual saliency detection, vision/language, visual aesthetics and creativity; and, we do so by applying our expertise in computer vision, machine learning, and deep learning. Our work has direct impact on many of Yahoo’s flagship products by powering core technologies such as multimedia metadata enrichment, search and recommendation, video highlight detection and summarization, and search by visual content.
Web Mail is a massive and complex ecosystem that offers unique challenges and opportunities in varied areas of computer science research such as personal search, data mining, scalable systems, advertisement, social computing or security/privacy. Yahoo Researchers have been pioneering this area in academic research by publishing their results at top research conferences. They are, in parallel, contributing novel solutions and features to Yahoo Mail and providing deep insights upon which future products and features could be built.
Metrics and User Engagement researchers at Yahoo Research collaborate with our Search and Personalization teams to study and define metrics for experimentation, user engagement, and product success. Statistical analysis and data-driven user understanding methods are employed to enable the optimization of products through rigorous experimentation, in addition to generative and exporatory statistical analysis. Researchers help to guide product development by unlocking large-scale data to provide deep, grounded analysis. We also aid Yahoo as a whole in driving data integrity and clean data practices. Our passion is a user-centric approach to Big Data, ensuring long-term happy and engaged users.
Natural Language Processing and Information Extraction at Yahoo Research involves close collaboration with the core content processing platforms across Search, Media, Mail, and Ads. The focus is on creating innovative, Web-scale solutions for the whole range of content processing including text mining, classification and clustering, structured and unstructured information extraction, and summarization. A key aim of the research is to enable fully- or semi-automated methods that scale to the Web, self-correct and improve automatically over time.
Machine Learning and Data Mining efforts at Yahoo Research focus on research and development of production-quality solutions to important data-intensive problems at Yahoo, including algorithms, frameworks and tools for large scale machine learning, data visualization, metrics reporting, and analytics. We live and breathe Big Data every day, and work to apply the most leading-edge machine learning technologies to Yahoo's toughest large-scale problems.
Search and Data Mining research at Yahoo Research involves regular Web search (including content acquisition, content understanding, query analysis, and content ranking) as well as mobile and mail search. As part of content acquisition, our research focuses on efficient means for crawling the Web, ensuring freshness, and performing topic-focused crawls or deep crawls. Content understanding focuses on document structural analysis, information extraction, categorization and clustering, etc. Query analysis and content ranking focus on how to predict user intent and how to match and rank the most relevant information satisfying user needs.
At Yahoo Research, Personalization joins traditional search and recommendation with the emerging idea of personalization within real-time experiences on mobile devices using rich context, adding anticipation to the mix. Anticipation depends on "hyper-personalized" and "hyper-contextualized" data, for example, as in predicting in real time that a person is about to begin their nightly jogging routine. A system could then open the user's favorite jogging app and execute the “start a run” action, and at the same time open a favorite music app and tune to a well-suited local radio station. Key problems in this area include contextual profiling, context-based intent prediction, personalized recommendations and real-time federated search.