Users contribute much data on the internet. They freely provide detailed information (i.e., data) about their activities, thoughts and opinions in social networks, blogs and product reviews, and much of that data is available to anyone else on the web. Corporations covering the gamut from gaming to news to fashion analyze user histories on their websites to understand how easy their websites are to navigate and which kinds of products and services are attractive to their customers. Advertisers may follow cookies to understand how display ads on different types of websites affect interest in their products. Users with nefarious interests who intend to infiltrate internal networks and so try to avoid leaving data tracks are obviously of interest to corporations and institutions with websites, and increasingly to the users who may be their targets. Moreover, users are contributing their own data to public websites in the interests of citizen science. Instead of just giving data passively through sensor networks over which they have no control, they share physical, social, personal and biological data they collect themselves, and often analyses of the shared data, with anyone who can access the global internet. This data-driven workshop will explore the challenges for inference, models, algorithms and graphical and analytical tools that these different aspects of user-centered modeling raise. The plan is to start with enabling, evaluating, and analyzing data that users actively contribute in citizen science, taking into account thorny issues like data aggregation, selection bias, data quality, and inferential uncertainty, then move on to data that users passively contribute or leave behind on the web, even when they are trying to hide, and asking the same questions in that context.