Technologies and applications for consumer digital media are evolving rapidly. Examples of these technologies are digital still and video
cameras, multimedia personal computers, broadband multimedia networks, and recordable CD/DVD. These technologies enable consumers to create and access ever-increasing amounts of content, from a wide variety of sources and formats. As a result, there are significant challenges to be overcome to effectively organize and access this media information.

The Multimedia Organization and Retrieval project at HP Laboratories was formed with the goal of making multimedia content accessible to non-expert users, by providing the architectural and technological basis for organizing and retrieving multimedia content in natural terms. Investigation areas included:

1. Metadata representation, generation, and query;
2. Retrieval methods (search, browse, and visualize) for large information sets, based on the use of rich metadata about content;
3. How to create effective annotation processes that model the user's goals and utilize automatic feature extractors to minimize the burden of annotation.
4. Understanding how multimedia content and associated metadata can be shared and retrieved over the Web.

To pursue our goal, we analyzed the strengths and weaknesses of existing multimedia information retrieval approaches, conducted user research to understand the consumer’s perspective, and built an experimental platform to explore and evaluate technology considerations. FotoFile, a Java-based multimedia content retrieval system, is the project's initial proof-of-concept prototype. FotoFile runs on both Win95 and NT, and it embodies most of the concepts the team is researching.

In this talk, I will describe the design and implementation of the FotoFile system, discuss our user research activities that informed our design, and identify the technology and usability issues related to the work.