Akimbo and

ChoiceStream's technology will power the recommendations. The growing inventory of digital content makes it important for download services to provide these tools that enable consumers to browse titles and discover new media each time.

The service to personalize recommendations requires technology that can determine not only what people like, but understand why they like it. "It is a way to help consumers find not only what they're looking for, but content they may not known they're looking for, other programming we offer as part of the service," said Jim Funk, vice president product and consumer marketing at Akimbo, a subscription services that offers nearly 9,000 programs from which to choose.

Personalization services can do a better job at sifting through mounds of media to recommend content if the technology supporting them understands how consumers think about media content and knows the underlying attributes attracting the consumers.

ChoiceStream's platform, written in C++, relies on "Attributized Bayesian Choice Modeling," with a layer of attributes, to provide personalization. ABCM uses techniques to classify content and products, linking to attributes people show the most interest in.

Compare ABCM with collaborative filtering, which requires more data to return accurate results, said Daren Gill, ChoiceStream's vice president of business development. "Our technology isn't limited to personalizing movie content," he said. "Akimbo has an IPTV solution, a <a = "http: www.techweb.com encyclopedia defineterm.jhtml?term="set-top+box"">set-top box that

delivers media through an IP cable, with content outside the realm of movies."