Search engines have made a boodle for outfits like Google, but when it comes to online entertainment — especially entertainment on demand — they may find their relevance waning.
“Search engine algorithms are based on text, information and links,” Daren Gill, vice president for business development for ChoiceStream in Cambridge, Mass., told TechNewsWorld. “From an entertainment perspective, we found that’s really not the paradigm that’s likely going to win.”
“It’s hard to search for entertainment using the concept of an empty search box,” he maintained.
A Different Kind of Relevance
For foraging the Web, he continued, search engines do a good job of finding results that meet one-size-fits-all standards of relevance. “In the entertainment space, relevance is a personal decision,” he opined.
When it comes to consumer online behavior toward entertainment, he continued, “pushing” content to seekers based on a sophisticated assessment of their tastes is more appealing to them than generic indexing systems.
At video sites, Gill contended, relevant video should be pushed as soon as possible because consumers often aren’t thinking of their behavior in terms of searching.
“They’re looking for something to watch, to check out,” he explained. “So you need an intelligent system to push based on some relevance that should be driven by users’ preferences.”
Rich Classification System
A key to providing relevant results to entertainment seekers is a rich classification system for the entertainment itself.
To do that, ChoiceStream has built an index around the movie, TV and premium video content available to users.
The index contains more than the kind of information a search engine would key on. For example, it contains latent attributes: Is a movie dark? Smart? Mindless?
“What you’re going to see over the next few years is more metadata coming to the table and more consistent metadata about video,” Gill observed. “There’s not a lot out there now.”
Metadata is data that describes other data. For example, 02895, without a context, is just a series of numbers. Add the metadata “ZIP code,” and 02895 can be associated with a place — Woonsocket, R.I.
“That’s one of the challenges for companies like Google,” Gill argued. “They don’t work so well with meta data. They work with direct data.”
Both Yahoo and Google declined to comment to TechNewsWorld for this story.
Once you create an index of content, you still have to determine an individual user’s tastes. That’s usually done through some kind of “collaborative filtering.”
Collaborative filtering tries to determine a consumer’s preferences by comparing what he or she buys to what others who bought the same object bought.
One notorious flaw in that kind of system is that it doesn’t scale very well. If you’ve bought a hundred items, you may find what the system recommends to you somewhat relevant, but if you’ve only bought an item or two, you may find its suggestions a tad goofy.
Fine Tuning Choice
That’s why ChoiceStream developed a system that it claims can produce relevant results with very few data points. It does that by using metadata to assign attributes to things like movies, TV shows and music.
In addition, it can further refine a user’s preferences by comparing his or her choices to those of other users in its system.
“We use all of our users in the system across our various customers to help us understand the preferences,” Gill explained. “By having a lot of users in the system that are interacting with movies and television, it’s easy to take a relatively few data points and start to build a profile and deliver relevance back.”
Wave of the Future
The system has garnered kudos from consumers in some quarters.
“We’ve gotten very positive user response,” Craig Michaels, director of content programming for Akimbo in San Mateo, Calif., told TechNewsWorld. Akimbo is a provider of online on-demand video content.
“When something gets recommended, people are investing and trying it,” he said.
He maintained that as entertainment choices increase, solutions like the one provided by ChoiceStream will become ever more important.
“As there is more choice,” he said, “there is the need for more guided decisions. ChoiceStream is poised to do that.”