MSearchGroove

I hinted at it yesterday and judging from the volume of emails, it’s going to be a topic of lively debate on this site and throughout the industry. I’m talking about recent research from Kartik Hosanagar, Wharton professor of operations and information management, which reveals the downside of content recommendations. I had the pleasure of meeting Kartik during RecSys, where I took the opportunity to deep-dive into the findings of his 43-page paper, Blockbuster Culture’s Next Rise or Fall: The Impact of Recommender Systems on Sales Diversity.

The research, which he completed in September, argues that recommenders “reinforce the blockbuster nature of media.” Put another way, they can limit the content exploration and discovery they were designed to encourage. “Because common recommenders recommend products based on sales and [consumer] ratings, they cannot recommend products with limited historical data, even if they would be rated favorably,” Kartik writes in his paper (which he co-wrote with Dan Fleder, a Wharton doctoral candidate). “This can create rich-get-richer effects for popular products and vice-versa for unpopular ones, which results in less diversity.”

I found this idea both intriguing and unsettling and quizzed Kartik on the impact on mobile, where recommenders will no doubt provide users with a useful shortcut to content and a means to by-pass tedious text input. Granted it’s early days, and his research did focus on online recommenders, but he is nonetheless convinced the same risks apply. Put simply, recommenders are good at pushing people to new content on mobile, but they may also push us toward the same new content. We’re discovering new items, but we’re all discovering the same items.

This effect of consumers being pushed toward similar products is shown in the figure below (excerpted from the research paper). In the graph, consumers are nodes equally spaced on the perimeter of a circle. Comparing the graphs, the increased density on the right shows that consumers have become more similar to one another in terms of products they consume. Put simply, sales diversity is limited and so is consumer choice.

But retailers shouldn’t give up on recommenders. To the contrary, recommenders, together with search, ensure that users can navigate the avalanche of digital content and services across the Internet and the mobile Web. Instead, companies should be careful of how they choose and design their recommenders.

AOL “gets it” and has come up with a design that not only pushes the envelope; it effectively solves problems associated with semantics and the pesky problems associated with telling technology that – when you express an interest in “bond” – you mean James and not finance, for example.

Jen Consalvo, AOL Director of Personalization, walked us through Mgnet, a content discovery and recommendation service with a differentiated design focused around an innovative “image cloud” of topics. (Mgnet is in public beta today in the U.S. – check it out at http://my.aol.com.) Put simply, rather then force users to grapple with text, the service lets users choose images. Click on images and stories you like, rate what you see and Mgnet will process this information to deliver personalized recommendations and bubble up sites and content you might not have found on your own.

Jen and I had lunch after the presentation and discussed the need for content discovery methods that deliver what users expect - plus a few surprises. In Jen’s view, serendipity is core – and this is reflected in Mgnet, which lets users in search of new stuff click “What else ya got” to get a new set of recommendations at any time.

The timeline for Mgnet isn’t set in stone, but plans are to roll it out to a global audience. AOL has already created an internal prototype of a similar content discovery and recommendation service for mobile, so it’s clear wireless access is next on the agenda. Smart move, since using images and the quality metadata associated with them is bound to overcome the poor usability common to all mobile devices. (Jen has kindly offered to put me in touch with the right people at AOL Mobile to discuss their vision for personalized recommendation, so you can count on me to circle back the moment I have a podcast in the bag.)

In related news, AOL took the wraps of a slew of new mobile services at CTIA. The main attraction is AOL MyMobile, a downloadable app for Windows Mobile phones that uses a carousel-like interface to access several AOL-owned services including AOL Mail, MapQuest and Moviefone. Yahoo Go Mobile also uses a carousel-like interface to give users access to services, so we can take this as a sure sign that AOL is going to give Yahoo a run for its money. (Actually, none of the services are brand-spanking new – but they do underline AOL’s determination to become a portal and compete on equal footing with Google & Co.)

AOL has also formally introduced a redesigned AOL.com portal page for mobile phones. (The beta has been available since September.) The portal includes much of the content that is available on the desktop version of the portal. AOL’s mobile search technology (which I reviewed here) delivers users a much more customized search experience, and emphasizes made-for mobile sites at the top of the results list. As of last week, AOL also includes paid listings (courtesy of mobile advertising and search company JumpTap) among the search results.

Other new mobile services from AOL include a shopping service that lets users enter a barcode number or product name and receive comparative prices for the product at various retailers. AOL Shopping Mobile also lets users employ click-to-call to find out whether a product is in stock or to make a purchase directly from their mobile phones.

Finally, AOL is bringing AIM instant messaging to every phone. With the new AIM via SMS service, users can text to AIMAIM to log in, send and receive text messages and connect with members of their buddy list. Regular text message charges apply, but there’s no extra charge for this service. A mobile widget is available (AIM users can download the application at http://www.where.com/aim) that lets users plot the locations of their AIM buddies, who also have the widget, on a map using GPS technology.

• Posted in
• Mobile Social Networking, Web 2.0, Recommendation, Personalization, Mobile Content, Mobile Search, Content Discovery

1 comment

share

These icons link to social bookmarking sites where readers can share and discover new web pages.

• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 
• 

permalink