Plenty of ink has been spilled detailing how wildly succesful Apple’s App Store has been, which now counts well over 100,000 applications. However, this huge number of apps presents a challenge: it’s hard to actually find interesting new apps when there are so many to choose from. Apple does its best to highlight quality applications in the store, and it offers a Genius feature, but that often leaves something to be desired. AppStoreHQ is a site that’s looking to help make app recommendations using a new source of data: Twitter. The site generates recommendations based on who is tweeting about iPhone apps, and can generate recommendations for you based on your own tweets.
The site looks at Twitter’s public stream, checking tweets for links back to the App Store. Every time someone links to an app, they treat that as a vote from that user. When you visit the site’s homepage, you can browse through the apps that are currently the hottest on Twitter or on the web (the latter is determined by apps with the most blog post mentions). Or, if you want personalised recommendations, you can tweet about an app or two that you enjoy. The one catch is that you have to include one of a few keywords like “iPhone” or “AppStoreHQ” in your tweet, otherwise the system won’t catch it. The service only resolves links with these keywords, because it would take a prohibitive amount of processing power to follow every short URL and Link to see if they’re linking to an iPhone app.
Once you’ve tweeted about a few apps, the system will make customized suggestions based on what other similar users have tweeted about. Obviously the more apps you tweet about, the better the recommendations become (which is also a good way for the service to grow virally). The system also takes into account a few other sources of data, including any apps you’ve Favorited on AppStoreHQ and which apps you’ve actually clicked through while browsing the site.
CEO Chris DeVore says that while the service could develop a sentiment engine to ensure that apps being linked to in tweets are actually being shared with a positive comment, it’s actually very rare for people to tweet a link to an app they don’t like, so they don’t do this processing.
I like the idea of using tweets for app recommendations — writing a tweet isn’t as involved as writing a blog post so the site will probably be able to get more data to analyze, and it will also be able to surface hot new apps quickly. That said, the algorithm will also likely have to put up with lots of spam from Twitter. And I think the AppStoreHQ layout could use some work, as its menu system seems clunky and bland.
AppStoreHQ has actually been around for a while, offering recommendations based on app links in blog posts. It also has a search engine for iPhone apps that it has white-labeled and is currently integrated in a number of other popular iPhone app news and community sites.