So, how many things have you liked today? Chances are that somewhere in the last 24 hours you’ve given a thumbs-up to a news article you came across on a friend’s Facebook post, a movie on Netflix, or a beer garden on Foursquare.

An application in beta from Cascaad, dubbed CircleMe, hopes to be the single source for hosting and managing all your likes.  “Typically you leave those traces all over the web but they aren’t leveraged,” says Erik Lumer, Cascaad founder and executive chairman. “It’s in your profile somewhere but you’re not getting much out of it.” Lumer says Cascaad is betting there’s value to help users manage the activity on their likes in one place, so that they can get more out of them such as more easily tracking new things underway that are connected to what they already like, or get recommendations from others with similar interests. And to do it with greater permanence, so to speak. As Lumer points out, you can potentially discover a new book on Facebook that one of your friends liked, but “two hours later it’s gone. There are hundreds of messages on top of it. There’s not a clean way to leverage that effectively, so in that sense I think we are very complementary” to Facebook likes.

Users can import likes from other services or add them directly from within CircleMe. What makes CircleMe semantic is the heritage brought to it from its initial two products, Cascaad Reader and Splice. That includes fast semantic analysis of content to extract topical relevance, which in Cascaad Reader is used to filter relevant entities in Twitter and Facebook feeds and surface new content based on personal interests. In the browser extension Splice, the semantic technology enriches feeds with relevant contextual information. The technology builds a semantic profile based on what users are interested in automatically and incrementally according to what they read, publish and share, and then extracts entities from Twitter and Facebook content that match. “We had to really optimize the custom entity extractor to work well on short pieces of text,” says Lumer, not to mention to process with great speed.

“It’s one thing to do that on a whole news article with a lot of text, redundancy, and context to disambiguate, but another to deal with tweets that basically have very little context, and grammar is virtually non-existent. The game there of matching entities requires much more fine-tuning and optimization for that kind of content.” It also leverages the social graph to understand whose content might be more relevant to a particular user.

Lumer notes that the base technology encompasses millions of well-defined entities at this point. A lot of work went into entity aggregation to understand, for instance, that the Cold Play being talked about on FourSquare and lastFM are one and the same. “In those two services these are different descriptions and we had to figure out they are the same, and remove the noise and duplication.” In CircleMe, such capabilities play into the discovery or recommendation process – as you start collecting things that you like, each disambiguated item has relations to other items. So, if you like a musician, that musician is related to a certain genre of music, and from there maps of new things consistent with a user’s interests can be drawn.

Interestingly, Cascaad Reader and Splice were thought of as being less consumer-oriented services than showcases for large content aggregators or social platforms that might be interested in white-labeling the technology. CircleMe, on the other hand, is a pure consumer play. “We felt that the B-to-B space has very interesting potential, but for what we were unique at there is only a few number of players of large size that can leverage our technology well,” says Lumer. Not to mention that some of them are pretty strong on the technology development front themselves. “So it’s an interesting market but a relatively limited one.”

The door’s still open to such opportunities, but to exclusively rely on them isn’t a particularly good idea for a VC-backed startup. “We also wanted to grow independently of those players,” says Lumer, and applying its knowledge and expertise to what it saw as a latent and unmet consumer need is the route it’s taking. Some things about CircleMe may remind you of other services out there, such as GetGlue, but Lumer says CircleMe is distinctive because for a couple of reasons. That includes the user experience and a focus on being a more broad, cross-category service rather than focused primarily on a single venue, such as entertainment. “You can express your likes across many categories, media content, passion, heroes, and build a very rich portrait, an extremely visually compelling portrait of what you like, and then keep up with it.”

By having all the like data in one place, users can search by category or date imported, for example. “But the real value proposition is around engaging with those. Once upu do that, or as you go along, you will collect more things you like,” he says. “It’s a way to keep up with new stories and new social activities around those likes and to discover new ones.”

For CircleMe, where in effect virtually everything you do is a statement of your affinity for a product or service, the business model proposition lies in its potential for sponsorships around items in the libraries of users, as well as commerce, such as facilitating access to transactions to buy tickets to shows by bands you have liked.

CircleMe right now is an invite-only Beta. Lumer expects it to move into open beta by year’s end.