Semantic social search engine TipTop Search has plans in the works to build up its shopping product. Launched at the end of last year, the shopping portal on the site directs queries to Amazon, much as its health, movies and other social search products plumb Twitter’s depths. It brings back specific products in the category of interest (handbags, perfumes, etc.) that pulls together ratings for key attributes such as overall quality and value for the money based on its extraction and analysis of the unstructured data for that entity culled from user reviews about it.

“If you do a search on Amazon you get price, star ratings, and so on, but in our case, because we have this technology we ‘read’ the reviews and use that to summarize how good the products are. That is what makes TipTop shopping different,” says Shyam Kapur
president and CEO of TipTop Technologies. Users can get a grid-view into selected products’ attributes (albeit somewhat limited to the above mentioned ones at this point) to understand how specific items line up against each other, and click on each to get the human input behind the data scores – the Top Tips, i.e. positive opinions and sentiments, and Top piTs (the not-so-good input that infuses the scores). “You can see the corresponding Tips and piTs snippets that are also extracted for comparison shopping without having to read through hundreds or thousands of reviews,” Kapur says. There might be 50 reviews for the thousand matches that could turn up for a shopping query on ‘boots,’ for instance, and no one has the time to process through all those 50,000 reviews, he says.

It’s on the roadmap to show more attributes and ratings than are currently accounted for now in the shopping space, and also to make them more explicit to the particular category than generic across all categories, on the idea that you probably care about different things when you’re buying toys than you do when you’re buying CDs, for instance.

While it’s still early on in the development of the shopping product, Kapur says to look for advances in TipTop’s comparison capabilities, so that users can compare any product to any product across different searches and different pages in a space it creates – and that what starts in the shopping context will “expand to anything else in the world that you want to compare. That is a unique notion,” he says.

Right now TipTop is keeping its infrastructure light, so it’s sticking with interfacing to services like Twitter and Amazon whose APIs enable it to do real-time searches without the need for storing data. “But clearly the goal is of course to make it such that a user can put in any data they want that they think will help them compare,” Kapur says. “At the moment we make decisions for the user – we’re saying if you’re looking for current events, and so on, Twitter is good. If you’re looking for products, we use Amazon. But the future vision is where the user in a sense works with us and by their behavior, as individuals or in aggregate, we figure out if you’re looking for movies [for example], these are the five best sources of information to be helpful in making decisions, so we get that data for you. The notion is that data evolves into whatever is appropriate for that particular issue, problem, or situation.”

Fundamentally, Kapur says, TipTop Search is really an analytics platform, so the direction is to give users insight about whatever decision they are trying to make and whatever the source might be. “The other part that is very importan is that we are social -- search is no longer just about passive information. Search has to be an interactive experience where you get some information, but also people you can talk to, who are in the same frame of mind, who share the same TOPS, or topics. So that’s exactly wherewe are headed.” Kapur says.

The user interface is one area that will likely see changes as TipTop moves forward. “The challenge is I know a lot more than what I am able to show because I still haven’t figured out the UI frankly, and how to make it work best for the average user,” he says. He’s not worried about TipTop’s prospects and advancing its capabilities even as big search competitors dive into more semantics-driven worlds such as real-time search. He sees them as being somewhat hampered by their scale and investment in legacy retrieval paradigms. “For my small startup it’s just a few changes, we’re not doing the whole web, and we don’t need to worry about scale right away,” he says. “They have to invent the best algorithms and make sure it’s scalable across the web so it will take them longer. I’m sure they will but they can’t do it as nimbly as it can happen in a startup like TipTop.”