Searching for stuff is sometimes tough. If you know what you’re looking for, and you phrased your search term just right, then you usually get good results. But if not, you’re in big trouble, doomed to endless sifting through the results, page by page until you find the thing that you were really looking for.
Search engines are good at finding terms, expressions, and pieces of text. But that’s where their world ends: They don’t understand the meaning of the text they are searching for, and they know nothing about objects, entities or relationships. In addition, they are not designed to find stuff in groups, but search for a single object each time.
For example, let’s say you are interested in seeing video clips of songs from the Dire Straits album “Brothers in Arms”. If you search for “Dire Straits Brothers in Arms Album” on YouTube, you will get many links to video clips of the song “Brothers in Arms”, and some links to other songs in that album (if the album name appears in the clip description). If you are lucky, you’ll get a link to a playlist called “Dire Straits Brother in Arms Album” prepared by some user who manually searched for these tracks by name.
YouTube search results for "Dire Straits Brothers in Arms Album"
But now look what happens if you execute the same query through Headup: Headup automatically digs into its database to find the tracks in the album, and searches for specific video clips of these tracks. Then, it returns a nice “video wall” where each thumbnail links to a different track in the “Brothers in Arms” album. The key here is that Headup “knows” what an album is, associates it with its tracks, and is smart enough to understand that YouTube hosts mainly videos of tracks, not full albums. This type of reasoning and “smart search” implementation is way beyond the power of other “topic search” engines that do nothing more than search forwarding.
Headup video results for "Dire Straits Brothers in Arms Album"
Let’s take another example. What if you are searching for a certain type of product by a certain brand – such as Samsung LED-backlit LCD TVs, or Sony Flash-based HD camcorders. If you try these search terms in a regular search engines, you will get scattered results of news announcements, product reviews, and maybe a link to a specific product page. But you’ll never get a list of actual TVs or camcorders that match these criteria, since the search engines can only search for the text you supplied, but don’t understand it.
When such a search is conducted through Headup, it queries its knowledge graph for items that match the requested criteria. Since in Headup objects have meaning, properties and relations to other objects, it is quite easy to go through all the “Products” by the “Company” Sony, find the “Camcorder” type products, and filter only those items that have “Memory Type” equal Flash, and “Resolution” equal “HD”. So executing such a query through Headup may result, for example, in a neat list of links to specific product pages, which may include media reviews, user reviews and price comparison with purchasing links.
Note that even though Headup currently does not support direct search, the “smart search” method is already implemented in the current pop-up widget and topic pages. When you look at images, news or videos of a certain object or topic, Headup’s “smart search” works behind the scenes to bring you the most relevant content for that object, by understanding and utilizing its relationship to other objects.
