Google, meet Facebook's OpenGraph Search

With the introduction of the OpenGraph Protocol, Facebook introduces a new concept of searching throughout the Web. Facebook's search works based on what they call "connections" between resources from their OG ontology. Their algorithms are capable of discovering related items to the input query, based on the individual's social neighborhood or perhaps on the frequency of hitting the (now famous) "Like" button. Sounds like a bundle of possibilities, doesn't it ?

FaceRank, the Social Relevance Algorithm

Of course, this is not something Facebook officially announced (it would sound corny, don't you think ?), but the point is, after long 10 years, finally there is a serious candidate to best the PageRank, or at least complement with it. But Google works fine, the whole world searches, people are happy! Why would anyone use Facebook's new lab gadget instead of tested, proven, mature, lightning-fast and precise tool? Well because, there are queries that Google Search simply cannot satisfy! Moreover, their results are based on statistical methods, no people are involved there. What makes Facebook different is the capability to deliver real-time results , fresh and relevant , without deploying complex calculations . If some event is popular, people will rapidly talk about it. Same as with Google, it will be up to the web masters to annotate their web pages with the metadata, but the key differential factor here is that Facebook has the feedback from the users. It can use the number of "Like" hits to give weight to popularity of some particular web page. What if someone puts false metadata? (One of the biggest problems in the Semantic Web, too). In this case, the answer is simple: people will not like it, they will simply ignore it if it is misleading, hence it will be less popular and will have lower positioning. Another advantage from using this approach is that metadata now contains the context of the resource, opening the gates for bringing the conventional Semantic Web Dream . Facebook  is now able to interpret user's query, does she search for related books, movies, sport teams, people... you name it, it finds it... in real time. As written in Times: Google, This Time, Its Personal.

Hey Mark, Recommend Me a Movie, Please

When someone says: "Yeah, the idea of the Semantic Web is great, but if it so wonderful, how come there are no applications to massively leverage it? You say the technology is available for a while.", usually made some point, but I think not anymore. With Facebook's ultimate way of Social Bookmarking, it becomes easily calculable of what users could want, on individual level ! How, you may ask ?

Here is what I am at. (This may be a real idea for semantic application, too). Suppose you want to watch a movie, but you are not really sure what you want to watch... Naturally, you would ask your friends or you would search through the Internet a bit to see where is the movie hype cloud at the moment... (did you realize I said, "at the moment"? Hang on.). Now imagine a widget, that simply communicates the Facebook via OpenGraph API, to check what movies do you like. The widget also supposes that since you like those movies, you have probably watched them, so it makes no sense to suggest them to you again. But how difficult it is, to write a query that says:

"Give me the most popular movies that are related to the comedies I like". We define "related to" as a simple rule: "A movie is related to another if X people that watched the first movie also watched the second. The movie gains ranking in relatedness if at least Y of that people are my friends. The movie gains ranking if there are at least Z pages with more than 50 likes on the Web". 

Hmmm, not so difficult to be written in a query language. For now some of these aspects are not covered in the OpenGraph ontology (I refer to the Movie Genre), but undoubtly, it could easily be added. On the other side, for the application user, it is as simple as logging in to Facebook, and pressing the "Recommend" button. Welcome to the Semantic reality, Neo. Btw, how do you write "My favorite movies" in Google ? :)

But appart from the interesting search ideas the OpenGraph brings, my deepest beliefs are that Facebook's reason number one to introduce this protocol has e-Marketing roots i.e. to deliberately interfere with Google's primary business model - with personalized, perfect ad targeting tool .

What do you think ? Will this Facebook API bring new methods of warfare between the web titans ? Will it provide better searching for end-users ? Will ultimately, data find us ? How will Google eventually respond ? Is this the final gate that needed to be opened, for semantic applications to be massively written ?

Facebook and the Semantic Web: Weaving the Social or the Advertising Graph?

You probably already heard about the Facebook's new OpenGraph Protocol. It represents a new way of making connections between topics people like around the web, thus embedding metadata within the webpages itself. Why is Facebook doing this ? Does it want (really) to act as a social hub platform for bridging the Semantic Web to reality?

Finally a big player enters the Semantic Web realm. One that is recognizable all over the world. One that people have confidence in. One that promises to be powerful enough, to integrate topics from different webpages and connect them to corresponding people. One that will get rid of the chicken-and-the-egg vicious circle of Semantic annotation and Semantic Applications. One graph to rule them all : Facebook's OpenGraph.

Facebook, the Chicken and the Egg

Facebook apparently is trying to motivate webmasters to start embedding semantics into webpages, similarly to how meta keywords and meta description tags are embedded today for SEO. That would eventually give the desired push and stable ground for Semantic Applications to be finally built. People are already familiar with this way of embedding metadata, thus the motivation for them lies in the fact that Facebook will utilize that metadata whenever someone puts the mouse over the link that describes how a person "likes" something. But what is happening in background ? Is this simplified mapping to Facebook's ontology one step driven by the desire for people to share what they really like around different platforms ?  Does Facebook have hidden intentions in this whole story ?

The Impact on Ordinary Users

Well, what do average users get from the Social Graph ? Of course, they could leverage this new feature in order to spread the word about services/products they prefer or offer, providing additional fuel to marketing in Social Media. From that aspect, users will get even more specific recommendations from friends about things that might interest them. Of course, friends have similar interests and there is a good chance that they will at least be intrigued about what one's friends like. Moreover, "like" web sites that aggregate Facebook page titles and groups have begin to emerge. Some users find this aggregation amusing.

Facebook Flaws in Semantics : Why ?

As it was recently published in a post on Read Write Web , Facebook did leave flaws in embedding semantics in web pages. Some of them are known to Semantic Web enthusiasts from long time ago, such as the ambiguity problem when identifying resources. In terms of the OpenGraph protocol, there is no means to denote that two resources on the Web refer to actually the same thing. Therefore, integration between heterogeneous systems is not easy at all. Secondly, items with same names refer to same things although they point to different terms. This means there is no way to denote that a page is relevant to the car Jaguar, not the animal jaguar. Furthermore, the OpenGraph leaves no way to build relations between resources, assuming that the only relation is : is_relevant_to . This relation applies to web pages and items and items to people, respectively. This conclusion comes since there is no way to embed multiple objects into a single web page.

The Open Advertising Protocol

This is not something that Facebook publicly says, but if one gets into little deeper thinking, becomes obvious. Facebook is not concerned about allowing people brag to the others what they like. The company is concerned about mapping the users' interests in another graph, which I take the freedom to name it Open Advertising Protocol. It refers to a graph that will try to make connections between topics that might interest the user and her social graph, individually and in groups. What this means is the following: Facebook is trying to gain information about the meaning of the things because it needs more precise targeting for its personalized ads! It is fairly simple. Every time a user presses the "Like" button, Facebook gains insight on that user's interests! By having this knowledgebase at hand, Facebook will soon have enough data to improve their Ad targeting algorithm. What is even scarier, even if one does not press the "Like" button, they will be able to map your interests roughly based on your friend's interests! One might think: Fine, then people will eventually stop hitting that button once they realize this. But hold on a second! Facebook was created to fulfill a human need for social interaction, an interaction that was not satisfied by any other media before ! The point is, people are not that inert as one might think! "Like" it or not, the Big Face will be able to find out who we are, who we hang out with, what are we interested in aaaand ... what companies have better chance of selling to us!

What do you think? What is the reason for Facebook to enter this Semantic Web Game ? Why does it leave flaws, although it has both knowledge and infrastructure to make it differently? Does it really want help people share things they like or this is just a preparation for the Perfect Advertising Tool and even bigger profit ?