I was invading other people's privacy using Graph Search the other day, and I hit a wall when I tried this:

So apparently Graph Search doesn't do incest.

I work on Graph Search and have been part of the Search Team for nearly two years(!) now, so I've been along for the ride as the product evolved from a small prototype to being fully rolled out to the hundreds of millions of users in English_US.

This is a big question that has been answered in detail by many on my talented team who specialize in various parts of the Graph Search stack. Here is some recommended reading:

Rolling out the Beta:
Under the Hood: Building Graph Search Beta

NLP Interface (my team!):
Under the Hood: The natural language interface of Graph Search

Indexing and Ranking:
Under the Hood: Indexing and ranking in Graph Search

Infra:
Under the Hood: Building out the infrastructure for Graph Search

Entities:
Under the Hood: The entities graph

I think it's little more than another typical Facebook cash grab. It's powered by Facebook's likes, which are already of dubious quality. With that in mind, more likes will always equal higher placement in the search results, so businesses can simply buy their way to the top by spending more of their marketing dollars on Facebook. That's not so great for the people actually using the search, but Facebook thinks they'll make a fortune off of it. Most Facebook likes are ill-gotten, they don't signify that a user actually likes something. This has been a problem on Facebook for a long time, and rather than fixing it before using the likes to power a search feature, they're trying to cash in on the problem, which by design will make the problem worse. The more successful (as defined by it's ability to make Facebook money) Graph Search becomes, the poorer the quality of its results will become. And the sad part is that users will likely have no idea when they're searching for something that the results are prioritized not by relevance to what they were searching for, but by how much someone was willing to pay to get their result in front of you.

Whew.

I'm just going to copy-paste chunks of my own Forrester blog post (Testing Facebook Graph Search: First Thoughts) here; see that post for more detail / examples and screenshots of my test of the service.

It’s clear this will be a powerful feature, but it feels like something Facebook should have built some time ago. What I predict to be the most common searches, such as “which of my friends live in London” or “people my friends are friends with who work at Ford Motors,” are powerful, but basic, features that users have been requesting for a long time. The first rollout will also be missing obvious road map features, including the ability to search for links and status updates that you or your graph have posted.

The success of any individual Graph Search reflects what data (and activities) users directly provide Facebook, and today, many of the online activities that Graph Search encompasses take place on other social properties. Facebook often facilitates the social graphs of the other social properties with Facebook Connect, but Graph Search cannot “see” into that data. The average Facebook “like” is also less meaningful than Facebook's development team hopes, as others have blogged.

Therefore, the Graph Search is most successful with Facebook’s core use features, including queries about where you work, where you live, how you’re connected to other people, and photos. Besides the successful searches I mention above, I threw “photos of my friends with bacon” at it and got useful results. However, it falls very flat with queries about music or restaurants, which were two things featured in the announcement. It did not return quality results for “restaurants in San Francisco my friends like,” since my graph of that information is contained on Yelp and Foursquare (which both use Facebook Connect). Similarly, searching for music my friends like returns either music groups my friends are active in or nothing at all, while Spotify would know the right exact answer and again uses Facebook Connect. (See screenshots in the linked blog post for examples of each query).

For Marketers:
There was no component specifically for marketers in the first beta product; however, if implemented well, this is an opportunity for Facebook to offer more targeted ads to marketers oriented around purchase intent in the same manner that has served Google AdWords well for years.

Many search queries, such as “photos of my friends with bacon” will not be relevant for marketers; however, there will be plenty of queries, ranging from travel (“my friends who live in London”) to job postings (“people my friends are friends with who work at Ford Motors”) with clear advertising potential.

You can't install graph search. Facebook is a web and app product so if you have Facebook you can access these features.

So far I've only seen queries like "Friends who live in San Francisco" trigger it on web but not on the iPhone app.

I would guess it is because of your language settings on Facebook. There is a place to designate which languages you want (or don’t want) translated. I would edit these settings if you want to prevent pages from being translated.

First, check your language settings again.

1. Click on the downwards arrow

2. Choose Settings

3. Make sure that language in which you want to use Facebook is English.

When folks hear the term "graph" they think of a chart with x/y corrdinates. That is not what's meant in graph data or graph databases. When Google or Facebook or Twitter say they use a "graph" or allow "graph search" or "knowledge graph" they mean the data is represented via "nodes" and connected via named arcs (lines)... like below. (Compare this with SQL tables and keys) Graph database are built on graph alegebra.

Disclaimer First: This answer will be mostly based on my understanding of the graph searching mechanism, instead of a specialized Facebook one since I don't know much about that.

Now to the answer:
Graphs are inherently visual. It’s not so difficult to understand how the technology works, even if you’re not that technical. Let’s take one of Facebook’s example Graph Search queries, which is to find all of the New York sushi restaurants that my friends like.
Getting the answer is a very simple matter for a graph database. You just need to formulate the question in a way that the database understands. Those who are more technically inclined can see an example below for the query that answers the question: “find New York sushi restaurants that my friends like”:

START philip=node:person(name = 'Philip'),city=node:location(name='New York') MATCH (philip)-[:IS_FRIEND_OF]->(friend)-[:LIKES]->(restaurant)-[:LOCATED_IN]->city WHERE restaurant.cuisine = "sushi" RETURN http://restaurant.name

Query Example: “Find New York Sushi Restaurants that my friends like”

Thinking in graphs is natural, and contagious. The more you think in terms of connections, the more you realize that graphs are the way that we implicitly think. What is a decision tree, for example, but a graph of possibilities? The more you look, the more you start to notice that graphs are, in fact, everywhere. Graph database users regularly use queries like the one above to answer questions, and the more you ask, the more you think of new questions that never occurred to you to ask previously. Graph queries can get quite elaborate, and it’s entirely possible to run queries that scan across a social network that is two, three, or more levels of friends removed.

1. When the Facebook page opens, you'll see the Facebook search box in the top left corner.
2. Type anything in the search box and hit Enter on your keyboard.
3. Now the advanced search page opens, giving you access to 11 search categories in the menu to the left:
4. Posts – Look for your friends’ posts or those mentioning your friends.
5. People – Find people based on location, education, or workplace.
6. Photos – Search for photos by type, location, year, or even person (the poster).
7. Videos – Look for videos by the date, location, or whether it’s an FB live.
8. Marketplace – This category allows you to search for products that are available on Facebook’s market. Please note that not all users can access this option, depending on the marketplace’s availability in your country
9. Pages – Use various filters to narrow down specific pages. For example, you can search for products or business pages and also stores from which you can purchase products and services.
10. Places – Here you can look for places like restaurants, clubs, takeout places, and more. You also get a map of your location for more convenient searching.
11. Groups – Narrow down groups by location, private or public status, and your membership status.
12. Apps – This category doesn’t have any detailed filters.
13. Events – Choose if you’re looking for an online or physical event. Set the location, set how many days in the future you’d like to search, and define what kind of event you’re looking for. Finally, you can select if you’re looking for a family-friendly event and if it’s popular with your friends.
14. Links – Just like the “Apps” category, this one doesn’t have additional filters.

While Facebook showed its demo working from a Facebook domain, the biggest game changer in this product is that Facebook has signaled that it now wants to enable the entire Web to give better personalized content, recommendations, etc. to subjective questions AND Facebook will be providing everyone the means to do so.

When Facebook enables Graph Search to occur via API across the web, any site would be able to more readily give poignant content and recommendations.

Facebook is essentially working toward giving all developers across the web an outsource-able analysis/recommendation engine based on users' behaviors and preferences across their online presence.

Facebook seems to be truly working toward enabling private data to be presented and utilized in a public-friendly way.

There are big product and business implications for many companies, particularly those consumer web companies providing personalized services.

For example, because of the ease of ability to implement, providing personalized content and recommendations will increasingly become must-have features, not nice-to-haves, since the competition will. A corollary becomes that personalized recommendations and content will become more commoditized; thus, companies will have harder times distinguishing themselves through this feature alone.

What's in it for Facebook?

- A much tighter ecosystem, defined by every player becoming more dependent on Facebook.

- Facebook will know more about users.

- Facebook will have unique advertising channels.

In all, it'll take some time for everything to be available, but I'm a big believer that Graph Search is a game changer.

With the old search, if you want to find out all female friends of your friend, You have to go to his account and look for his females friends. There is no way to see only female friends as a list

But with graph search you can just type Female friends of <Friend's Name> in the search box from your home page itself and it will show all female friends of your friend. You don't have to go to his Facebook page or you don't have to click anywhere. Just type and hit enter. If you want to find out all female friends of your friend that is not your friends and studying in the college that you and your friend studying. You can find out that with Graph search.

Let us take an another simple example.
My friend living at Los Angeles said that he need a graphic artist to design his website but have no time to search. So what? how can I help? I'm living at a small village in India with a laptop, internet connection and my Facebook account wide opened. I don't know anyone at that place except my friend and I don't know where is LA, what is LA and How it is looks like.

No I couldn't help earlier. but now I have graph search. Just search 'Graphic artists in Los Angeles' and hit Enter. Bhoom!
More than thousand Graphic Artists who are working at LA in front of me just a message away distance.
I don't know anyone but some of them have mutual friends with me. So I can message them about this. Then I can also message to those mutual friends to know more about these artists. Select the one that got green signal from those mutual friends and I can give my friend's phone number and asked them to call him.

So now you know how powerful is Graph Search.

You can do very complicated searches very easily with a lot of parameters in Graph Search. It is so powerful. It will become more easy to search and connect with people once it is fully released.

In graph search you will only get the information that is shared as public or shared with you.

You can filter your results with all these parameters. See below