Facebook is changing the graph and introducing a new, keyword-based search engine. This may result a way to run more complex Boolean syntaxes in the future, however, as an immediate effect, has blocked the old-way graph search possibilities. Remember the story when we asked questions like ‘people who studied at xyz’? Yep, that has totally gone with the new search engine…
This change has created the need to summarize the old-way graph search method as that is still running – but in the back-end. You though have to memorize and use the below commands to retrieve the results but hey! that is not a rocket science just bear with me for some time.
In my last graph search post I was mainly focusing on leveraging FB IDs in the search. Now I am showing you how to use the /str command (and so) to retrieve the most and best of the old keyword-based graph search.
Before all, learn this /intersect command. This is the heart of the system!
The graph is built on set theory and intersection is the method how it generates results. A set is anything which members have something in common. For example, ‘people who speak English’ is a certain set. ‘People who speak Spanish’ is a different set. ‘People who speak both English and Spanish’ is the intersection of these two previous sets. The /intersect command helps you create the cut (the mutual set) between two sets.
The graph is reading everything in a linear way, always. You have to apply a very simple way of command creation: only one thing after one thing and then intersect them to get the mutual set. Do not try to make it too complex – even if you are searching complex things. The way how you get there has to remain as simple as possible.
If you apply this logic carefully you will be able to look for cases like ‘people who speak German and are any type of engineers and have ever visited any type of NASA entities while currently working for Boeing’.
Its graph is indicated on the left – the target person is marked with the ‘sun’ icon in the middle. Wherever you see two sets being intersected by each other that is where you will have to build in the /intersect command in your URL.
So this little command is probably the most powerful one and I try to explain why…
If you are running a certain ID-based search you always narrow your results to some specific cases. For instance, you want to find Google employees and run this one: /104958162837/employees/present/intersect where the long number is the Facebook ID of Google. With this search you find only those people who are using this exact (104958162837) Google page as their employer. You miss all the other Google companies that use different IDs.
The turnaround can be to use the /str command and run a search like this: /str/google/pages-named/ employees/current/intersect. With this command you get every person who uses Google as a word on their ‘Work and Education’ profile (that can be either their employer or a job title).
The basic structure of the /str command is as follows – remember this one, you can use this command in all the cases below!
- /str at the beginning and then
- add /your keyword(s) and then
- add /pages-named.
What follows this basic structure will depend on the rest of your query… but let us see some samples below!
The /employees command can be helpful in two cases:
1. This is how you find employees of a certain company: /str/your company name/pages-named (as above!) and then /employees/intersect.
You can search for current and past and ‘ever’ (anytime) employees by adding the /present /past and /ever commands after the /employees.
2. This is how you can search for job title with the /str command: str/your job title/pages-named/employees/intersect. I recommend you to use one-word-long job title keywords such as engineer or java or recruiter or so…
Note1: we use these commands to locate employer or job title searches, however, Facebook is doing nothing else than looking for these keywords on the ‘Work and Education’ tab under the ‘About’ section. So if someone’s job title contains the word ‘Google’ they will appear in the Google ‘employer’ search.
Note2: Facebook is not running a verbatim search. In some cases, Facebook is bringing false positives (aka synonyms). On the right, run a search for prezi with /str/prezi/pages-named/employees/intersect and you will get profiles containing the word ‘presentation’ but not ‘prezi’.
3. Now it is time to combine these two ‘strings’. Pretty basically, you just put together these two commands each after each to get the results.
Here you go some current employees from Facebook who are former engineers. These are people who currently work for Osram but are former Philips employees. These guys have worked for both Google, Yahoo and Facebook (uhmmm….most of them are probably fake profiles, what do you think?).
This is certainly an obvious and a rather practical search term. Use the /str again like: /str/your language/pages-named/speakers.
Again, you can combine all the above to find people who speak Spanish and deal with android as a current Google employee.
This was a very useful filter in the old graph (and I am totally confused why this has gone…).
Use the /str to target certain colleges and/or universities and/or just certifications. The way must be already familiar: /str/your uni/edu keyword/pages-named/students.
Curious to see how many Cisco engineers Avaya has on Facebook? Click here then!
Note: you can add general keywords, too, to the studies. Use this /str/mba/pages-named/students to get everyone who mentions MBA on their education profile. Or use this /str/college/pages-named/students to get every type of colleges.
Regarding the schools another helpful search can be to see all the universities and colleges in a certain location. This is where you can use the /places command.
/108051929285833/places/106078429431815/places-in/intersect – this brings every educational institute’s Facebook page out of London, UK.
You can also combine the /places with the above commands. This string 108051929285833/places/students/intersect brings everyone who has studied at a college or a university. The long number here (108051929285833) is the general page of College & University that is displayed on every other educational page (see right).
- Add their name into the current search engine
- Go to Pages
- Click on the text saying n like this
- A new window opens and you will find the ID in the URL
Since Facebook has acquired Instagram (and with that: removed Foursquare) I think, people are more open (or simply softly forced?) to check in on Facebook. This is certainly good for searching.
By using the /visitors command you can easily check who has been and where – and this info can reveal the professional interest and build the professional profile of a candidate.
This is how you can see which engineer has checked in at any NASA entities. Use the /str command as many times above… /str/nasa/pages-named/visitors/str/engineer/pages-named/employees/intersect
And if you remember one of the very first samples above (‘people who speak German and are any type of engineers and have ever visited any type of NASA entities while currently working for Boeing’) this is how that string looks like:
Another cool command: shows you who has liked a page. It can be very easily used: just add the /likers command at the end of the ID. These are the folks who have liked our Sourcing Centre in Budapest.
…or these are the guys who have liked every page named iPhone.
…or these might be some potential candidates from Google who have liked Yahoo.
The /residents command works as before. You can also combine with the /str command such as /str/your location/pages-named/residents/intersect.
Here you identify some Alexander Mann Solutions employees out of Krakow.
- /str/alexander mann/pages-named/employees/intersect
and some final ones
You can search for groups by using the /groups command. Use this at the end of a string.
As a sample check which groups Sourcing Recruiters belong to who work for Amazon. Wow, you will find Amazon’s closed social recruiting group there…
You can still add /males or /females at the end of your string. Its use is pretty obvious.
Note: this gender-search is massively against the data privacy and legislation regulations in many countries on the globe. #ThinkBeforeYouSearch
Hope you enjoyed (well, at least appreciated) this long tutorial – in my next post I will come up with some super challenging real samples (and of course, smart solutions).
Happy New Year, Sourcers! May the spirit of the Goat be with you in 2015!