Wednesday, 25 September 2013

Facebook graph search - something beyond search.

'Search' on internet has been evolving from ages. It also changed and defined the way we search over internet and also made us provide information according the way we search. And now it took a big turn towards semantic search. Facebook graph search is one of its byproduct.

What is Facebook graph search?
Technically speaking it is semantic search engine where we get results for our natural language queries rather than links. To put in simple terms it gives results within the data  from facebook user and his/her networks and relationships. Additional results are provided by Bing search engine. .

It automatically gives few functions to choose from and extend our search.

How it is different from normal search engines?
It is different. It is different in terms of understanding the query. And so the accuracy and trust in the results vary. Because the results are collected from the information given either by the user himself or his networks and relationships. It depends on the interests expressed by the user and his network keeping their privacy settings into consideration. Whereas traditional search engines like google give the results by matching the search terms with the terms in webpages.

What difference it makes?
As said above, accuracy and trust are key points along with enormous opportunity it gives to its users to search deep with really big queries. Accuracy because the results are generated from the information users gave(believing that majority of active users on social networks are not faking) and trust because the information is given by his network.

Are graph search results prioritized?
Absolutely! They are customized as well. They are prioritized by showing the results from your network first and then results above your network as they are shared public. Remember facebook is binded to privacy settings. And customized as you can refine or narrow your search.

For instance we will do a simple sample search. "Photos taken in Varanasi, India"
there will be a list of photos of various sizes taken by different people will be shown there. Also giving options for two new searches on the top. Here in this instance it gave Recent and Before 2006. when we click on any of it, it gives new results accordingly.


There you can also see refine this search option where you can narrow the search

Few sample search terms you can try and see its extend-ability to refine your search:
Search1: Photos taken in Varanasi, India
Search2: Photos taken in Varansi, India after september 2013
Search3: Photos taken in Varanasi, India in september 2013
Search4:  Photos taken in Varanasi, India from september 2012 to january 2013
Search5:  Photos taken in Varanasi, India from september 2012 to january 2013 and photos by XXX
Search6: Search4:  Photos taken in Varanasi, India from september 2012 to january 2013 and by my friends who like M S Subhalakshmi
.... and the possibilities and list continues.

Search the web through facebook?
Any facebook user can search the web using facebook. By placing "web search:" before the search term.
These web search results are from microsoft bing search engine. One can also see web search results popping up during usual searches as well.

Just go to facebook graph search and discover different possibilities it gives, listen what Marc says.
If you don't have a graph search - you can hit try graph search button given in the facebook graph search page and see the magic!

Also to see the different possibilities of what graph search shows you by just do a simple search music my friends like and on the right column of the results page you can see refine search option, below that extend this search and below that discover something new. Just keep hitting discover something new and see the combinations of search keywords it takes and results it gives. you will love it!




Internet, web has been evolving from ages and it keeps evolving as long as man keeps on evolving and discovering the enormous power within himself.









No comments:

Post a Comment