Do you know how social networking sites build personalized networks?

I think we all know that the social graph is a record of our social media connections. From search results to coupons, services based on social maps are now commonplace. But at the same time, a more accurate, personalized network is starting to rise--a network of interest profiles for users. There have been some scattered interest maps, such as Google's personalized search, or Amazon's recommendation engine. Some companies want this service to cover all of our online lives, but that means that where we go and what we see, it will be just our hearts. That means we no longer need to search the news sites for favorite stories, or to find what we need on the shopping site-because once you land on these sites, the system will automatically identify who we are, what we like, and then present the information we need.

No matter what this idea will bring to your surprise or panic, I think we should all look at how this completely personalized network works.

Find out what you like

In fact, social networking sites such as Twitter and Facebook may not be as effective as clicking or commenting on other sites when it comes to determining people's real interests. Many people share information on Twitter because of career needs, and the information that users often share on Facebook is limited to the social circle. In either case, "most of the time, people are just making a show to impress people." This leads to incomplete user data. ”

So how does the map of interest work? Start-up company Gravity is a company based on this service, its CTO Jim Benedetto briefed us on how the site works. Gravity at the same time for a number of Web services, it will track the user in all of their services site behavior, so that users log on any site, can automatically get the characteristics of content. You can click here to see the site to build a map of the specific process.

Technically speaking, gravity is a large data engine based on multiple datasets, such as Freebase and DBpedia, to determine the actual point of interest when a user clicks on an article or publishes a comment. For example, if a user publishes a tweet from Vanessa Laine (Lakers NBA player Kobe Bryant's ex-wife), he should be more interested in basketball than Laine's date of birth or other exact but irrelevant information.

Carry "personal label" with you

One day, people can be based on their interests regardless of what page they are browsing.

This is indeed a compelling vision: our interest map will affect what we see on any Web site. If you like skiing, you may be able to see the relevant trading information on the ski equipment on the ecommerce website. Also, chances are that you will not have to log in to see this information, because these sites will take a holistic view of your behavior across the network.

Although there are already some scattered interest maps, the real difficulty is how to integrate these fragmented, fragmented interests into a single personal label about who we are and what we like. It is possible that such methods as OAuth (allowing APIs to be shared among different services) will work. This way you can log on online only once, your interest map is in the drive, and the entire network based on your needs to serve you.

Really for you.

Of course, while many people will enjoy this personalized Internet experience, there will be concerns about privacy issues. Also, some people would like to be able to surf the internet, rather than the site to predict what they like. For this reason, this personalized network must be an optional service, not a required one.

Privacy is a thorny issue. Therefore, users should have the right to decide which online records they have become part of their personal label. For example, most users who prefer pornographic video do not want their preferences to affect their recommended news content, or when he or she is watching a game, the site suddenly recommends pornographic video. Therefore, the website must set the appropriate authorization button, as well as the "Do Not track" button to ensure that this personalized network will not become a stumbling block for users.

Although we may like this personalized network very much, but everything should have a degree. The so-called personalized network services, in addition to ensure that the network information presented personalized, but also should ensure that the service itself user-friendly, really for users.