Flipboard for it by analyzing the user in the social network, including friends like Facebook, all social networking sites and Twitter, Sina microblogging microblogging service above information as a source of information, direct crawl behind some of the links renumbered the articles or pictures of the entire organization, and presented to the new layout. SNS is based on friends, family and may have a common interest or common concern of friends, so the correlation of these content will be higher, closer personal interest.
Based personalized recommendation algorithm recommended (Interest Graphs)
Zite for forward / ignored / shield and other behavioral data to calculate the user's preferences by analyzing a collection of different sources / topics / primary screening data on the user's social concerns / news subscription Again screening recommended for users with more relevant information.
The disadvantage of mainstream read mode
SNS and SNS-based source of information not suitable for reading
The fly based on the recommendation of the social network, in addition to the number of friends is not caused by too much information narrower social network information flow as read information source essentially is debatable. Corrosion because many people already SNS platform of this group, like Jean-Marie Le Pen's "rabble" said, "groups of poor reasoning, is anxious to action" personal microblogging integration into the group after the emotional and thoughts will turn to groups of all public direction, easier to impulsive, fickle, credulous, irritable, paranoid, overbearing, emotional, extreme, does not allow the existence of doubt and uncertainty, like low biological state. This has nothing to do with the individual qualities of the constituent groups, this time its decisive role is instinctive and emotional, is an "unconscious" level, rather than rational.
Therefore, in many cases, the information you see is around the particular microblogging staggering SNS big vat, after a lot of information into the implied and amplified the various comments, forwarding, modify, re- forwarding will become mutilated, distorted, rumors become even because of the missing and out of context. Excessive amplification of any power will blur the truth behind, so we need to be more calm, rational and objective information, rather than this source of information can not be guaranteed to be around.
Recommendation algorithm is a panacea?
Or would you say, we have algorithms think that may come in the future based on semantic search, data mining, and intelligent match the Web3.0 completely personalized era, many people are optimistic about it. It is true that the present and future trends are based algorithm recommended, but in the moment, the algorithm is not a panacea. Act as recommended roles such as watercress, most the domestic quality algorithms with known or your friendly neighbors, but not often do not fly watercress guess. The moment, the algorithm also can not completely take on the important task of recommended, after all, the algorithm is dead, often a little accidents will result in severe matching offset. For example, reading applications need to provide individuals with inside "source" or after polymerization to form the customization of information "flow" from the definition of "source", algorithm-based selection process, single and repeat multiple applications "source" so customizable difficulty increases, the run time becomes longer, Once you interested in a little bit of change, occasional interruptions will make the filtering mechanism flaws, because the algorithm is not smart enough to be Identify some exceptions.