Weibo recommendation system

The emergence and http://www.aliyun.com/zixun/aggregation/17298.html of micro-blogging "> The form of communication information is relatively new, the increase in the amount of information, will face the problem of spam and duplication, the recommendation system will also be useful."

After the new registration, will recommend a popular user for initialization, the common practice in SNS is to import a mailbox or MSN contact, or like Kik message Mobile client scan the user's phone contact. Serious social applications need to use existing networks to complete product initialization, and the formation of products in the network spread, resulting in the exponential growth of product users, such as the early QQ, mailbox or mobile phone such as a manual add contact is no longer applicable, users do not have this patience. The more friends participate, the higher the user viscosity, the greater the cost of the product conversion.

And according to the completed place of origin, education and professional information recommended users, this is a cliché, the form of a community network is not necessarily suitable for micro-blogging, which relies on interest and topic to bring people together, not as a social network to look at other people's albums and avatars to decide whether to add as friends, at least Weibo's social composition will be lower. For ordinary users, perhaps it's just a chat-and-watch tool that is opportunistic about access to information, and whether more high-end users expect more valuable information.

From the personal experience of short-term use of Weibo, it is expected to be able to produce a high quality of information to the user, usually to view the user before the speech recognition, their own artificial filtering. Furthermore, the expert as a clue to see the expert fo users, and then choose whether the same fo. Weibo itself will also set up official spokesmen (such as Sina Technology), which will increase operating costs, the resulting information is too popular, the same can be obtained in other media or RSS information, the official Spokesman's value will be reduced.

Sina Weibo's products are more fragmented, more functional, there is little effect of actual use. Recommended popular users do not see any basis, the topic of personal concern for the recommended user can not produce filter. Manually added the label after the recommended user value is not high, rely on whether there is a common topic and whether in a school can not immediately evaluate whether FO, after viewing usually found that the user's number of fans is relatively small, or is not high quality of speech. Weibo itself wants these ordinary users to be FO, increasing their user's viscosity, and the frustration that comes with viewing is damaging to the user experience.

In the Sina Weibo developer conference, Wang Yuquan's repetitive messages and similar spam messages are annoying, and individuals do not care where these strangers go, and the messages that have been read for pure forwarding should be clearly filtered out. If you develop a more professional filter for Third-party applications, showing the most valuable 50 tweets a day, there may be some good prospects for development.

Source: http://daichuanqing.com/index.php/archives/2055