Data mining method and system for high-quality user-generated content

Abstract

This application discloses a data mining method and system for high-quality user-generated content (UGC), including: calculating the quality score of historical UGC published by each account and its correlation with various purposes, according to the quality score and correlation After receiving the newly published UGC of the high-quality account, calculate the quality score of the UGC and the correlation between the UGC and the category of the high-quality account that published the UGC ; Determine whether the quality score of the UGC is greater than the preset quality score threshold, and whether the correlation between the UGC and the category of the high-quality account that published the UGC is greater than the preset correlation threshold of the category, and if so, the The UGC is regarded as a high-quality UGC in the category of the high-quality account that published the UGC. Utilizing the present invention can improve the correlation between high-quality UGC and content concerned by target users and the category, and improve the real-time performance of high-quality UGC.

Description

technical field [0001] This application relates to the technical field of computer and Internet data processing, and in particular to a data mining method and system for high-quality User Generated Content (UGC, User Generated Content). Background technique [0002] At present, with the development of Internet technology, the Internet has gradually become an important source of information for people, especially after the Internet enters the Web2.0 era, users are not only the viewers of website content, but also the creators of website content. The content created by users is called UGC, such as logs and photos published by users. In the era of Web 2.0, due to the emergence of a large number of UGC, the amount of network information has grown rapidly at a geometric level, forming a situation of multiple, wide, and specialized, which has played a very important role in the accumulation and dissemination of human knowledge. [0003] A website system that can publish UGC is ge...

AI Technical Summary

Problems solved by technology

Therefore, the high-quality accounts set in the prior art are not based on the objective UGC content published by the accounts, but based on other factors such as the number of "fans" or subjective settings. Therefore, the so-called high-quality accounts sent The high-quality UGC content selected from UGC has poor correlation with the content that target users care about, and also has poor correlation with this category. For example, in the Weibo system, a high-quality "entertainment" account contains The content of the popular Weibo published does not necessarily belong to the "entertainment" category, and may belong to other categories

[0009] Secondly, the existing high-quality UGC data mining process is mainly based on the ranking and selection of UGC reposts and comments published by high-quality accounts of various categories, and is not selected based on the content of UGC, resulting in the final selected high-quality The content of UGC is less relevant to the content that target users care about, and also less relevant to this category

[0010] Thirdly, because the existing high-quality UGC data mining process is mainly based on the number of reposts and comments of UGC published by high-quality accounts in various categories, the UGC with a higher number of reposts and comments is often published. Older UGC, and the number of reposts and comments of the latest UGC is often very small, so the probability of the latest UGC being selected as high-quality UGC is very low, resulting in the poor real-time performance of high-quality UGC, which cannot adapt to some real-time needs Real-time requirements for higher categories such as news categories

[0011] In short, the high-quality UGC selected by the existing high-quality UGC data mining technology for various purposes has poor correlation with the content that target users care about, and the correlation with this category is also poor, and the real-time performance is also poor. poor

It makes it inconvenient for target users to quickly browse the UGC they care about, and it takes too much time and energy to find the content they care about

Images