我有一个反应票制标记我的文章有这样的话：LOL，跆拳道，妈呀，等等...

I have a reactions vote system to "tag" my articles with things like: LOL, WTF, OMG, and so on...

我想知道什么是检测什么反应是趋势该文章中的最佳解决方案。

I would like to know what would be the best solution to detect what "reactions" are "trending" in that article.

例如： http://www.buzzfeed.com/mjs538 /哈巴狗童车和 - 一个婴儿它，S标记WIN， 可爱，LOL

For example: http://www.buzzfeed.com/mjs538/pug-stroller-and-a-baby it,s tagged with WIN, CUTE, LOL

通过这个投票：

不过这一个： HTTP：// WWW。 buzzfeed.com/daves4/11-animals-having-their-hearts-tickled-out

只有标记为CUTE

由于可爱和其他选项之间的性差异是pretty的大，无论票本身的数目

Because the diference between CUTE and the others options is pretty big, regardless of the number of votes itself.

我已经有票了我的数据库中每一个反应是这样......

我想知道什么是最好的方式（PHP或MySQL查询？）来检测什么是最重要的反应...

我使用PHP和MySQL

谢谢！

推荐答案

你要做到这一点的方法是完全个人的 - 也就是说，它会反映你分析的响应方式。举例来说，我不认为第一个例子值得标签WIN（但这当然只是我的意见，）。

The way you are going to do this is utterly personal - that is, it will reflect your way of analyzing the responses. For instance, I don't think that the first example deserves the tag WIN (but that is only my opinion, of course).

我会尝试采用模糊逻辑。您可以根据得票率定义模糊集合，并根据这些定义的规则。该解决方案的最好的内在品质是，你可以微调规则库以便根据自己的愿望来进行分析。

I would try using fuzzy logic. You can define the fuzzy sets according to the percentage of votes and define your rules according to them. The best inherent quality of this solution is that you can fine tune your rule base in order to perform the analysis according to your desires.

如果您不熟悉模糊集合，三角集是简单的一个，简直是一条线从下限将最大值和另一个下行线从最大​​到设定的上限（检查为例这里）。

If you are not familiar with fuzzy sets, the triangular set is the simpler one and is simply a line going from the lower limit to the maximum and another descending line from the maximum to the upper limit of the set (check an example here).

模糊规则可以定义专案，根据你的思维。例如： 如果比例是非常高的，然后趋势为高 如果比例低则趋势是低的。

The fuzzy rules can be defined ad hoc, according to your thinking. For instance: IF percentage is VERY HIGH then trending is HIGH IF percentage is LOW then trending is LOW.

之后，您可以defuzzify你的趋势的结论，为了建立一个采伐限额来确定是否存在是一个趋势，该元素。

Afterwards, you can defuzzify your "trending" conclusion and establish a cutting limit in order to define whether or not there is a trend to that element.

这似乎有点势不可挡，但执行非常简单（在互联网上快速搜索让我找到本网站），您将收到的有力工具whathever使用后你会发现。

It might seem a bit overwhelming, but the implementation is straightforward (a quick search in the internet made me find this site) and you will have a powerful tool available for whathever posterior use you may find.