推理的几个概念

首先,这个是我们组师弟 Shawn Guo在一次组会上的报告,给我的启发。




image.png

推理   蕴含  蕴意


看到 inference 主要就是我现在SCT-mc做的这一块,但是Squad类似的东西还是Entailment。


Entailment 也叫蕴含,就是结果一定能被推出来,总统被刺杀了,他一定死了;孩子们在篮球场打球,孩子们一定在运动。


推理有三种,Deduction/Induction/Abduction  演绎、归纳、溯因  


对这三个比较好的解释,在这里 https://www.merriam-webster.com/words-at-play/deduction-vs-induction-vs-abduction


总结来说,就是


溯因:For example, if you find a half-eaten sandwich in your home, you might use probability to reason that your teenage son made the sandwich, realized he was late for work, and abandoned it before he could finish it.


看到了一件事情,这件事可能有很多种原因造成,你要对此进行猜测。


推理:Say you have a 10 o'clock appointment with the dentist and you know that it takes 30 minutes to drive from your house to the dentist's. From those two facts, you deduce that you will have to leave your house at 9:30, at the latest, to be at the dentist's on time.


就是根据已有的一些元素做出合理的推测,接下来发生的事情,元素必须都是正确的


Deductive reasoning always follows necessarily from general or universal premises. If a sandwich is defined as "two or more slices of bread or a split roll having a filling in between," and a hot dog is defined as "a frankfurter; especially : a frankfurter heated and served in a long split roll" then one must deduce that any hot dog served in a split roll is a sandwich.


induction是从一些很小的事件里面归纳出来一些statement,或者有概率的事实。比如你在天安门广场,碰到6个人,四个人都是小偷,那么天安门广场都是小偷,类似MLE


总结:

  • deduction: finding data to support an argument.

  • induction: finding an argument to explain some data;

  • abduction: supplying a warrant(保证) that enables us to move from data to argument.

Abductive reasoning (also called abduction,[1] abductive inference,[1] or retroduction[2]) is a form of logical inference which starts with an observation then seeks to find the simplest and most likely explanation. In abductive reasoning, unlike in deductive reasoning, the premises do not guarantee the conclusion. One can understand abductive reasoning as inference to the best explanation,[3] although not all uses of the terms abduction and inference to the best explanation are exactly equivalent



留下您的评论

回复列表:

    god发表于 Jan. 15, 2019, 6:10 p.m.

最近又要用到这个了,过来看看

By王炳宁 on Sept. 1, 2017 | 类别 NLP

关于本站