这条路线的搜索空间 (universe)是句子长度 n 的这样的一个函数:可以 assume n 中每两个词都必须发生7种二元关系之一。三种是实关系但是有方向(父父子子),所以“原子化”后就是6种实关系,即,是二元排列不是组合。第7种是:无关系。无关系也算关系,就一网打尽了。任意两词只允许发生7种关系之一,不能多也不能少。在 n 不大的时候,搜索空间爆炸得不算厉害。
白:ordered pairs,A跟B和B跟A可以有不同的关系标签。
李:对,有这个二元循环的可能,忘了这茬了。不过那很罕见,对于搜索空间影响不大。能想到的只有 定语从句谓词与中心词有二元循环关系,一个 mod 一个 arg 方向相反。
what 似乎也不齐全,只是展示结构的 what,没有展示结构的功能性(角色)。所以,作为学习,这里有两个空白需填补,一个是 how,尤其是语义相谐机制,怎么招之即来挥之即去的。另一个是逻辑语义,逻辑语义怎么在句法或逻辑的链接基础上得出的。当然这二者是相关的,前者是条件,后者是结论。目前展示的结构树图就是个架子和桥梁。
“boys go to Jupiter to get more stupider, girls go to college to get more knowledge.”
这是取笑男孩的。饶舌的甜甜现场发挥,富于夸张和强调:“what do you want me to say now? boys go to Jupiter , do you know the planet Jupiter? they go to the planet Jupiter, once they get there, they get supider and supider every second. And girls they go to college to get more knowledge and knowledge into their brain on their head.”
"Eeny, meeny, miny, moe, Catch a tiger by the toe. If he hollers, let it go, Eeny, meeny, miny, moe.
My mother told me/says to pick the very best one, and you are not it."
这是非常流行的“选择”童谣。小孩子面对两个或多种选择的时候,不知道选哪一样好,就口中念念有词,一边用手在选择物之间轮流数着,道理上应该是童谣完了手落在哪个选择上,就选择哪个。可是,儿童的心理是微妙的,很多时候内心其实有了一个所指,为了最终得到自己想得到的,表面上还跟着童谣走,孩子们学会在童谣后面,打着家长的名号,用肯定或否定来保证自己不要落到自己不要选的东西上:如果最后落到中意的选项上,就说 “My mother told me/says to pick the very best one, and that is YOU”. 否则就改口说:“My mother told me/says to pick the very best one, and you are not it.”
"You know what Kick your butt All the way to Pizza Hut
While you're there, Comb your hair Don't forget your underwear!"
里面有个片段说学校的事儿。回家说的这个故事是小女孩玩家家的,也有微妙的儿童心理:
"I said that I am the Princess of Jewelry because one of my friends and buddy said that she looked at my jewelry I brought to school. What happened is she was so surprised and she loved it ... she said that I am Princess of Jewelry and she is the Queen of Makeup. Next time I am going to bring new jewelry, she said that I am the Queen of Jewelry...... No,Daddy, Jessica said I am the Queen of Jewelry if I bring some new jewelry tomorrow."
看目前 Siri 的水平,相当不错了,蛮impressed,毕竟是 Siri 第一次把自然语言对话推送到千千万万客户的手中,虽然有很多噱头,很多人拿它当玩具,毕竟有终端客户的大面积使用和反馈的积累。尽管如此,后出来的 Google Assistant 却感觉只在其上不在其下,由于搜索统治天下20年的雄厚积累,开放类知识问答更是强项。
所有话术都那么具有可爱的欺骗性,until 最后一句,莫名其妙回应说 this isn't supported.
(顺便一提,上面终于发现一个语音转写错误,我跟 Google Assistant 说的是,you are both funny and sometimes amusing. 她听成了 and sometimes I'm using. 从纯粹语音相似角度,也算是个 reasonable mistake,从句法角度,就完全不对劲了,both A and B 要求 A 和 B 是同类的词啊。大家知道,语音转写目前是没有什么语言学句法知识的,为了这点改错,加上语言学也不见得合算。关键是,其实也没人知道如何在语音深度神经里面融入语言学知识。这个让深度学习与知识系统耦合的话题且放下,以后有机会再论。)
2 短语:VP = Verb Phrase; AP = Adjective Phrase; NP = Noun Phrase; VG = Verb Group; NG = Noun Group; NE = Named Entity; DE = Data Entity; Pred = Predicate; CL = Clause;
3 句法:H = Head; O = Object; S = Subject;M = Modifier; R = Adverbial; (veryR = Intensifier-adverbial;possM = possessive-modifier); NX = Next; CN = Conjoin; sCL = Subject Clause;oCL = Object Clause; mCL = Modifier/Relative Clause; Z = Functional; X = Optional Function
2 短语:VP = Verb Phrase; AP = Adjective Phrase; NP = Noun Phrase; VG = Verb Group; NG = Noun Group; NE = Named Entity; DE = Data Entity; Pred = Predicate; CL = Clause;
3 句法:H = Head; O = Object; S = Subject;M = Modifier; R = Adverbial; (veryR = Intensifier-Adverbial); NX = Next; CN = Conjoin; sCL = Subject Clause;oCL = Object Clause; mCL = Modifier/Relative Clause; Z = Functional; X = Optional Function