Skip to content

Commit

Permalink
release candidate for noun.reference_to and referenced_by api
Browse files Browse the repository at this point in the history
  • Loading branch information
@spencermountain
spencermountain committed May 18, 2015
1 parent bc492d8 commit 97fef78
Show file tree
Hide file tree
Showing 6 changed files with 295 additions and 42 deletions.
197 changes: 163 additions & 34 deletions client_side/nlp.js
View file
Original file line number Diff line number Diff line change
Expand Up @@ -5427,7 +5427,7 @@ var date_extractor = (function() {
// console.log(date_extractor("1999"))

//wrapper for value's methods
var Value = function(str, next, last, token) {
var Value = function(str, sentence, word_i) {
var the = this
the.word = str || '';

Expand Down Expand Up @@ -5932,11 +5932,14 @@ var inflect = (function() {
// console.log(inflect.singularize('mayors of chicago')=="mayor of chicago")

//wrapper for noun's methods
var Noun = function(str, next, last, token) {
var Noun = function(str, sentence, word_i) {
var the = this
var token, next;
if(sentence!==undefined && word_i!==undefined){
token=sentence.tokens[word_i]
next=sentence.tokens[word_i+i]
}
the.word = str || '';
the.next = next
the.last = last

if (typeof module !== "undefined" && module.exports) {
parts_of_speech = require("../../data/parts_of_speech")
Expand Down Expand Up @@ -6089,7 +6092,11 @@ var Noun = function(str, next, last, token) {
"ss",
"of",
"the",
"for"
"for",
"and",
"&",
"co",
"sons"
]
l= blacklist.length
for (i = 0; i < l; i++) {
Expand All @@ -6104,22 +6111,119 @@ var Noun = function(str, next, last, token) {
return true
}
}
//see if noun has a first-name
var names = Object.keys(firstnames)
l = names.length
var firstname=the.word.split(' ')[0].toLowerCase()
for (i = 0; i < l; i++) {
if (names[i]===firstname) {
return true
}
//see if noun has a known first-name
var names=the.word.split(' ').map(function(a){
return a.toLowerCase()
})
if(firstnames[names[0]]){
return true
}
//(test middle name too, if there's one)
if(names.length> 2 && firstnames[names[1]]){
return true
}

//if it has an initial between two words
if(the.word.match(/[a-z]{3,20} [a-z]\.? [a-z]{3,20}/i)){
return true
}
return false
}

//decides if it deserves a he, she, they, or it
the.pronoun=function(){

//if it's a person try to classify male/female
if(the.is_person()){
var names=the.word.split(' ').map(function(a){
return a.toLowerCase()
})
if(firstnames[names[0]]==="m" || firstnames[names[1]]=="m"){
return "he"
}
if(firstnames[names[0]]==="f" || firstnames[names[1]]=="f" ){
return "she"
}
//test some honourifics
if(the.word.match(/^(mrs|miss|ms|misses|mme|mlle)\.? /,'i')){
return "she"
}
if(the.word.match(/\b(mr|mister|sr|jr)\b/,'i')){
return "he"
}
//if it's a known unisex name, don't try guess it. be safe.
if(firstnames[names[0]]==="a" || firstnames[names[1]]=="a" ){
return "they"
}
//if we think it's a person, but still don't know the gender, do a little guessing
if(names[0].match(/[aeiy]$/)){//if it ends in a 'ee or ah', female
return "she"
}
if(names[0].match(/[ou]$/)){//if it ends in a 'oh or uh', male
return "he"
}
if(names[0].match(/(nn|ll|tt)/)){//if it has double-consonants, female
return "she"
}
//fallback to 'singular-they'
return "they"
}

//not a person
if(the.is_plural()){
return "they"
}

return "it"
}

//tokens that refer to the same thing. "[obama] is cool, [he] is nice."
the.referenced_by = function() {
//if it's named-noun, look forward for the pronouns pointing to it -> '... he'
if(token && token.pos.tag!=="PRP"){
var prp=the.pronoun()
//look at rest of sentence
var interested=sentence.tokens.slice(word_i+1, sentence.tokens.length)
//add next sentence too, could go further..
if(sentence.next){
interested=interested.concat(sentence.next.tokens)
}
//find the matching pronouns, and break if another noun overwrites it
var matches=[]
for(var i=0; i<interested.length; i++){
if(interested[i].pos.tag==="PRP" && interested[i].normalised===prp){
matches.push(interested[i])
}else if(interested[i].pos.parent==="noun" && interested[i].analysis.pronoun()===prp){
break
}
}
return matches
}
return []
}

// a pronoun that points at a noun mentioned previously '[he] is nice'
the.reference_to = function() {
//if it's a pronoun, look backwards for the first mention '[obama]... <-.. [he]'
if(token && token.pos.tag==="PRP"){
var prp=token.normalised
//look at starting of this sentence
var interested=sentence.tokens.slice(0, word_i)
//add previous sentence, if applicable
if(sentence.last){
interested=sentence.last.tokens.concat(interested)
}
//reverse the terms to loop through backward..
interested=interested.reverse()
for(var i=0; i<interested.length; i++){
//it's a match
if(interested[i].pos.parent==="noun" && interested[i].pos.tag!=="PRP" && interested[i].analysis.pronoun()===prp){
return interested[i]
}
}
}
}

//specifically which pos it is
the.which = (function() {
//posessive
Expand All @@ -6143,9 +6247,12 @@ if (typeof module !== "undefined" && module.exports) {
// console.log(new Noun('farmhouse').is_entity())
// console.log(new Noun("FBI").is_acronym())
// console.log(new Noun("Tony Danza").is_person())
// console.time('h')
// console.log(new Noun("Tonys h. Danza").is_person())
// console.timeEnd('h')
// console.log(new Noun("Tony Danza").pronoun()=="he")
// console.log(new Noun("Tanya Danza").pronoun()=="she")
// console.log(new Noun("mrs. Taya Danza").pronoun()=="she")
// console.log(new Noun("Gool Tanya Danza").pronoun()=="she")
// console.log(new Noun("illi G. Danza").pronoun()=="she")
// console.log(new Noun("horses").pronoun()=="they")

//turns 'quickly' into 'quick'
var to_adjective = (function() {
Expand Down Expand Up @@ -6212,11 +6319,9 @@ var to_adjective = (function() {
// console.log(to_adjective('marvelously') === 'marvelous')

//wrapper for Adverb's methods
var Adverb = function(str, next, last, token) {
var Adverb = function(str, sentence, word_i) {
var the = this
the.word = str || '';
the.next = next
the.last = last

if (typeof module !== "undefined" && module.exports) {
to_adjective = require("./conjugate/to_adjective")
Expand Down Expand Up @@ -7077,11 +7182,14 @@ var verb_conjugate = (function() {
// console.log(verb_conjugate("have tried"))

//wrapper for verb's methods
var Verb = function(str, next, last, token) {
var Verb = function(str, sentence, word_i) {
var the = this
var token, next;
if(sentence!==undefined && word_i!==undefined){
token=sentence.tokens[word_i]
next=sentence.tokens[word_i+i]
}
the.word = str || '';
the.next = next
the.last = last

if (typeof module !== "undefined" && module.exports) {
verb_conjugate = require("./conjugate/conjugate")
Expand Down Expand Up @@ -7183,7 +7291,7 @@ var Verb = function(str, next, last, token) {
if (the.word.match(/n't$/)) {
return true
}
if ((modals[the.word] || copulas[the.word]) && the.next && the.next.normalised === "not") {
if ((modals[the.word] || copulas[the.word]) && next && next.normalised === "not") {
return true
}
return false
Expand Down Expand Up @@ -7602,11 +7710,9 @@ var adj_to_adv = (function() {
// console.log(adj_to_adv('direct'))

//wrapper for Adjective's methods
var Adjective = function(str, next, last, token) {
var Adjective = function(str, sentence, word_i) {
var the = this
the.word = str || '';
the.next = next
the.last = last

if (typeof module !== "undefined" && module.exports) {
to_comparative = require("./conjugate/to_comparative")
Expand Down Expand Up @@ -8591,6 +8697,23 @@ var Sentence = function(tokens) {
})
}

//find the 'it', 'he', 'she', and 'they' of this sentence
//these are the words that get 'exported' to be used in other sentences
the.referables=function(){
var pronouns={
he:undefined,
she:undefined,
they:undefined,
it:undefined
}
the.tokens.forEach(function(t){
if(t.pos.parent=="noun" && t.pos.tag!="PRP"){
pronouns[t.analysis.pronoun()]=t
}
})
return pronouns
}

return the
}

Expand Down Expand Up @@ -9154,21 +9277,25 @@ var pos = (function() {
})
}

//make them Sentence objects
sentences = sentences.map(function(s) {
var sentence = new Sentence(s.tokens)
sentence.type = s.type
return sentence
})
//add analysis on each token
sentences = sentences.map(function(s) {
s.tokens = s.tokens.map(function(token, i) {
var last_token = s.tokens[i - 1] || null
var next_token = s.tokens[i + 1] || null
token.analysis = parents[token.pos.parent](token.normalised, next_token, last_token, token)
token.analysis = parents[token.pos.parent](token.normalised, s, i)
return token
})
return s
})

//make them Sentence objects
sentences = sentences.map(function(s) {
var sentence = new Sentence(s.tokens)
sentence.type = s.type
//add next-last references
sentences = sentences.map(function(sentence,i) {
sentence.last=sentences[i-1]
sentence.next=sentences[i+1]
return sentence
})
//return a Section object, with its methods
Expand All @@ -9191,7 +9318,7 @@ var pos = (function() {
// console.log(pos("may live").tags())
// console.log(pos("may 7th live").tags())
// console.log(pos("She and Marc Emery married on July 23, 2006.").tags())
// console.log(pos("spencer quickly acked").sentences[0].tokens)
// console.log(pos("Toronto is fun. Spencer and heather quickly walked. it was cool").sentences[0].referables())
// console.log(pos("a hundred").sentences[0].tokens)
// console.log(pos("Tony Reagan skates").sentences[0].tokens)
// console.log(pos("She and Marc Emery married on July 23, 2006").sentences[0].tokens)
Expand All @@ -9205,6 +9332,8 @@ var pos = (function() {
// pos("Sather tried to stop the deal, but when he found out that Gretzky").sentences[0].tokens.map(function(t){console.log(t.pos.tag + " "+t.text+" "+t.pos_reason)})
// pos("Gretzky had tried skating").sentences[0].tokens.map(function(t){console.log(t.pos.tag + " "+t.text+" "+t.pos_reason)})

// console.log(pos("i think Tony Danza is cool. He rocks and he is golden.").sentences[0].tokens[2].analysis.referenced_by())
// console.log(pos("i think Tony Danza is cool and he is golden.").sentences[0].tokens[6].analysis.reference_to())

//just a wrapper for text -> entities
//most of this logic is in ./parents/noun
Expand Down
6 changes: 3 additions & 3 deletions client_side/nlp.min.js
View file

Large diffs are not rendered by default.

47 changes: 47 additions & 0 deletions src/parents/noun/index.js
View file
Original file line number Diff line number Diff line change
Expand Up @@ -244,6 +244,53 @@ var Noun = function(str, sentence, word_i) {
return "it"
}

//tokens that refer to the same thing. "[obama] is cool, [he] is nice."
the.referenced_by = function() {
//if it's named-noun, look forward for the pronouns pointing to it -> '... he'
if(token && token.pos.tag!=="PRP"){
var prp=the.pronoun()
//look at rest of sentence
var interested=sentence.tokens.slice(word_i+1, sentence.tokens.length)
//add next sentence too, could go further..
if(sentence.next){
interested=interested.concat(sentence.next.tokens)
}
//find the matching pronouns, and break if another noun overwrites it
var matches=[]
for(var i=0; i<interested.length; i++){
if(interested[i].pos.tag==="PRP" && interested[i].normalised===prp){
matches.push(interested[i])
}else if(interested[i].pos.parent==="noun" && interested[i].analysis.pronoun()===prp){
break
}
}
return matches
}
return []
}

// a pronoun that points at a noun mentioned previously '[he] is nice'
the.reference_to = function() {
//if it's a pronoun, look backwards for the first mention '[obama]... <-.. [he]'
if(token && token.pos.tag==="PRP"){
var prp=token.normalised
//look at starting of this sentence
var interested=sentence.tokens.slice(0, word_i)
//add previous sentence, if applicable
if(sentence.last){
interested=sentence.last.tokens.concat(interested)
}
//reverse the terms to loop through backward..
interested=interested.reverse()
for(var i=0; i<interested.length; i++){
//it's a match
if(interested[i].pos.parent==="noun" && interested[i].pos.tag!=="PRP" && interested[i].analysis.pronoun()===prp){
return interested[i]
}
}
}
}

//specifically which pos it is
the.which = (function() {
//posessive
Expand Down
Loading

0 comments on commit 97fef78

Please sign in to comment.