org.aksw.hawk.number

Class UnitEnglish

java.lang.Object

org.aksw.hawk.number.UnitEnglish

All Implemented Interfaces:

IUnitLanguage

public class UnitEnglish
extends Object
implements IUnitLanguage

This class converts natural word numbers to digit numbers and, if a unit is given, converts this unit to a base unit specified in a resource. The input can be any string and can have multiple occurrences of numbers and units.

 Examples:
 "Give me two hundred birds" -> "Give me 200 birds"
 "$80 thousand and three hundred four" -> "$ 80304"
 "Show the first ten screws with a length of two inches" -> "Show the first 10 screws with a length of 0.0508 m"
 "10 miles" -> "1609.344 m"
 
 "one" can be a numeral, but it does not have to be. To handle those cases,
 The Stanford pipeline will be started on sentences containing "one". It will only be converted to a number,
 if its not in a relation "nmod" with any other word. With this, we can handle sentences like:
 
 "The color is different from the old one."
 
  This sentence will be the same after parsing.

 

Field Summary

Fields

Modifier and Type	Field and Description
(package private) HashMap<String,Double>	identifierToMultiplier
(package private) HashMap<String,org.apache.commons.lang3.tuple.ImmutablePair<Double,String>>	identifierToUnit
private static org.slf4j.Logger	log
(package private) StanfordNLPConnector	stanford

Constructor Summary

Constructors

Constructor and Description
UnitEnglish(StanfordNLPConnector stanford) Stanford is needed to handle sentences containing "one"

Method Summary

Modifier and Type	Method and Description
String	convert(String q) Converts all occurring natural language numerals to digits.
private String	convertToBaseUnit(String str) Converts occurring units to their base unit.
private String	insertWhitespacebeforePunctuation(String input)
private void	loadResource()
static void	main(String[] args)
private Double	parseWord(String s)
private String	prettyAppendDouble(String out, Double val) Appends a String representation of a Double to input string.
private String	replaceNumerals(String replaceThis) Converts occurring natural language numerals to digits in any given String.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

log

private static org.slf4j.Logger log

identifierToUnit

HashMap<String,org.apache.commons.lang3.tuple.ImmutablePair<Double,String>> identifierToUnit

identifierToMultiplier

HashMap<String,Double> identifierToMultiplier

stanford

StanfordNLPConnector stanford

Constructor Detail

UnitEnglish

public UnitEnglish(StanfordNLPConnector stanford)

Stanford is needed to handle sentences containing "one"

Method Detail

convert

public String convert(String q)

Description copied from interface: IUnitLanguage

Converts all occurring natural language numerals to digits. If a unit is given, depending numeral will be converted to a base unit.

 "$80 million" - "$ 80000000"
 "10 miles" - "1609.344 m"
 

Specified by:

convert in interface IUnitLanguage

Parameters:

q - any string which may or may not contain numerals or units to convert.

Returns:

Same string with all numerals as digits and all units converted to a base unit.

replaceNumerals

private String replaceNumerals(String replaceThis)

Converts occurring natural language numerals to digits in any given String. For examples: UnitEnglish This only works when natural language numerals are stored in identifierToMultiplier. Call UnitEnglish#loadTabSplit(File) beforehand, to load data. To add numerals which should be recognized, edit resource file.

Parameters:

replaceThis - String which may or may not contain something to convert.

Returns:

InputString, but with replaced natural language numerals.

convertToBaseUnit

private String convertToBaseUnit(String str)

Converts occurring units to their base unit. Iterates over given string. if a unit is found, and the word left from unit is a number, conversion takes place.

Parameters:

str - Any string which may or may not contain a unit

Returns:

Input string but with unit converted to base unit.

insertWhitespacebeforePunctuation

private String insertWhitespacebeforePunctuation(String input)

prettyAppendDouble

private String prettyAppendDouble(String out,
                                  Double val)

Appends a String representation of a Double to input string. If its parseable to an integer, decimals will not be printed. All zeros will be printed. Scientific representation will not be used.

Parameters:

out - String to append to.

val - Val to append.

Returns:

String with appended Double val.

parseWord

private Double parseWord(String s)

Parameters:

s - String to be parsed to Double. Also accepts all numeral words defined in English file

Returns:

If String is parsable to Double, this will be returned. otherwise null.

loadResource

private void loadResource()
                   throws IOException

Throws:

IOException

main

public static void main(String[] args)