Upload
berenice-kennedy
View
225
Download
0
Tags:
Embed Size (px)
Citation preview
UN
DL F
OU
ND
ATIO
N
UNLUNLA Language for A Language for
ComputersComputers
Hiroshi Uchida
UNL Foundation
Why a Language for
computers is necessary ?
A computer needs a capability to make knowledge processing.
A computer needs to shoulder thought and judgement of human using human knowledge.
A computer needs to make a processing based on contents.
UN
DL F
OU
ND
ATI O
N
Why a Language for
computers is necessary ?
A computer needs to have knowledge for knowledge processing.
A computer needs to have knowledge to shoulder human.
A computer needs to have knowledge to process contents.
A computer needs a language to store, use, manage, create knowledge like human.
UN
DL F
OU
ND
ATI O
N
Advantage of having a common language for
computers
A language for computers greatly reduces cost of developing knowledge or contents by sharing knowledge and contents of each system.
UN
DL F
OU
ND
ATI O
N
What should be a language for
computers ? Artificial language without any
ambiguity Consists of;
syntaxvocabulary (words)semantics
Express information by classifying objectivity and subjectivity.
UN
DL F
OU
ND
ATI O
N
How a Language for
Computers is developed ?
Define syntax
Develop vocabulary
Define semantics
UN
DL F
OU
ND
ATI O
N
Role of Syntax in a
Language for Computers
Syntax specifies roles of words
Syntax specifies subjectivity
UN
DL F
OU
ND
ATI O
N
How Syntax of a Language for Computers
is developed ?
A role of a word in a sentence decided by the syntax (marker) and the related word with this syntax (marker).
Syntax (marker) should be defined so as to specifies a role without any ambiguity.
UN
DL F
OU
ND
ATI O
N
How Syntax of a Language for Computers
is developed ?
When a word has relations between more than one other word, each relation should be set as to be able to identify each relation on the premise that we have enough knowledge about the concept of each word expressed.
UN
DL F
OU
ND
ATI O
N
How Syntax of a Language for Computers
is developed ?
When there are relations between words, each relation should be set so as to be able to understand the role of each UW only by referring to the relation label.
Collect every subjectivity used in NL.
UN
DL F
OU
ND
ATI O
N
Role of Words in a
Language for Computers
A word should express a concept.
That word should be non-ambiguous.
A word should have a well defined semantics.
(understandable for computers)
UN
DL F
OU
ND
ATI O
N
How Words of a Language for Computers is
developed ?
Words of a language can be developed by every people
Methodology to develop words should be clear enough.
UN
DL F
OU
ND
ATI O
N
How Words of a Language for Computers is
developed ?
Common symbol are necessary to express concepts for developers.
Combination of common symbols restrict meaning.
Using combination of common symbols, words of a language can be developed.
UN
DL F
OU
ND
ATI O
N
Role of Semantics in a
Language for Computers
Meanings of words should be defined so as to computers can understand them
UN
DL F
OU
ND
ATI O
N
How Semantics of a Language for Computers
is developed ?
Semantics of a word is defined by defining possible relations with other words.
Collect possible relations between words.
UN
DL F
OU
ND
ATI O
N
UNL: a Language for
Computers Artificial language (Semantic Network)
A node represents a concept An arc represents a relationA node can be annotated by attributes
Consists of;Universal Words: vocabularyRelations, Attributes: syntaxKnowledge Base: semantics
UN
DL F
OU
ND
ATI O
N
Long ago, in the city of Babylon, the people begun to build a huge tower, which seemed about to reach the heavens.
buildbegun tower
long ago people
seemedreach
heaven
huge
tim agt agt aoj
plc obj obj
gol
obj obj
Babylon
city
modobj
UN
DL F
OU
ND
ATI O
N
Long ago, in the city of Babylon, the people begun to build a huge tower, which seemed about to reach the heavens.
{unl}tim(begin(agt>thing,obj>thing).@entry.@past,long ago(icl>ago))mod(city(icl>region).@def,Babylon(icl>city))plc(begin(agt>thing,obj>thing).@entry.@past,city(icl>region).@def)agt(begin(agt>thing,obj>thing).@entry.@past,people(icl>person).
@def)obj(begin(agt>thing,obj>thing).@entry.@past,build(icl>do).@past)agt(build(agt>thing,obj>thing),people(icl>person).@def)obj(build(agt>thing,obj>thing),tower(icl>building))aoj(huge(icl>big),tower(icl>building))aoj(seem(icl>be).@past,tower(icl>building))obj(seem(icl>be).@past,reach(icl>come).@begin.@soon)obj(reach(icl>come).@begin.@soon,tower(icl>building))gol(reach(icl>come).@begin.@soon,heaven(icl>region).@def.@pl){/unl}
UN
DL F
OU
ND
ATI O
N
RelationRelation
Constitutes syntax of the UNL Expresses objectivity together with
UWs Expresses how concepts(UW)
constitutes a sentence related each other
38 relationsagt and aoj bas cag cao cnt cob con coo dur fmt frm ins man met mod nam obj or per plc plf plt pof pos ptn pur qua rsn scn seq src tim tmf tmt to via
UN
DL F
OU
ND
ATI O
N
AttributeAttribute
Constitutes syntax of the UNLExpresses subjectivity
evaluation of the speaker for the concepts(sentence)
Tense, aspect, mood, etc.
UN
DL F
OU
ND
ATI O
N
AttributeAttribute Time with respect to speaker
@past @present @future
Speaker’s view of aspect@begin-soon @begin-just @progress
@end-soon @end-just @complete @state @repeat
Speaker’s view of reference@generic @def @indef @not @ordinal
Speaker’s focus@emphasis @entry @qfocus @theme @title @topic
Speaker’s attitude@affirmative @confirmation @exclamation @imperative @interrogative @invitation @politeness @respect @vocativeU
ND
L F
OU
ND
ATI O
N
AttributeAttribute Speaker’s viewpoint
@ability, @ability-past, @admire, @although,
@ask-back, @conclusion, @custom, @ doubt,@expectation, @grant, @grant-not, @induce,
@inevitability, @insistence, @intention, @may,@obligation, @obligation-not, @possibility,@probability, @regret, @request, @should,@unexpected-presumption, @unexpected-consequence, @will
Convention@angle_bracket, @double_parenthesis,
@double_quotation, @parenthesis, @pl,@single_quotation, @square_bracket
UN
DL F
OU
ND
ATI O
N
UniversalUniversal WordWord
Vocabulary of the UNLAdopt English words as common
symbols for developers A UW represents a concept 1) Basic UW
ex) spring
2) Restricted UW
ex) spring(icl>tool)
spring(icl>season) spring(agt>person,obj>person)
UN
DL F
OU
ND
ATI O
N
How to define an UWHow to define an UW(1)(1)
One must decide to which category listed below each concept (meaning) belongs
Nominal Concept Verbal Concept Adjective Concept Adverbial Concept
UN
DL F
OU
ND
ATI O
N
How to define an UWHow to define an UW(2)(2)
Nominal Concept
swallow(icl>thing) Verbal Concept
change(agt>thing) I changed my mind.
change(obj>thing) The weather will change.
remember(agt>thing) I cannot remember his name.
remember(aoj>thing) Do you remember me?
UN
DL F
OU
ND
ATI O
N
How to define an UWHow to define an UW(3)(3)
Adjective Conceptpositive(aoj>thing) Are you positive abo
ut that?
a positive fact
only(mod<thing) the only person
Adverbial Conceptweekly(icl>how) This class is held
weekly.
only(icl>how) The child only cried.
UN
DL F
OU
ND
ATI O
N
If the ambiguity of an UW cannot be solved or still remains by attaching each of the above category labels, UW hierarchy or case relations will be used.
Attaching above categoryswallow(icl>bird) the bird
“One swallow does not make a summer”
swallow(icl>action) the action of swallowing“at one swallow”
swallow(icl>quantity) the quantity“take a swallow of water”
UN
DL F
OU
ND
ATI O
NHow to define an UWHow to define an UW
(4)(4)
Using (case) relations and what it takesspring(agt>thing,obj>wood) bending or dividing
something
spring(agt>thing,obj>mine)) blasting something
spring(agt>thing,obj>person, escaping (from) prison
src>prison))
spring(agt>thing,gol>place) jumping up
“to spring up”
spring(agt>thing,gol>thing) jumping on
“to spring on”
spring(obj>liquid) gushing out
“to spring out”
UN
DL F
OU
ND
ATI O
NHow to define an UWHow to define an UW
(5)(5)
Role of Master
Definition of a concept
A Master Definition defines a label for a concept (UW) and possible relations with other concepts (UW)
How to connect Uws to another UWs.
UN
DL F
OU
ND
ATI O
N
How to develop MD
First decide a label for a concept (UW)
Define possible relations with other concept (UW)
UN
DL F
OU
ND
ATI O
N
How to develop MD
ex)
Dixie(icl>jazz{>music})Dixie(icl>jazz)—icljazz(icl>music))
bark(agt>dog{>mammal,icl>sound(agt>thing)})
bark(agt>dog)—agtdog(icl>manmal)
bark(agt>dog)—iclsound(agt>thing)
UN
DL F
OU
ND
ATI O
N
UNL Knowledge BaseUNL Knowledge Base
Collection of concept from each language
Define every possible relations between concepts (UWs)
Ex)
dog(icl>mammal)—iclanimal(icl>living thing)bite(agt>animal,obj>thing)—agt
animal(libing thing)eat(agt>volitional thing)—obj
food(icl>functional thing)
UN
DL F
OU
ND
ATI O
N
UNL Knowledge BaseUNL Knowledge Base
UW system (hierarchy) is introduced for:1. to generate a word when a concept
is not included in a language
2. to reduce KB entries which can be deductively inferred
ex)dog(icl>mammal)—iclmammal (icl>animal)
KB give semantics of UWs
UN
DL F
OU
ND
ATI O
N
Communication with Communication with human languagehuman language
A language for computers needs to have a bridge to human languages to:
get knowledge or information
expressed in human languages express knowledge or content of
computers in human languages.
UN
DL F
OU
ND
ATI O
N
UNL-Language UNL-Language Master Master DictioDictio
narynary
Defines the link between a word of a language and a UW
Defines grammatical features when a word express a concept (UW)
Define KB when someone introduce a new UW
UN
DL F
OU
ND
ATI O
N
Development UNL
UNL should be developed by all the people in the world
• Universal words necessary for each language
• UNL center will develop at least 200,000 UWs in this tear.
UN
DL F
OU
ND
ATI O
N
What we expect to be developed by people
in the world
UNL should be developed by all the people in the world
• Universal words necessary for each language
UN
DL F
OU
ND
ATI O
N
What we expect to be developed by people in
the world Language Servers for new languages
and new domains
Application systems such as:
Information Retrieval System
UNL based Search Engines
Browsers
Editors/Word Processors
Machine translation Systems
UNL Web Pages
UN
DL F
OU
ND
ATI O
N
Top 10 Languages by Top 10 Languages by PopulationPopulation
RANK LANGUAGE POPULATION_______________________________________
1. CHINESE, MANDARIN885,000,000
2. SPANISH 332,000,000
3. ENGLISH 322,000,000
4. BENGALI 189,000,000
5. HINDI 182,000,000
6. ARABIC, ALL COUNTRIES 177,000,000
7. PORTUGUESE 170,000,000
8. RUSSIAN 170,000,000
9. JAPANESE 125,000,000
10. GERMAN, STANDARD 98,000,000
11. CHINESE, WU 77,175,000
___________________________________________Ethnologue:Languages of
the World
UN
DL F
OU
ND
ATI O
N