Detection of Relations in Textual Documents

1
Detection of Relations in Textual Documents

• Manuela Kunze,
• Dietmar Rösner

University of Magdeburg C Knowledge Based Systems
and Document Processing
2
Introduction
http//en.wikipedia.org/wiki/Unsupervised_learning
3
Introduction

• to extract information from text, you can use
  techniques like simple pattern matching etc.

• additional knowledge is required
• 'Thursday' a day of a week
• meaning of
• (implicit) open' vs. close'
• Pay-what-you-wish'
• text understanding / techniques of NLP
• Exhibition of over 30 color photographs and
  stories of life in China's Yunnan Province '

4
Introduction

• ontologies contain information about
• definition/description of concepts and
• description of instances
• kind of relation (name, type),
• definition of domain and range values,
• characteristic of the relation cardinality,
  transitivity, ...,

5
Natural Language Processing

• NLP techniques
• case frame analysis
• exploiting syntactic structures
• corpus-based IE for an initial ontology
• corpus
• autopsy protocols (400 protocols)
• different document parts
• findings
• histological findings
• background
• discussion
• short linguistic structures
• typical attribute-value structures

6
Overview

• Case Frame
• Analysis of Specific Syntactic Structures
• Discussion/Conclusion

7
Case Frames

• resources
• results from syntactic parser
• ltNP TYPE"COMPLEX" RULE"NPC3" GEN"MAS" NUM"SG"
  CAS"NOM"gt       ltNP TYPE"FULL" RULE"NP1"
  CAS"NOM" NUM"SG" GEN"MAS"gt        
  ltNgtFlachschnittlt/Ngt       lt/NPgt       ltPP
  RULE"PP1" CAS"AKK"gt         ltPRP
  CAS"AKK"gtinlt/PRPgt         ltNP TYPE"FULL"
  RULE"NP2" CAS"AKK" NUM"SG" GEN"NTR"gt
            ltDETDgtdaslt/DETDgt          
  ltNgtZungengewebelt/Ngt         lt/NPgt       lt/PPgt
      lt/NPgt
• results from semantic tagger
• description of case frames

8
Case Frames

• (corpus-based) definition of roles for a concept
• Flachschnitt' (flat cut)
• location'
• sem. category tissue'
• PP, case of NP accusative, preposition in'
• Herausschleudern' (skidding)
• patient'
• sem. category body-hum'
• NP case of NP genitive
• location'
• sem. category vehicle'
• PP, case of NP dative, preposition aus'

9
Case Frames

• ltCONCEPT TYPE"medicalOperation"gt        
  ltWORDgtFlachschnittlt/WORDgt        
  ltDESCgtmedizinischer Schnittlt/DESCgt        
  ltSLOTSgt                 ltRELATION
  TYPE"LOCATION"gt                        
  ltASSIGN_TOgtTISSUElt/ASSIGN_TOgt                
          ltFORMgtP(akk, fak, in)lt/FORMgt        
                  ltCONTENTgtin das
  Zungengewebelt/CONTENTgt                
  lt/RELATIONgt         lt/SLOTSgt lt/CONCEPTgt
• ltCONCEPT TYPE"traffic-event"gt        
  ltWORDgtHerausschleudernlt/WORDgt        
  ltDESCgteventlt/DESCgt         ltSLOTSgt        
          ltRELATION TYPE"PATIENT"gt        
                  ltASSIGN_TOgtBODY-HUMlt/ASSIGN_TOgt
                          ltFORMgtN(gen, fak)lt/FORMgt
                          ltCONTENTgtdes
  Koerperslt/CONTENTgt                 lt/RELATIONgt
                  ltRELATION TYPE"LOCATION"gt
                          ltASSIGN_TOgtVEHICLElt/ASSIG
  N_TOgt                         ltFORMgtP(dat, fak,
  aus)lt/FORMgt                        
  ltCONTENTgtlt/CONTENTgt                 lt/RELATIONgt
          lt/SLOTSgt lt/CONCEPTgt

10
Case Frames

• coverage of phrases like fracture of elbow
  joint'?
• abstraction
• fracture' (sem. category trauma')
• role patient' sem. category bone'
• bruise' (sem. category trauma')
• role patient' sem. category organ'
• hematoma' (sem. category trauma')
• role patient' sem. category tissue'
• concept x (sem. category trauma')
• role patient' sem. category body-part'

11
Case Frames

• results
• relations are defined by the case frame
• name/type of relation
• domain, range
• corpus-based abstractions
• redefinition of semantic restriction
• use the least general hypernym as semantic
  restriction
• not yet extracted
• information about the characteristic of a relation

12
Overview

• Case Frame
• Analysis of Specific Syntactic Structures
• Discussion/Conclusion

13
Analysis of Specific Syntactic Structures

• from general to specific information
• resources
• results from syntactic parser
• results from semantic tagger
• description of interpretation of syntactic
  structures
• Which word class can be interpreted as
  concept/instance?
• Which word class describes a relation?
• adjective in a NP describes the noun in the NP ?
  relation prop
• negations negate concepts, verbs, or properties
  of a concept
• particle modification of adjectives

14
Analysis of Specific Syntactic Structures
CLMed? N ADJ
N interpreted as concept ADJ interpreted as
concept
prop(N, ADJ)
results prop_catadj(N,ADJ)
15
Analysis of Specific Syntactic Structures

• liver tissue bloodless

Steps

• nouns and adjectives are interpreted as
  concept/instance

liver_tissue
tissue
liver tissue
bloodless
blood concentration
bloodless
concept instance relation

• adjectives describe a relation
• in general 'prop'

16
Analysis of Specific Syntactic Structures

• liver tissue bloodless

ltowlClass rdfID"lebergewebe"gt
ltrdfssubClassOfgtltowlClass rdfID"tissue"/gtlt/rdf
ssubClassOfgtlt/owlClassgt ltowlClass
rdfID"blood-concentration"/gt ltowlClass
rdfID"blutleer"gt ltrdfssubClassOf
rdfresource"blood-concentration"/gtlt/owlClassgt
ltowlObjectProperty rdfID"prop_blood-concentra
tion"gt ltrdfsdomain rdfresource"tissue"/gtltr
dfsrange rdfresource"blood-concentration"/gtlt/o
wlObjectPropertygt ltlebergewebe
rdfID"Lebergewebe_6"gt ltprop_blood-concentrat
iongtltblutleer rdfID"blutleer_7"/gtlt/prop_blood-co
ncentrationgtlt/lebergewebegt
17
Analysis of Specific Syntactic Structures
"kaum wahrnehmbare Unterblutungen" (Engl. "hardly
detectable hematomas")
results of syntactic parser ltNP TYPE"FULL"
RULE"NP4" CAS"_" NUM"PL" GEN"FEM"gt
ltADJP RULE"ADJP1"gt ltADVgtkaumlt/ADVgt
ltADJgtwahrnehmbarelt/ADJgt lt/ADJPgt
ltNgtUnterblutungenlt/Ngt lt/NPgt

• resources for interpretation
• N concept/instance
• ADJ
• concept/instance
• rel prop
• ADV
• concept/instance
• rel mod

• results of semantic tagger
• kaum' weak-graduation
• wahrnehmbar' unknown token
• Unterblutung' trauma

18
Analysis of Specific Syntactic Structures

• hardly detectable hematomas

Steps

• nouns, adjectives and adverbs are interpreted as
  concept/instance

• adjectives and adverbs describe relations

concept instance relation
hematoma
trauma
hematoma
detectable
unspecified
hardly
hardly
weak-graduation
19
Analysis of Specific Syntactic Structures

• hardly detectable hematomas

ltowlClass rdfID"unterblutung"gtltrdfssubClassOf
rdfresource"trauma"/gtlt/owlClassgt ltowlClass
rdfID"trauma"/gt ltowlClass rdfID"wahrnehmbar"
gt ltrdfssubClassOf rdfresource"unspecified"/gtlt
/owlClassgt ltowlClass rdfID"unspecified"/gt lto
wlClass rdfID"kaum"gt ltrdfssubClassOf
rdfresource"weak-graduation"/gtlt/owlClassgt ltow
lClass rdfID"weak-graduation"/gt
20
Analysis of Specific Syntactic Structures

• hardly detectable hematomas

ltowlObjectProperty rdfID"mod_weak-graduation"gt
ltrdfsdomain rdfresource"unspecified"/gt ltrdfs
range rdfresource"weak-graduation"/gtlt/owlObje
ctPropertygt ltowlObjectProperty
rdfID"prop_unspecified"gt ltrdfsdomain
rdfresource"trauma"/gt ltrdfsrange
rdfresource"unspecified"/gtlt/owlObjectPropertygt
ltunterblutung rdfID"Unterblutungen_5"gt ltprop
_unspecified rdfresource"wahrnehmbare_4"/gtlt/unt
erblutunggt ltwahrnehmbar rdfID"wahrnehmbare_4"gt
ltmod_weak-graduation rdfresource"kaum_3"/gtlt/w
ahrnehmbargt ltkaum rdfID"kaum_3"gtlt/kaumgt
21
Analysis of Specific Syntactic Structures

• Phrases like
• NP ? NP NP
• NP ? N Adj Conj Adj
• NP ? N conj N Adj

concept instance relation
Protégé Plugin for Visualization Ontoviz
22
Analysis of Specific Syntactic Structures

• results
• definition of concepts/instances
• corpus-based definition/concretion of relations
• prop ? prop_catADJ
• information about domain, relation
• not extracted
• information about the characteristic of a
  relation

23
Overview

• Case Frame
• Analysis of Specific Syntactic Structures
• Discussion/Conclusion

24
Conclusion

• NLP techniques for extraction of information
• analyse syntactic structures
• information about semantic categories
• result corpus-based description of an initial
  ontology
• case frame analysis
• relations are described in the case frame
• disadvantage creation of case frames
• advantage a definition of the relation
• analysis specific syntactic structures
• a general interpretation of tokens and the
  syntactic structures
• redefined by results from the semantic tagger
• disadvantage in some case, only the general
  relation definition is delivered
• advantage less effort to describe the resources

25
Conclusion

• no information about the characteristic of a
  relation (cardinality, )
• solutions
• analyse occurrences in the corpus
• corpus-based assumption about cardinality
• integration of additional knowledge
• initial domain specific ontology

26
Key Aspects for IE

• conceptual preprocessing steps Names of
  concepts occur in different linguistic
  structures compound vs. complex noun phrase
  (like liver tissue and tissue of liver)
• handle only one canonical linguistic structure as
  a representative for all paraphrases
• treatment of generalisation within local contexts
  
• The token liver may occur in the first sentence
  of a paragraph. In the next sentences of the
  paragraph, only the hypernym organ is used.
• concept or instance which term in a linguistic
  structure has to be interpreted as a concept and
  which as an instance of a concept resp.
• definition of the scope for a concept
• a paragraph starts with a description of an organ
  (e.g. organ liver in The liver shows ... .
  Bloodrichness of the tissue. ), after this
  follows a description of parts of the organ
  (e.g., Gewebe). In such cases, additional
  knowledge about the domain has to be employed
  (for example, about meronyms or holonyms)
• tissue part-of liver vs tissue part-of concept X