Towards a Distributional Semantic Web 
Stack 
André Freitas, Edward Curry, Siegfried Handschuh 
Insight Centre for Data Analytics 
University of Passau 
URSW 2014 
Riva del Garda
 Position paper 
 Model targeting semantic approximations 
(from a praxis perspective) 
 Interested in collecting references / creating 
bridges with this community 
2
Outline 
 Motivation 
 Distributional Semantic Models (DSMs) 
 Distributional-Relational Models (DRMs) 
 Applications 
 Take-away message 
3
Motivation 
 Semantic intelligent behaviour is highly dependent on 
(commonsense, semantic) knowledge scale 
Semantics 
= 
Formal meaning representation model 
(lots of data) 
+ 
inference model 
4
Motivation 
 Scalability problems 
1st Hard problem: Acquisition 
Semantics 
= 
Formal meaning representation model 
(lots of data) 
+ 
inference model 
5
Motivation 
 Scalability problems 
2nd Hard problem: Consistency 
Semantics 
= 
Formal meaning representation model 
(lots of data) 
+ 
inference model 
6
Semantics for a Complex World 
 “Most semantic models have dealt with particular types of 
constructions, and have been carried out under very simplifying 
assumptions, in true lab conditions. 
 If these idealizations are removed it is not clear at all that modern 
semantics can give a full account of all but the simplest 
models/statements.” 
Baroni et al. 2013 
7
Distributional Semantic Models 
 Semantic Model with low acquisition effort 
(automatically built from text) 
Simplification of the representation 
(vector-based) 
 Enables the construction of comprehensive 
commonsense/semantic KBs 
 Trades formal structure for volume of commonsense 
knowledge 
 What is the cost? 
Some level of noise 
(semantic best-effort) 
8
Distributional Hypothesis 
“Words occurring in similar (linguistic) contexts tend 
to be semantically similar” 
 He filled the wampimuk with the substance, passed it 
around and we all drunk some 
9
Distributional Semantic Models (DSMs) 
“The dog barked in the park. The owner of the dog put him on the 
leash since he barked.” 
contexts = nouns and verbs in the same 
sentence 
10
Distributional Semantic Models (DSMs) 
“The dog barked in the park. The owner of the dog put him on the 
leash since he barked.” 
bark 
dog 
park 
leash 
contexts = nouns and verbs in the same 
sentence 
bark : 2 
park : 1 
leash : 1 
owner : 1 
11
Distributional Semantic Models (DSMs) 
car 
dog 
bark 
run 
leash 
12
Semantic Similarity & Relatedness 
dog 
car 
bark 
run 
leash 
13 
Query: cat
Semantic Similarity & Relatedness 
θ 
dog 
cat 
car 
bark 
run 
leash 
14 
Query: cat
Distributional Semantic Models (DSMs)
DSMs as Commonsense Reasoning 
Commonsense data is here 
θ 
car 
dog 
cat 
bark 
run 
leash 
Semantic Approximation is here 
16
Distributional-Relational Models (DRMs) 
 Hybrid distributional + structured data 
 Semantic approximation as a first-class citizen 
 Structured data + user query provides a contextual 
support for the semantic approximation 
17
Distributional-Relational Models (DRMs) 
Your Algorithm 
goes here 
DRM 
Structured Data 
Text Collection 
Distributional 
Semantic 
Model 
Heuristics to minimize 
the approximation 
errors 
18
Distributional-Relational Models (DRMs) 
Your Algorithm 
goes here 
DRM 
Structured Data 
Structured Data (the 
same or another one) 
Distributional 
Semantic 
Model 
Heuristics to minimize 
the approximation 
errors 
19
DRM 
20
Application: Flexible Querying / Semantic 
Search 
Freitas et al., ICSC 2011 Freitas & Curry, IUI 2014
Application: Selective Reasoning (1) 
 Speer et al. AAAI 2009 
 Freitas et al, NLDB 2014 
22
Application: Distributional Semantics and 
Logic Programming 
 Pereira da Silva & Freitas, FOIKS 2014 
23
Application: Knowledge Discovery 
 Entity similarity/Entity consolidation 
 Relationship discovery 
 Novacek et al. ISWC 2011 
 Cohen et al. T. AMIA Annu Symp 2009 
 Speer et al. AAAI 2009 
24
Distributional Semantics / 
Semantic Web Stack? 
25
26
Take-away message 
Effective semantic approximation that works 
+ 
Automatic construction of comprehensive semantic 
models from unstructured data 
+ 
Simple to use 
 Powerful semantic pattern in practice. 
27
Do-it-yourself 
http://easy-esa.org 
28

Towards a Distributional Semantic Web Stack

  • 1.
    Towards a DistributionalSemantic Web Stack André Freitas, Edward Curry, Siegfried Handschuh Insight Centre for Data Analytics University of Passau URSW 2014 Riva del Garda
  • 2.
     Position paper  Model targeting semantic approximations (from a praxis perspective)  Interested in collecting references / creating bridges with this community 2
  • 3.
    Outline  Motivation  Distributional Semantic Models (DSMs)  Distributional-Relational Models (DRMs)  Applications  Take-away message 3
  • 4.
    Motivation  Semanticintelligent behaviour is highly dependent on (commonsense, semantic) knowledge scale Semantics = Formal meaning representation model (lots of data) + inference model 4
  • 5.
    Motivation  Scalabilityproblems 1st Hard problem: Acquisition Semantics = Formal meaning representation model (lots of data) + inference model 5
  • 6.
    Motivation  Scalabilityproblems 2nd Hard problem: Consistency Semantics = Formal meaning representation model (lots of data) + inference model 6
  • 7.
    Semantics for aComplex World  “Most semantic models have dealt with particular types of constructions, and have been carried out under very simplifying assumptions, in true lab conditions.  If these idealizations are removed it is not clear at all that modern semantics can give a full account of all but the simplest models/statements.” Baroni et al. 2013 7
  • 8.
    Distributional Semantic Models  Semantic Model with low acquisition effort (automatically built from text) Simplification of the representation (vector-based)  Enables the construction of comprehensive commonsense/semantic KBs  Trades formal structure for volume of commonsense knowledge  What is the cost? Some level of noise (semantic best-effort) 8
  • 9.
    Distributional Hypothesis “Wordsoccurring in similar (linguistic) contexts tend to be semantically similar”  He filled the wampimuk with the substance, passed it around and we all drunk some 9
  • 10.
    Distributional Semantic Models(DSMs) “The dog barked in the park. The owner of the dog put him on the leash since he barked.” contexts = nouns and verbs in the same sentence 10
  • 11.
    Distributional Semantic Models(DSMs) “The dog barked in the park. The owner of the dog put him on the leash since he barked.” bark dog park leash contexts = nouns and verbs in the same sentence bark : 2 park : 1 leash : 1 owner : 1 11
  • 12.
    Distributional Semantic Models(DSMs) car dog bark run leash 12
  • 13.
    Semantic Similarity &Relatedness dog car bark run leash 13 Query: cat
  • 14.
    Semantic Similarity &Relatedness θ dog cat car bark run leash 14 Query: cat
  • 15.
  • 16.
    DSMs as CommonsenseReasoning Commonsense data is here θ car dog cat bark run leash Semantic Approximation is here 16
  • 17.
    Distributional-Relational Models (DRMs)  Hybrid distributional + structured data  Semantic approximation as a first-class citizen  Structured data + user query provides a contextual support for the semantic approximation 17
  • 18.
    Distributional-Relational Models (DRMs) Your Algorithm goes here DRM Structured Data Text Collection Distributional Semantic Model Heuristics to minimize the approximation errors 18
  • 19.
    Distributional-Relational Models (DRMs) Your Algorithm goes here DRM Structured Data Structured Data (the same or another one) Distributional Semantic Model Heuristics to minimize the approximation errors 19
  • 20.
  • 21.
    Application: Flexible Querying/ Semantic Search Freitas et al., ICSC 2011 Freitas & Curry, IUI 2014
  • 22.
    Application: Selective Reasoning(1)  Speer et al. AAAI 2009  Freitas et al, NLDB 2014 22
  • 23.
    Application: Distributional Semanticsand Logic Programming  Pereira da Silva & Freitas, FOIKS 2014 23
  • 24.
    Application: Knowledge Discovery  Entity similarity/Entity consolidation  Relationship discovery  Novacek et al. ISWC 2011  Cohen et al. T. AMIA Annu Symp 2009  Speer et al. AAAI 2009 24
  • 25.
    Distributional Semantics / Semantic Web Stack? 25
  • 26.
  • 27.
    Take-away message Effectivesemantic approximation that works + Automatic construction of comprehensive semantic models from unstructured data + Simple to use  Powerful semantic pattern in practice. 27
  • 28.