talking to your data: natural language interfaces for a schema-less world (keynote at nliwod, iswc...
DESCRIPTION
The increase in the size, heterogeneity and complexity of contemporary Big Data environments brings major challenges for the consumption of structured and semi–structured data. Addressing these challenges requires a convergence of approaches from different communities including databases, natural language processing, and information retrieval. Research on Natural Language Interfaces (NLI) and Question Answering systems has played a prominent role in stimulating a multidisciplinary approach to the problem that has moved the field from a futuristic vision to a concrete industry-level technological trend. In this talk we distill the key principles of state-of-the-art approaches for data consumption using NLI. Particular attention is paid to the maturity and effectiveness of each approach together with discussion on future trends and active research questions.TRANSCRIPT
Talking to your Data:
Natural Language Interfaces for a
schema-less world
André Freitas
NLIWoD at ISWC 2014
Riva del Garda
Outline
Shift in the Database Landscape
On Schema-agnosticism & Semantics
Distributional Semantics to the Help
Case Study: Treo QA System
Living in a Schema-less World
Take-away Message
Shift in the Database
Landscape
3
Big Data (Data Variety)
Vision: More complete data-based picture of the world for
systems and users.
4
The Long Tail of Data Variety
The Long Tail of Data Variety
6
Data variety +
Data
Programs
Full data coverage
Full automation
Full knowledge
The Long Tail of Data Variety
7
Data variety +
Data
Programs
Full data coverage
Full automation
Full knowledge
The Long Tail of Data Variety
Data generation
8
Very-large and dynamic “schemas”
10s-100s attributes1,000s-1,000,000s attributes
circa 2000circa 2014
9
Semantic Heterogeneity
Decentralized content generation.
Multiple perspectives (conceptualizations) of the reality.
Ambiguity, vagueness, inconsistency.
10
Data variety +
Data
Programs
Full data coverage
Full automation
Full knowledge
The Long Tail of Data Variety
Data generation
Data consumption
11
Databases for a Complex World
How do you query data at this scale?
12
Schema-agnosticism
Ab
str
ac
tio
n
La
ye
r
User
13
First-level independency
(Relational Model)
“… it provides a basis for a high level data language which will yield maximal independence between programs on the one hand and representation and organization of data on the other”
Codd, 1970
Second-level independency
(Schema-agnosticism)
14
On Schema-agnosticism
& semantics
15
Vocabulary Problem for Databases
Query: Who is the daughter of Bill Clinton married to?
Semantic Gap
Possible representations
Schema-agnostic query
mechanisms
Abstraction level differences
Lexical variation
Structural (compositional) differences
Operational/functional differences
16
Robust Semantic Model
Semantic intelligent behaviour is highly dependent on knowledge scale (commonsense, semantic)
Semantics
=
Formal meaning representation model
(lots of data)
+
inference model
17
Robust Semantic Model
Not scalable!
1st Hard problem: Acquisition
Semantics
=
Formal meaning representation model
(lots of data)
+
inference model
18
Robust Semantic Model
Not scalable!
2nd Hard problem: Consistency
Semantics
=
Formal meaning representation model
(lots of data)
+
inference model
19
“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.”
Formal World Real World
Baroni et al. 2013
Semantics for a Complex World
20
Distributional Semantic Models
Semantic Model with low acquisition effort(automatically built from text)
Simplification of the representation
Enables the construction of comprehensive commonsense/semantic KBs
What is the cost?
Some level of noise(semantic best-effort)
21
Distributional Hypothesis
“Words occurring in similar (linguistic) contexts tend to be semantically similar”
He filled the wampimuk with the substance, passed itaround and we all drunk some
22
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
23
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
24
Distributional Semantic Models (DSMs)
car
dog
bark
run
leash
25
Context
Semantic Similarity & Relatedness
car
dog
bark
run
leash
26
Query: cat
Semantic Similarity & Relatedness
θ
car
dog
cat
bark
run
leash
27
Query: cat
DSMs as Commonsense Reasoning
Commonsense is here
θ
car
dog
cat
bark
run
leash
28
Semantic Approximation is here
DSMs as Commonsense Reasoning
θ
car
dog
cat
bark
run
leash
...
vs.
Semantic best-effort
Case Study: Treo QA
System
30
Approach Overview
Query Planner
Ƭ-Space
Large-scale
unstructured data
Commonsense
knowledge
Structured
Data
Distributional
semantics
Core semantic approximation &
composition operations
Query AnalysisQuery Query Features
Query Plan
31
Approach Overview
Query Planner
Ƭ-Space
Wikipedia
RDF Data
Explicit Semantic
Analysis (ESA)
Core semantic approximation &
composition operations
Query AnalysisQuery Query Features
Query Plan
Commonsense
knowledge
32
Ƭ-Space
e
p
r
33
Core Operations
Search &
Composition
Operations
Query
34
Does it work?
35
Addressing the Vocabulary Problem for
Databases (with Distributional Semantics)
Gaelic: direction
36
Solution (Video)
37
More Complex Queries (Video)
38
Treo Answers Jeopardy Queries (Video)
http://bit.ly/1hWcch939
Relevance
Test Collection: QALD 2011.
DBpedia.
Dataset (DBpedia + YAGO links): 45,767 predicates, 9,434,677
instances, more than 200,000 classes
40
Transform natural language queries into triplepatterns.
“Who is the daughter of Bill Clinton married to?”
Query Pre-Processing
(Question Analysis)
41
Step 1: POS Tagging- Who/WP
- is/VBZ
- the/DT
- daughter/NN
- of/IN
- Bill/NNP
- Clinton/NNP
- married/VBN
- to/TO
- ?/.
Query Pre-Processing
(Question Analysis)
42
Step 2: Core Entity Recognition- Rules-based: POS Tag + TF/IDF
Who is the daughter of Bill Clinton married to?(PROBABLY AN INSTANCE)
Query Pre-Processing
(Question Analysis)
43
Step 3: Determine answer typeRules-based.
Who is the daughter of Bill Clinton married to?(PERSON)
Query Pre-Processing
(Question Analysis)
44
Step 4: Dependency parsing- dep(married-8, Who-1)
- auxpass(married-8, is-2)
- det(daughter-4, the-3)
- nsubjpass(married-8, daughter-4)
- prep(daughter-4, of-5)
- nn(Clinton-7, Bill-6)
- pobj(of-5, Clinton-7)
- root(ROOT-0, married-8)
- xcomp(married-8, to-9)
Query Pre-Processing
(Question Analysis)
45
Step 5: Determine Partial Ordered Dependency Structure
(PODS)
- Rules based.
• Remove stop words.
• Merge words into entities.
• Reorder structure from core entity position.
Query Pre-Processing
(Question Analysis)
46
Bill Clinton daughter married to
(INSTANCE)
ANSWER
TYPE
Person
QUESTION FOCUSLower level of ambiguity,
vagueness, synonimy
Question Analysis
Transform natural language queries into triplepatterns
“Who is the daughter of Bill Clinton married to?”
Bill Clinton daughter married to
(INSTANCE) (PREDICATE) (PREDICATE) Query Features
PODS
47
Query Plan
Map query features into a query plan.
A query plan contains a sequence of core operations.
(INSTANCE) (PREDICATE) (PREDICATE) Query Features
Query Plan
(1) INSTANCE SEARCH (Bill Clinton)
(2) p1 <- SEARCH PREDICATE (Bill Clintion, daughter)
(3) e1 <- NAVIGATE (Bill Clintion, p1)
(4) p2 <- SEARCH PREDICATE (e1, married to)
(5) e2 <- NAVIGATE (e1, p2)
48
Instance Search
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data:
Instance Search
49
Predicate Search
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data::Chelsea_Clinton
:child
:Baptists:religion
:Yale_Law_School
:almaMater
...(PIVOT ENTITY)
(ASSOCIATED
TRIPLES)
50
Predicate Search
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data::Chelsea_Clinton
:child
:Baptists:religion
:Yale_Law_School
:almaMater
...
sem_rel(daughter,child)=0.054
sem_rel(daughter,child)=0.004
sem_rel(daughter,alma mater)=0.001
Which properties are semantically related to ‘daughter’?
51
Predicate Search
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data::Chelsea_Clinton
:child
:Baptists:religion
:Yale_Law_School
:almaMater
...
sem_rel(daughter,child)=0.054
sem_rel(daughter,child)=0.004
sem_rel(daughter,alma mater)=0.001
Which properties are semantically related to ‘daughter’?
(In the context of Bill Clinton)
52
Navigate
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data::Chelsea_Clinton
:child
53
Navigate
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data::Chelsea_Clinton
:child
(PIVOT ENTITY)
54
Predicate Search
Bill Clinton daughter married to
:Bill_Clinton
Query:
Linked
Data::Chelsea_Clinton
:child
(PIVOT ENTITY)
:Mark_Mezvinsky
:spouse
55
Results
56
Core Principles
Minimize the impact of Ambiguity, Vagueness, Synonymy with
semantic pivoting.
Semantic pivoting: Address the simplest matchings first
(heuristics).
Semantic Relatedness as a primitive semantic approximation
operation.
Distributional semantics as commonsense/semantic
knowledge.
Natural Language Queries over Heterogeneous Linked Data Graphs: A Distributional-
Compositional Semantics Approach, IUI 2014
Living in a
Schema-less World
58
How do we build systems today?
Structure the domain
59
Generalize and encode some rules
How do we build systems today?
Allow some constrained interaction
How do we build systems today?
Query is here
61
Siloed Systems
62
Data variety +
Data
Full data coverage
Full automation
Full knowledge
63
Linked Data: Datasets are easier to integrate and to
consume (data model level). However, the semantic
barrier for consumption is still there
Data variety +
Data
Full data coverage
Full automation
Full knowledge
65
Distributional DBMS
Natural Language Queries over Heterogeneous Linked Data Graphs: A Distributional-
Compositional Semantics Approach, IUI 2014
Data variety +
Data
Full data coverage
Full automation
Full knowledge
67
Simplification of Information Extraction
A Semantic Best-Effort Approach for Extracting Structured Discourse Graphs, WoLE, 2012
Simplification of Information Extraction
General Electric Company, or GE , is an American multinational conglomerate
corporation incorporated in Schenectady , New York
69
Data variety +
Data
Full data coverage
Full knowledge
Full automation
70
Schema-agnostic programs
Towards An Approximative Ontology-Agnostic Approach for Logic Programs, FOIKS 2014
Data variety +
Data
Full data coverage
Full knowledge
Full automation
72
Reasoning with Distributional Semantics
A Distributional Semantics Approach for Selective Reasoning on Commonsense Graph
Knowledge Bases, NLDB 2014
Data variety +
Data
Full data coverage
Full automation
Full knowledge
74
Existing semantic technologies can address today major data
management problems
Muiti-disciplinarity is one key (and NLI people are very good at it!):- NLP + IR + Semantic Web + Databases
Schema-agnosticism is a central property/functionality/goal!
Distributional Semantics + semantics of structured data =
schema-agnosticism
Schema-agnosticism brings major impact for information systems.
We can tame the long tail of data variety!
The wave is just starting. Be a part of it!
Take-away Message
75
Want to play with Distributional
Semantics?
http://easy-esa.org
76
Any Queries?