Semantically Conceptualizing and Annotating Tables

1
Semantically Conceptualizing and Annotating Tables

• Stephen Lynn David W. Embley
• Data Extraction Research Group
• Department of Computer Science
• Brigham Young University

Supported by the
2
Overview

• Context
• WoK Web of Knowledge
• TANGO Table ANalysis for Generating Ontologies
• MOGO Mini-Ontology GeneratOr
• Semantic Enrichment via MOGO
• Implementation
• Experimentation
• Enhancements
• Challenges Opportunities

3
WoK a Web of Knowledge
4
TANGO
fleck velter velter
fleck gonsity (ld/gg) hepth(gd)
burlam 1.2 120
falder 2.3 230
multon 2.5 400
TANGO repeatedly turns raw tables into conceptual
mini-ontologies and integrates them into a
growing ontology.
Growing Ontology
5
MOGO
fleck velter velter
fleck gonsity (ld/gg) hepth(gd)
burlam 1.2 120
falder 2.3 230
multon 2.5 400
TANGO repeatedly turns raw tables into conceptual
mini-ontologies and integrates them into a
growing ontology.
Growing Ontology
MOGO generates mini-ontologies from interpreted
tables.
6
MOGO Overview

• Table
• Interpretation
• Yields a canonical table
• Canonical Table
• Concept/Value Recognition
• Relationship Discovery
• Constraint Discovery
• Yields a semantically enriched conceptual model
• Mini-ontology
• Integration into a growing ontology

MOGO
7
Sample Input
Region and State Information Region and State Information Region and State Information Region and State Information
Location Population (2000) Latitude Longitude
Northeast 2,122,869
Delaware 817,376 45 -90
Maine 1,305,493 44 -93
Northwest 9,690,665
Oregon 3,559,547 45 -120
Washington 6,131,118 43 -120
Sample Output
8
Concept/Value Recognition

• Lexical Clues
• Labels as data values
• Data value assignment
• Data Frame Clues
• Labels as data values
• Data value assignment
• Default
• Recognize concepts and values by syntax and
  layout

9
Concept/Value Recognition

• Lexical Clues
• Labels as data values
• Data value assignment
• Data Frame Clues
• Labels as data values
• Data value assignment
• Default
• Recognize concepts and values by syntax and
  layout

Region
State
10
Concept/Value Recognition

• Lexical Clues
• Labels as data values
• Data value assignment
• Data Frame Clues
• Labels as data values
• Data value assignment
• Default
• Recognize concepts and values by syntax and
  layout

Region
State
11
Relationship Discovery
2000

• Dimension Tree Mappings
• Lexical Clues
• Generalization/Specialization
• Aggregation
• Data Frames
• Ontology Fragment Merge

12
Relationship Discovery

• Dimension Tree Mappings
• Lexical Clues
• Generalization/Specialization
• Aggregation
• Data Frames
• Ontology Fragment Merge

13
Constraint Discovery

• Generalization/Specialization
• Computed Values
• Functional Relationships
• Optional Participation

Region and State Information Region and State Information Region and State Information Region and State Information
Location Population (2000) Latitude Longitude
Northeast 2,122,869
Delaware 817,376 45 -90
Maine 1,305,493 44 -93
Northwest 9,690,665
Oregon 3,559,547 45 -120
Washington 6,131,118 43 -120
14
Validation

• Concept/Value Recognition
• Correctly identified concepts
• Missed concepts
• False positives
• Data values assignment
• Relationship Discovery
• Valid relationship sets
• Invalid relationship sets
• Missed relationship sets
• Constraint Discovery
• Valid constraints
• Invalid constraints
• Missed constraints

Precision Recall F-measure
Concept Recognition 87 94 90
Relationship Discovery 73 81 77
Constraint Discovery 89 91 90
15
Concept Recognition

• Counted
• Correct/Incorrect/Missing Concepts
• Correct/Incorrect/Missing Labels
• Data value assignments

16
Relationship Discovery

• Counted
• Correct/incorrect/missing relationship sets
• Correct/incorrect/missing aggregations and
  generalization/specializations

17
Constraint Discovery

• Counted
• Correct/Incorrect/Missing
• Generalization/Specialization constraints
• Computed value constraints
• Functional constraints
• Optional constraints

18
Concept Recognition

• Successes
• 98 of concepts identified
• Missing label identification
• 97 of values assigned to correct concept
• Common problems
• Finding an appropriate label
• Duplicate concepts

19
Relationship Discovery

• Recall of 92 for relationship sets
• Missing aggregations and gen./spec.s (only found
  in label nesting)
• Unnecessary rel. sets generated (are computable)

20
Constraint Discovery

• F-measure of 98 for functional relationship sets
• Computed value discovery
• Funtional/non-functional ? lists in cells

21
MOGO Contributions

• Tool to generate mini-ontologies
• Accuracy encouraging

Precision Recall F-measure
Concept Recognition 87 94 90
Relationship Discovery 73 81 77
Constraint Discovery 89 91 90
22
Opportunities Challenges MOGO

• Enhancements
• Check for inter-label relationships
• Check for more complex computations
• Check for lists in cells
• Wish List
• Data-frame library
• Atomic knowledge components
• Instance recognizers
• Library of molecular components
• Semi-automatic construction of a WordNet-like
  resource for knowledge components

23
Summary

• MOGO
• Semantic Enrichment
• Encouraging Results
• But More Possible
• Broader Implications Vision Challenges
• TANGO
• WoK
• Web of Data
• Semantic Annotation
• User-friendly Query Answering

www.deg.byu.edu embley_at_cs.byu.edu
24
Opportunities Challenges TANGO

• Table Interpretation
• Transforming tables to F-logic Pivk07
• Layout-independent table representation Jha08
• Table interpretation by sibling tables Tao07
• Semantic Enhancement / Ontology Generation
• Naming unnamed table concepts Pivk07
• MOGO Lynn09
• Semi-automatic Ontology Integration
• Ontology Matching Euzenat07
• Ontology-mapping tools Falconer07
• Direct and indirect schema mappings for TANGO
  Xu06

25
Opportunities Challenges WoK

• Web of Data
• The Semantic Web is a web of data. W3C
• Upcoming special issue of Journal of Web
  Semantics
• Enabling a Web of Knowledge Tao09
• Information Extraction
• Domain-independent IE from web tables
  Gatterbauer07
• Open IE Banko07

26
Opportunities Challenges WoK

• Semantic Annotation wrt Ontologies
• Linking Data to Ontologies Poggi08
• TISP Tao07
• FOCIH Tao09
• Reasoning Query Answering
• Description Logics Baadar03
• NLIDB Community
• AskOntos Ding06
• SerFR Al-Muhammed07

27
References

• Al-Muhammed07 Al-Muhammed and Embley,
  Ontology-Based Constraint Recognition for
  Free-Form Service Requests, Proceedings of the
  23rd International Conference on Data
  Engineering, 2007.
• Baader, Calvanese, McGuinness, Nardi and
  Patel-Schneider, The Description Logic Handbook,
  Cambridge University Press, 2003.
• Banko07 Banko, Cafarella, Soderland, Broadhead
  and Etzioni, Open Information Extraction from
  the Web, Proceedings of the International Joint
  Conference on Artificial Intelligence, 2007.
• Ding06 Ding, Embley and Liddle, Automatic
  Creation and Simplified Querying of Semantic Web
  Content An Approach Based on Information-Extracti
  on Ontologies, Proceedings of the First Asian
  Semantic Web Conference, 2006.
• Euzenat07 Eusenat and Shvaiko, Ontology
  Matching, Springer Verlag, 2007.
• Falconer07 Falconer, Noy and Storey, Ontology
  MappingA User Survey, Proceedings of the Second
  International Workshop on Ontology Mapping, 2007.
• Gatterbauer07 Gatterbauer, Bohunsky, Herzog and
  Pollak, Towards Domain-Independent Information
  Extraction from Web Tables, Proceedings of the
  Sixteenth International World Wide Web
  Conference, 2007.
• Jha07 Jha and Nagy, Wang Notation Tool Layout
  Independent Representation of Tables,
  Proceedings of the 19th International Conference
  on Pattern Recognition, 2007.
• Pivk07 Pivk, Sure, Cimiano, Gams, Rajkovic and
  Studer, Transforming Arbitrary Tables into
  Logical Form with TARTAR, Data Knowledge
  Engineering, 2007.
• Poggi08 Poggi, Lembo, Calvanese, DeGiacomo,
  Lenzerini and Rosati, Linking Data to
  Ontologies, Journal on Data Semantics, 2008.
• Tao07 Tao and Embley, Automatic Hidden-Web
  Table Interpretation by Sibling page Comparison,
  Proceedings of the 26th International Conference
  on Conceptual Modeling, 2007.
• Tao09 Tao, Embley and Liddle, Enabling a Web
  of Knowledge, Technical Report
  tango.byu.edu/papers, 2009.
• Xu06 Xu and Embley, A Composite Approach to
  Automating Direct and Indirect Schema Mappings,
  Information Systems, 2006.