TANGO

1
TANGO

• Table Analysis for Generating OntologiesDavid W.
  Embley (BYU) George Nagy (RPI)under NSF
  Awards 0414644 and 0414854 INFORMATION
  KNOWLEDGE MANAGEMENTDr. Maria Zemankova (a)
  Table Interpretation
• (b) Query by Table

2
TANGO STEPS
TABLE
Wang Notation Tool
INTERPRETED TABLE
Wang Notation XML
MINI ONTOLOGY
Ontology Editor
GROWING ONTOLOGY
Annotated Semantic Web Pages
Standard Ontology Language (OWL)
Ontology Based Web Services
Form Based Specification
Extraction Ontologies
Relational Databases
Query By Table
3
This presentation
TABLE
Wang Notation Tool
INTERPRETED TABLE
Wang Notation XML
MINI ONTOLOGY
Ontology Editor
GROWING ONTOLOGY
Annotated Semantic Web Pages
Standard Ontology Language (OWL)
Ontology Based Web Services
Form Based Specification
Extraction Ontologies
Relational Databases
Query By Table
4
(a) Table Interpretation
Confirm or correct
HTML web pages
Extract table
Matlab table
XMLtable
Construct Wang notation
Wang Notation
Confirm orcorrect
Mini Ontology
5
Median Income tablehttp//www40.statcan.ca/l01/cs
t01/famil108a.htm?sdimedian20income
6
Median Income table displayed from Canada
Statistics displayed in TANGO Wang Notation Tool
7
Wang Notation

• Abstract table is specified by ordered pair (C,?)
  - (category, delta)
• C is a finite set of labeled domains (header, sub
  headers of tables, etc)
• ? represents each individual value within a table
  corresponding to C.

8
Categories

• Two categories in previous table.
• CATEGORY 1 (Region_Virtual,(Canada,phi),
  (Newfoundland and Labrador,phi), (Prince Edward
  Island,phi), (Nova Scotia,phi), (New
  Brunswick,phi), (Quebec,phi), (Ontario,phi),
  (Manitoba,phi), (Saskatchewan,phi),(Alberta,phi),(
  British Columbia,phi),(Yukon Territory,phi),
  (Northwest Territories,phi), (Nunavut,phi))
• CATEGORY 2 (Year_Virtual, (2001,phi),
  (2002,phi), (2003,phi), (2004,phi), (2005,phi))
  

9
Content (leaf) cells

• Delta Notation for two (of 15) rows
• delta(Year_Virtual.2001,Region_Virtual.Canada)5
  3,500
• delta(Year_Virtual.2002,Region_Virtual.Canada)5
  5,000
• delta(Year_Virtual.2003,Region_Virtual.Canada)5
  6,000
• delta(Year_Virtual.2004,Region_Virtual.Canada)5
  8,100
• delta(Year_Virtual.2005,Region_Virtual.Canada)6
  0,600
• delta(Year_Virtual.2001,Region_Virtual.Newfoundla
  nd and Labrador)41,400
• delta(Year_Virtual.2002,Region_Virtual.Newfoundla
  nd and Labrador)43,200
• delta(Year_Virtual.2003,Region_Virtual.Newfoundla
  nd and Labrador)44,800
• delta(Year_Virtual.2004,Region_Virtual.Newfoundla
  nd and Labrador)46,100
• delta(Year_Virtual.2005,Region_Virtual.Newfoundla
  nd and Labrador)47,600

10
XML RepresentationSchema for (1) table (2)
categories (3) data cells (4) augmentation

• ltInterpretedTable xsinoNamespaceSchemaLocation"G
  \RPI\XML\02_TableInterface.XS.070803.xml"
  xmlnsxsi"http//www.w3.org/2001/XMLSchema-instan
  ce"gt
• ltTable TableOID"Table2" Number"2"
  DocumentCitation"Wang's Thesis" Title"Wang
  table" Caption"Grades in 1991 and 1992"gt
• ltCategoryNodesgt
• ltCategoryNode CategoryNodeOID"C1"
  Label"Median Total Income"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C11"
  Label"Canada"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C12"
  Label"Newfoundland and Labrador"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C13"
  Label"Prince Edward Island"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C14"
  Label"Nova Scotia"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C15"
  Label"New Brunswick"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C16"
  Label"Quebec"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C17"
  Label"Ontario"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C18"
  Label"Manitoba"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C19"
  Label"Saskatchewan"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C110"
  Label"Alberta"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C111"
  Label"British Columbia"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C112"
  Label"Yukon Territory"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C113"
  Label"Northwest Territories"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C114"
  Label"Nunavut"gtlt/CategoryNodegt
• ltCategoryNode CategoryNodeOID"C2"
  Label"Year (Virtual)"gtlt/CategoryNodegt

11
Verification tool category headers for a
selected content cell
12
Verification toolcontent cells for a selected
header
13
Verification toolhierarchical category
structure for a selected content cell
14
(b) Query by Table
Income 2002 4500 2003 3300 2004 1240 2
005 3400
Income 2002 2003 2004 2005
QBT
InterpretQuery Table
Database
Ontology from many tables
15
Query Table
Composed in MS-Excel by a person seeking
information from an ontology compiled from many
web tables
16
Display of automatically processed Query Table
for human verification
17
Wang notation for Query Table
18
QBT identifies requested data
19
URLs of tables in the Example Database

• Median Total Income http//www40.statcan.ca/l0
  1/cst01/famil108a.htm?sdimedian20income
• Number of Induced Abortions http//www40.statcan
  .ca/l01/cst01/health40a.htm?sdiabortions
• Number of Divorces http//www40.statcan.ca/l
  01/cst01/famil02.htm?sdinumber20divorces
• Infant Mortality Rate http//www40.statcan.ca
  /l01/cst01/health21a.htm?sdiinfant20mortality20
  rate
• Trips By Canadians in Canada http//www40.statc
  an.ca/l01/cst01/arts26a.htm
• Number of Homicideshttp//www40.statcan.ca/l01/c
  st01/legal12a.htm?sdihomicide
• Populationhttp//www40.statcan.ca/l01/cst01/demo
  02a.htm?sdipopulation
• Number of Persons with Diabetes
• http//www40.statcan.ca/l01/cst01/health54a.
  htm?sdidiabetes
• Number of Persons with Asthma
• http//www40.statcan.ca/l01/cst01/health50a.htm?sd
  iasthma
• University Degrees Awarded to Males
• http//www40.statcan.ca/l01/cst01/educ51b.ht
  m
• University Degrees Awarded to Females
• http//www40.statcan.ca/l01/cst01/educ51c.ht
  m
• Food services and drinking places (13
  tables)http//www40.statcan.ca/l01/cst01/serv24j
  

20
Fields in the Example Database

• IDENTIFIER
• REGION
• YEAR
• NUMBER_OF_ABORTIONS
• ABORTION_RATE
• NUMBER_OF_DIVORCES
• INFANT_MORTALITY_RATE
• NUMBER_OF_TRIPS
• MEDIAN_TOTAL_INCOME
• POPULATION
• NUMBER_OF_HOMICIDES
• GENDER
• INCIDENCE_OF_DIABETES
• UNIVERSITY_DEGREES_AWARDED
• INCIDENCE_OF_ASTHMA
• RESTAURANT_OPERATING_REVENUE
• RESTAURANT_OPERATING_EXPENSES
• RESTAURANT_OPERATING_PROFIT_MARGIN
• RESTAURANT_OPERATING_WAGES

21
QBT fills in requested data from Example Database
22
A current puzzle
Year Region Gender Diabetics
2002 Alberta Male XX
2002 Alberta Female XX
2002 Ontario Male XX
2002 Ontario Female XX
Year Region Diabetics Diabetics
Year Region Male Female
2002 Alberta XX XX
2002 Ontario XX XX

• How can QBT tell that these two query tables
  represent the same request? NB Although
  plausible, both of these tables exemplify poor
  layout.

23
Next steps

• Complete the conversion of Wang/XML table
  descriptions to mini ontologies
• Improve the interface for generating cumulative
  ontology from mini ontologies
• Implement database generation from ontology
• Embed logging routines for statistical evaluation
  of time/error trade-offs