Semi-Automatic Generation of Mini-Ontologies from Canonicalized Relational Tables

1
Semi-Automatic Generation of Mini-Ontologies from
Canonicalized Relational Tables

• Chris Hathaway

Supported by NSF
2
Introduction

• Ontologies are an important tool for realizing
  the vision of the semantic web
• Major setback - creation and upkeep
• Created by experts
• Experts are biased in knowledge, agreement needed
• Ontologies continually change
• Some automation is needed

3
Introduction (contd)

• Current attempts at automatic generation of
  ontologies not successful, because extracted from
  free-form, unstructured text.
• A more effective alternative is to extract
  ontologies from structured data on the web
  (tables, charts, etc.)
• TANGO project
• Part 1 Extract tables from the web
• Part 2 Define mini-ontologies from tables
• Part 3 Merge into growing domain ontology

4
Process Overview

• Start with canonicalized table
• Generate candidates for
• Object Sets
• Relationship Sets
• Functional Constraints
• Inclusion Constraints/Hierarchical Structure
• Get help from user
• Choose best candidate for the ontology

5
Example 1 Generate Concepts
Create list of candidate concepts (usually column
names)
6
Example 1 Generate Concepts
Current ontology
7
Example 1 Generate Relationships

• Decide relationship sets
• Exponential number of combinations
• Basic assumption one main concept relates to all
  others (attributes)
• Goal find central column of interest

8
Example 1 Generate Relationships
Look for mapping between one column and title of
table
9
Example 1 Generate Relationships
Current ontology
10
Example 1 Generate Constraints

• FDs and Participation Constraints
• FD definition X ? Y iff (Xi Xj) ? (Yi
  Yj) for all row indexes i and j.
• Unless solid case (two or more same values), only
  consider FDs from central object to attributes
• Use heuristics for setting exact participation
  (01,1, etc)

11
Example 1 Generate Concepts
Numerical values are usually functionally
determined by column of interest and have 0
participation constraint.
12
Example 1 Generate Constraints
Completed mini-ontology
13
Example 2 Generate Concepts

• SubFamily, Group, and SubGroup are generic types
• Enumerate column values as object sets because
  less than 5 divisions (recursively)

14
Example 2 Generate Relationships

• Found mapping of central column of interest to
  title (Language)
• Create ISA hierarchy from table structure

15
Example 2 Generate Relationships
Current ontology
16
Example 2 Generate Hierarchical Constraints

• Assign members to each object set for easy
  calculation
• Find inclusion dependencies
• Union All members of parents are members of one
  or more child
• Intersection (Less common) Child members are
  always in both parents
• Mutual exclusion Intersection of any two child
  members is empty.

17
Example 2 Generate Hierarchical Constraints
Completed mini-ontology
18
Getting Help from the User

• Sometimes human intervention is required
• Effective use of the users input will rely on
  IDS statements
• Issue explains the problem (Ex. No central
  object was found in the table)
• Default describes default behavior (Ex. A new
  non-lexical object named Object will be created)
• Suggestion suggests an action for the user to
  follow (Ex. Either choose the column central to
  describing the table, or name the new object set
  something appropriate)

19
Conclusion

• Successful transformation from table to ontology
• Develop a set of rules, assumptions, heuristics,
  etc. to automate most accurately
• Greater ease for the ontology creator