An Overview of Schema Matching Approaches

1
An Overview of Schema Matching Approaches

• Alsayed Algergawy

2
Outline

• What is schema matching?
• Where is schema matching used?
• Schema matching challenges
• Generic schema matching system architecture
• Schema matching approaches
• Existing prototypes and comparison
• Conclusions

3
Why surveying?

• Survey is useful when
• Comparing different approaches of schema matching
• Developing a new match algorithm
• Implementing a schema matching component

4
Schema Matching

• Schema matching is define as the task of
  finding the semantic correspondences between
  elements of two schemas.

S1
Match
Match Result
S2
Auxiliary information
5

• Input information
• Schema information
• Instance data
• Auxiliary information
• Output information
• A mapping is defined as a set of mapping
  elements each of which specifies that certain
  elements of S1 are mapped to certain elements of
  S2 ltID, Si1 , Sj2 ,Rgt

Element names, data types, constraints,
Used to characterize the content and semantics of
schema elements
Dictionaries, user input, previous match result
Mapping (match result)
6
Where is schema matching used?

• To motivate the importance of schema
  matching, we summarize its use is several
  application domains
• Database application domains
• Data integration
• Data warehouse
• E-commerce
• Query processing
• Peer data management
• Model management
• Semantic Web
• Semantic web services
• Xml/html to an ontology

7
Data Integration

• Problem Construct a global view from a set of
  independently constructed schemas.
• - Different structure and terminologies
• In the AI setting, this is the problem of
  integrating independently developed ontologies
  into a single ontology.
• Solution Schema Matching is performed to find
  relationships between concepts in each schema.
  Then the matching elements can be unified.

8
Data Warehouse

• Problem Integrating data sources into a data
  warehouse.
• - Different formats between the source and
  warehouse.
• Solution Use matching to find the elements of
  the source that are also present in the
  warehouse. Then the details of the semantics can
  be examined to integrate the two.

9
E-commerce

• Problem Message translation.
• -Each trading partner uses its own message
  format.
• Solution A match operation would reduce the
  amount of manual work to specify how the formats
  are related.

10
Query Processing

• Problem The terms used in the users query may
  be different from those in the database.
• Solution Matching is used to map the
  user-specified concepts in the query to schema
  elements.

11
Challenges of Schema Matching

• Despite its pervasiveness and importance,
  schema matching remains an extremely difficult
  problem
• The semantics of the involved elements can be
  inferred from only a few information sources,
  typically the creators of data, documentation,
  and associated schema and data.
• Schema elements are matched based on clues in the
  schema and data.
• Schema and data clues are often incomplete.
• To decide that element s of schema S matches
  element t of schema T, one must examine all other
  elements of T to make sure that there is no other
  element that matches s better than t.
• To make matters worse, matching is often
  subjective, depending on the application.

12
Challenges (cont.)

• Currently, the schema matching is largely
  performed manually, then it is a
• tedious
• time consuming
• error prone, and
• expensive
• process.
• To reduce the amount of manual effort as much
  as possible, approaches to semi automatically are
  required.
• It is not possible to fully automatically
  determined all correspondences between two
  schemas due to of their semantic heterogeneity

13
Generic Schema Matching System Architecture
Application/Tool1 (semantic web)
Application/Tool 2 (E-commerce)
Application/Tool 3 (warehousing)
Application/Tool 4 (schema integration)

Schema import/export
Generic match implementation
General libraries
Internal schema representation
14
General Schema Matching Procedure

• The schema matching process requires the
  following main steps
• Importing external schemas
• Identifying the elements to be matched
• Applying the match algorithm(s)
• Exporting the match results

15
Schema Matching Approaches

• The schema matching approaches differ from
  each other according to
• The input information
• The way the input information is processed, and
• The characteristics of the output information
• It can be
• Individual / combining
• For individual matchers
• Schema vs. instance
• Element vs. structure
• Language vs. constraint
• Matching cardinality
• Auxiliary information
• For the combining matchers
• Hybrid vs. composite

16
Classification of Schema Matching Approaches
Schema Matching Approaches
Individual Matchers
Combining Matchers
Schema-based
Instance-based
Hybrid Matchers
Composite Matchers
Structure Level
Element Level
Element Level
Constraint-based
Linguistic
Constraint-based
Constraint-based
Linguistic
Further Criteria -Match Cardinality
-Auxiliary information used

• Word Frequency

• Name Similarity
• Description Similarity
• Global Namespaces

• Group Matching

• Type Similarity
• Key Properties

• Value Pattern and Ranges

Sample Approaches
Bernstein P, Rahm E. A survey of approaches to
automatic schema matching
17
Schema-based Approaches

• considering only schema information
• including different properties of schema elements
  and relationships between them
• applying either on element level or structure
  level
• The two main approaches are linguistic-based and
  constraint-based

18
Linguistic-based Approaches
Syntactically Comparing name strings
Semantically Comparing name meaning
Exact string matching (equality of names)
approximate string matching (n-gram, soundex,)
Make use of auxiliary source, terminological
relationships, (synonyms, )
19
The issues affecting name matching

• Multi-word names
• Cryptic names
• Homonyms

Preprocessing (tokenization, removing of stop
words,)
Abbreviation dictionary are required
20
Instance-based Approaches

• Why?
• When useful schema information is limited
• To complement schema-based approaches
• To match instance-level data
• Limitation
• The availability of representative data
• The quality of the provided instance data
• The execution time

21
Instance-based approaches
Element-level
Structure-level
Heuristic comparison
AI techniques
Text-based
Numerical/string
Using IR techniques keywords applying name
matching
Constraint-based chts data length, data type,
value range, value distribution
22
Combination Approaches
23
Existing Prototypes
Schema-based Prototypes
Instance-based Prototypes

• SEMINT 94,2000
• AutoPlex 2001
• CLIO 2001
• LSD 2001,2002
• Corpus-based matching 2003
• IMAP 2004

• DELTA 95
• DIKE 98
• MOMIS 99
• CUPID 2001
• SF 2002
• COMA 2001
• COMA 2005

24
Prototype Comparison

• Architecture
• Schema representation
• Result (mapping) representation
• Input information
• Matching algorithms
• Match execution

25
(No Transcript)
26
(No Transcript)
27
(No Transcript)
28
Conclusions

• Schema matching is such a pervasive, important,
  and extremely difficult problem
• Schema matching is primarily studied as a piece
  of other applications
• It is should be studied independently
• Schema matching is still performed manually
• The existing approaches are incomplete solutions,
  exploiting at most a few approaches

29
References

• Aumüller, D., H.H. Do, S. Massmann, E. Rahm
  Schema and Ontology Matching with COMA
  (Software Demonstration). Proc. 24. ACM SIGMOD
  Intl. Conf. Management of Data, 2005
• Berlin, J., A. Motro Database Schema Matching
  Using Machine Learning with Feature Selection.
  Proc. 14. Intl. Conf. Advanced Information
  Systems Engineering (CAiSE), 2002
• Clifton, C., E. Housman, A. Rosenthal Experience
  with a Combined Approach to Attribute-Matching
  Across Heterogeneous Databases. Proc. IFIP 2.6
  Working Conf. Database Semantics, 1996
• Do, H.H., E. Rahm COMA - A System for Flexible
  Combination of Schema Matching Approach. Proc.
  Intl. Conf. Very Large Databases (VLDB), 2002
• Doan, A.H., A. Halevy Semantic Integration
  Research in the Database Community A Brief
  Survey. AI Magazine, Special Issue on Semantic
  Integration, 2005
• Rahm, E., P.A. Bernstein A Survey of Approaches
  to Automatic Schema Matching. VLDB Journal,
  10(4), 2001

30
Thank you