1. TOQL: Temporal Ontology Querying Language E. Baratis, E. G. M
TOQL: Temporal Ontology Querying Language E. Baratis, E.G.M. Petrakis, S. Batsakis, N. Maris, N. Papadakis
Intelligent Systems Laboratory
Dept. of Electronic and Computer Engineering
Technical University of Crete (TUC)
Chania Crete, Greece

2. Query on information that changes over time in ontologies
Problem Definition
12/1/2018
Query on information that changes over time in ontologies
The price of a product at time 5
Representation of concepts / events that change in time in ontologies
TOQL tackles both these problems
SSTD'09 Aalborg, Denmark

3. OWL: binary relationships
State-Of-The-Art
12/1/2018
OWL: binary relationships
Produces(C,P)
Produces(C,P,t) ?
OWL-Time: ontology of temporal concepts
SPARQL for static ontologies
No temporal operators
The user needs to be aware of the representation of time
SSTD'09 Aalborg, Denmark

4. Representation of Time
12/1/2018
Representation of the evolution of concepts (events) in ontologies
Reification : extends n-ary relations to binary relations
Versioning: new ontology for each change
Extensive searches, restricted use of OWL language constructs (inverse, transitive, functional), reasoning can’t be applied
SSTD'09 Aalborg, Denmark

5. 4D Fluent (Perdurantist) Aproach
12/1/2018
Classes TimeSlice, TimeInterval are introduced
Dynamic classes become instances of TimeSlice
Temporal properties of dynamic classes become instances of TimeInterval
A time slice object is created each time a (fluent) property changes
SSTD'09 Aalborg, Denmark

6. Sample Ontology (Static)
12/1/2018
Employee
Company
Product
string
int
Has Employee
Produces
Name
Price
Static classes/properties
Dynamic classes/properties
SSTD'09 Aalborg, Denmark

7. Dynamic Ontology Employee Company Product tsTimeSliceOf tsTimeSliceOf
12/1/2018
Employee
Company
Product
Employee
Name
Company
Name
string
string
tsTimeSliceOf
tsTimeSliceOf
tsTimeSliceOf
TimeSlice
SSTD'09 Aalborg, Denmark
IS-A
IS-A
IS-A
EmployeeTimeSlice
Company
TimeSlice
Product
TimeSlice
Has Employee
Produces
Product
Name
Price
int
string
tsTimeInterval
tsTimeInterval
TimeInterval

8. Disadvantages: data redundancy
Comments
12/1/2018
The 4D perdurantist (fluent) solves problems of reification, versioning
Changes affect only related objects not the entire ontology
Reasoning mechanisms and semantics of OWL fully supported
Disadvantages: data redundancy
SSTD'09 Aalborg, Denmark

9. Handles ontologies like rel. DBs
TOQL
12/1/2018
Handles ontologies like rel. DBs
SQL-like
Relational Database
Ontology
Table - concept
Class
Table - relationship
Object properties
Attributes
Datatype properties
SSTD'09 Aalborg, Denmark
SELECT
<Object/Datatype property> …
FROM
<Class> …
WHERE
<Condition>

10. Introduces the AT operator for querying on temporal information
TOQL FEATURES
12/1/2018
Implements ALLEN operators between time instants and intervals
before, after, equals, meets, overlaps, during, starts, ends
Introduces the AT operator for querying on temporal information
AT(time instant), AT(time interval)
Supports nested queries and connecting clauses
minus, union, intersect, exist, all, any, in
Wildcards (*)
SSTD'09 Aalborg, Denmark

11. Return companies producing “x” at time 4
Simple TOQL Example
12/1/2018
Return companies producing “x” at time 4
SELECT Company.companyName
FROM Company, Product
WHERE Company.produces:Product
AND Product.productName LIKE “x”
AT (4)
SSTD'09 Aalborg, Denmark

12. Without TOQL Functionality
12/1/2018
SELECT Company.companyName
FROM Company, Employee,
TimeSlice AS T1, TimeSlice AS T2
WHERE T1.tsTimeSliceOf:Company AND T2.tsTimeSliceOf:Employee AND T1.hasEmployee:T2 AND
Employee.employeeName LIKE “x“ AT (4)
SSTD'09 Aalborg, Denmark

13. Return companies which employed “x” during (3,5)
Temporal Example
12/1/2018
Return companies which employed “x” during (3,5)
SELECT Company.companyName
FROM Company, Employee
WHERE Company.hasEmployee:Employee AT(3,5) AND Employee.employeeName LIKE “x"
SSTD'09 Aalborg, Denmark

14. Example with ALLEN operator
12/1/2018
Find companies which employed “x” before “y”
SELECT Company.companyName
FROM Company, Employee AS E1, Employee AS E2
WHERE Company.hasEmployee:E1 BEFORE Company.hasEmployee:E2
AND E1.employeeName LIKE “x” AND E2.employeeName LIKE “y”
SSTD'09 Aalborg, Denmark

15. Translated into equivalent SeRQL ones
Query Translation
12/1/2018
Translated into equivalent SeRQL ones
Ontology Parser
Query Interpreter
Knowledge Base (KB)
KB Querying
OWL ontology
TOQL query
SeRQL
query
Answer
Error message
Reasoner
SSTD'09 Aalborg, Denmark

16. Find products produced by company “C1” at time 2
Transation Example
12/1/2018
Find products produced by company “C1” at time 2
SELECT Product
FROM Company, Product
WHERE Company.produces:Product AT(2)
AND Company.companyName LIKE "C1"
SSTD'09 Aalborg, Denmark

17. Equivalent SeRQL query
12/1/2018
SELECT Product
FROM {Company} rdf:type {ex1:Company},
{Product} rdf:type {ex1:Product},
{CompanySlice_1} rdf:type {ex1:TimeSlice},
{interval1} rdf:type {ex1:TimeInterval},
{CompanySlice_1} ex1:produces {ProductSlice_1},
{Company} ex1:companyName {companyName_Company},
{CompanySlice_1} ex1:tsTimeSliceOf {Company},
{ProductSlice_1} ex1:tsTimeSliceOf {Product},
{interval1} ex1:startValue {startValue1},
{interval1} ex1:endValue {endValue1},
{ProductSlice_1} ex1:tsTimeInterval {interval1}
WHERE startValue1 <= "2"^^xsd:int AND endValue1 >= "2"^^xsd:int AND companyName_Company Like "C1"
USING NAMESPACE ex1= < owl#>
SSTD'09 Aalborg, Denmark

18. Infer information from ontology schema
Reasoning in TOQL
12/1/2018
Infer information from ontology schema
A query on a class will return the instances of its sub-classes
e.g., a query on class COMPANY will also return instances of COMUTER COMPANIES
The PELLET reasoner applies to the ontology schema at loading time to derive these relationships
SSTD'09 Aalborg, Denmark

19. Infer information from existing information
Temporal Reasoner
12/1/2018
Infer information from existing information
e.g., the price of a product remains the same since the last time it was changed
Implemented in PROLOG based on Event Calculus
Applies to functional fluent properties i.e., properties that have one value at each time instant
SSTD'09 Aalborg, Denmark

20. Example of Reasoner use
12/1/2018
If the price of “P4” is 50 at time 2 and becomes 60 at time 4, then TOQL will return null
SELECT Product
FROM Product
WHERE Product.price LIKE “50” AT(9)
Without the reasoner the TOQL will return “P4”
SSTD'09 Aalborg, Denmark

21. Summary of Query Translation
12/1/2018
Loads the ontology
Applies PELLET for class / subclass info
Parses the query and report error messages
Translates the query to SeRQL
Uses the dynamic part of ontology to answer queries addressing fluent properties
Applies the temporal reasoner on functional fluent properties
Uses the static part of ontology to answer queries addressing static properties
SSTD'09 Aalborg, Denmark

22. Conclusions
12/1/2018
TOQL is new query language for retrieving temporal information from OWL ontologies
Simple SQL – like syntax, with more powerful temporal operators than SPARQL or SeRQL
Independent of representation of time
SSTD'09 Aalborg, Denmark

23. Solve performance issues (indexing) Query optimization
Future Work
12/1/2018
Representation of Spatial / Spatio- temporal information representation and operators
Solve performance issues (indexing)
Query optimization
Open schema query support (class as variable in FROM)
Addressing ontology structure (support of CLASS / SUBCLASS keywords)
SSTD'09 Aalborg, Denmark