Xem mẫu
A Basis for the Semantic Web and E-Business
IUDPH ZH VHOHFW WKH ³3HUVRQB2QWRORJ\´ DQG LQWKH³&KLOGRUJUDQGFKLOG2QWRORJLHV´IUDPH ZHLQSXWWKHRQWRORJ\QDPH³6WXGHQW´DQGWKH QDPHVSDFH ³VWX´ IRU WKLV 6WXGHQWB2QWRORJ\ :KHQFOLFNLQJWKH³,QKHULWDQFH´EXWWRQDVLPSOH Student_Ontology is automatically created which ZLOOEHVKRZQLQWKHULJKW³&RGHV´IUDPH,QDG-GLWLRQDVZHVHOHFWWKH³3HUVRQB2QWRORJ\´LQWKH ³3DUHQW2QWRORJLHV´IUDPHDOOWKHFRQFHSWVRI ³3HUVRQB2QWRORJ\´ZLOOEHOLVWHGLQWKH³6HOHFW &RQFHSWV´ FRPER LQ WKH ³3DUHQW 2QWRORJLHV´ IUDPH VRPH FRQFHSWV RI ³3HUVRQB2QWRORJ\´ DUHGH¿QHGLQ)LJXUHWKHQZHFDQVHOHFWRQH FRQFHSWDQGE\FOLFNLQJWKH³%ORFN´EXWWRQD certain concept of the parent ontology is blocked in the child ontology. Also, after selecting a con-cept from the parent ontology and clicking the ³0XWDWLRQ´EXWWRQZHFDQLQGLFDWHWKDWFHUWDLQ concept of the parent ontology is mutated in the FKLOGRQWRORJ\)URPWKHULJKW³&RGHV´IUDPHRI )LJXUHZHFDQVHHWKDW³RI¿FHBSKRQH´LVEORFNHG E\6WXGHQWB2QWRORJ\DQG³FRQWDFWBQR´LVPXWDWHG LQ6WXGHQWB2QWRORJ\,IZHFOLFNWKH³6DYH´EXW-ton, the new Student_Ontology will be saved in D¿OHEXWLIWKHUHDUHSUREOHPVZHFDQUROOEDFN
10 steps. When a new ontology is created, it will EHDXWRPDWLFDOO\OLVWHGLQWKH³6HOHFW2QWRORJLHV´ FRPERVRI³*UDQGSDUHQW2QWRORJLHV´DQG³3DUHQW Ontologies” frames. Similarly, we can use the atavism operation to indicate that some concepts of the grandparent ontology are atavismed in the grandchild ontology or in the offspring ontolo-gies of the grandchild ontologies. Note that the ³JPRH´LQ)LJXUHLQGLFDWHVWKDWWKHRSHUDWLRQV LQWKH³2QWRORJ\/DQJXDJH2UJDQL]DWLRQ´VHFWLRQ are a genetic model.
This tool is a prototype to indicate that the inheritance, block, atavism, and mutation opera-tions really work in organize ontology language and ontologies. This prototype tool can be further improved for commercial use.
Next we summarize the guidelines of how to organize information in ontologies, that is, different information should be put at different hierarchies of ontologies.
The general concepts in a domain should be put in the highest level ontologies, for example, O1 in Figure 8. Here O represents Ontologies. If VRPHFRQFHSWVDUHVSHFL¿FWKH\VKRXOGEHSXWLQ the lower level ontologies, for example, O2 and
Figure 7. A graphical tool for ontology language and ontology organization
2314
A Basis for the Semantic Web and E-Business
Figure 8. Architecture of building ontology systems
O1 …
O2 O3 …
O4 O5 O6 O7 O8 O9 …
O3 in Figure 8. When some concepts are more VSHFL¿FWKH\VKRXOGEHSXWLQHYHQORZHURQWROR-gies, for example, O4-O9 in Figure 8. Figure 8 shows the hierarchy of ontologies. We allow multiple inheritance in ontology organizations, for example, O6 inherits both O2 and O3. In practice, the hierarchies can be more than three levels.
The hierarchy of ontologies is similar to the hierarchy of ontology languages. However, because the concepts in ontologies will change (add in, move out, and update), next we mainly GLVFXVVKRZWRUHVROYHWKHFRQÀLFWVLQRQWRORJ\ organizations.
5HVROYH&RQÀLFWVLQOntology Organization
Kalfoglou and Schorlemmer (2003) survey the related works on ontology mapping and indicate WKDWPRVWRIWKHSUHYLRXVZRUNVDUHDERXW¿QGLQJ the similarities and differences among ontolo-gies, then the ontologies can be accessed from a common layer. There are no related works on UHVROYLQJWKHFRQÀLFWVLQGHVLJQRQWRORJLHV+HUH ZHGLVFXVVVRPHWHFKQLTXHVWRUHVROYHFRQÀLFWV
When designing ontologies with hierarchies, it is important to keep the ontologies consistent. $FRQFHSWLVVSHFL¿HGLQDQRQWRORJ\LILWLV HLWKHUGH¿QHGRUUHGH¿QHGIRUWKHRQWRORJ\$ UHGH¿QHGFRQFHSWRYHUORDGVDVLPLODUFRQFHSW in some ancestor ontologies. Figure 9 shows the hierarchies of ontologies. The O in Figure 9 represents ontologies which are displayed as rounded rectangles, and the C in Figure 9 rep-UHVHQWVFRQFHSWVGH¿QHGLQRQWRORJLHVZKLFKDUH displayed as parallelograms.
In this section, we discuss how to resolve the FRQÀLFWV$QLQKHULWHGFRQFHSWLVZHOOGH¿QHGLI LWLVVSHFL¿HGLQRQHDQGRQO\RQHDQFHVWRURQWRO-RJ\SRVVLEO\LQGLUHFW$FRQÀLFWVLWXDWLRQH[LVWV ZKHQDQLQKHULWHGFRQFHSWLVQRWZHOOGH¿QHG that is, two or more ancestor ontologies specify the same concept. For example, from Figure 9, we can see that concept C1 of ontology O2 LVUHGH¿QHGLQRQWRORJLHV22DQG2& FRQWULEXWHVWRDFRQÀLFWVLWXDWLRQLQ2EXW& LVZHOOGH¿QHGLQ2
We have the following methods to solve the FRQÀLFWSUREOHP
in designing ontologies with hierarchies. 5HGH¿QLQJRURYHUULGLQJ
2315
A Basis for the Semantic Web and E-Business
The C2 in O9 and O2 in Figure 9 have the VDPHQDPHWKXVLWPD\EHDFRQÀLFW+RZHYHU LI&LQ2LVGH¿QHGWRRYHUULGHWKH&LQ2
3. Redesigning the organizations of ontologies (e.g. factoring)
DQGUHGH¿QHG&ZLWKGLIIHUHQWPHDQLQJWKHQ WKHUHDUHQRFRQÀLFWV
We use the ontology hierarchies shown in )LJXUH WR LQWURGXFH WKLV FRQÀLFW UHVROYLQJ approach. The two Cs in ontologies O2 and O3
2. Explicitly selecting or renaming have the same semantics, and they have the same name. Obviously, there will be confusion when O4
We use an example to show how to use explic-LWO\VHOHFWLQJRUUHQDPLQJWRVROYHFRQÀLFWV
Example 9. If the two C4 in O3 and O1 of Figure 9 have the different semantics, there will EHDFRQÀLFWLQ27RVROYHWKLVFRQÀLFWZHKDYH WZRRSWLRQV7KH¿UVWRSWLRQKDVWKHRQWRORJ\ designer explicitly mention that the C4 in O9 is inherited from the C4 in O3. However, explicitly
inherits C from O2 and O3. In ontology design, the semantics of each concept in the ontology should be clear without any ambiguities because the concepts are shared by the Semantic Web or e-business applications for semantic information processing.
7RSURFHVVWKLVFRQÀLFWWKHUHDUHWZRFDVHV to consider.
selecting has a problem, that is, some informa-tion will be lost. If O9 explicitly mentions that O9 uses the C4 in O3, the information of the C4 in O1 can not be inherited by O9, which is a loss of information. The second option to process this FRQÀLFWLVUHQDPHWKH&LQHLWKHU2RU2RU both; in this way, all the information can be kept without lost.
1. If O1= O2 O3, Figure 11. shows that we can factor C to the parent ontology of Q2 and O3, that is, O1. In this way, O4 inherits concept C from a single ancestor ontology, WKHUHIRUHWKHUHDUHQRFRQÀLFWV
2. If O1 O2 O3, then we create ontology O5 such that O5 = O2 O3, and factor C to
)LJXUH&RQÀLFWVLQRQWRORJ\GHVLJQ
O1
O2
C4
O3 C1 C2 O4
C4
C1 O5 C3
O6
C1 O7 O8
C1
O9
C2
2316
A Basis for the Semantic Web and E-Business
)LJXUH5HVROYHFRQÀLFWVE\UHGHVLJQLQJWKHRUJDQL]DWLRQVRIRQWRORJLHV
O1
O2 O3
C C
O4
Figure 11. Factor to parent ontology
O1
C1
Figure 12. Factor to an intermediate level of ontology
O1
O2 O3 O5
C1
O4
O2 O3
O4
O5. Figure 12 shows this approach. In this ZD\WKHFRQÀLFWFDQEHUHVROYHGDQGWKH& is at an appropriate level.
$OJRULWKPWR5HVROYH&RQÀLFWV
)LJXUHVKRZVWKHDOJRULWKPWRUHVROYHFRQÀLFWV which is a formal summary of the cases in the ³5HVROYH&RQÀLFWVLQ2QWRORJ\2UJDQL]DWLRQ´ section.
:LWKWKHVHFRQÀLFWSURFHVVLQJDSSURDFKHV when inserting concepts into or deleting concepts
from ontologies, we should be careful to make WKHRQWRORJLHVFRQVLVWHQWZLWKRXWFRQÀLFWV/LQJ & Teo, 1993).
SEMANTIC INFORMATION PROCESSING IN THE SEMANTIC WEB AND E-BUSINESS
The present Web exists in the HTML and XML formats for persons to browse. Recently there is a trend towards the Semantic Web where the
2317
A Basis for the Semantic Web and E-Business
)LJXUH$OJRULWKPWRUHVROYHFRQÀLFWV
Given ontologies with hierarchies
FOR each conflict situation in the hierarchy DO
Let the conflict situation be ontologies A, B1, …, Bn (n > 1) where B1, …, Bn are the nearest ancestor ontologies of A that specify a property p.
/* Note that a ancestor ontology of some Bi may itself specify a property p. */ /* Check the semantics of p in B1, …, Bn */
IF semantics of p is the same in B1, …, Bn THEN IF intersection of B1, …, Bn is empty THEN
***Design error, since ontology A (which is the intersection of B1, …, Bn) is empty ELSE
******/* same semantics (Factoring) */
IF there exists a more general ontology K which is UNION of B1, …, Bn THEN Factor p to ontology K
ELSE
Resolve the conflict by either:
(a) creating a general ontology K that is the UNION of B1, …, Bn and factoring p to K.
OR
(b) Explicitly choosing one parent ontology to inherit the property. ENDIF
ENDIF ELSE
/* different semantics */
Let G1, G2, …, Gm be sets of mutually exclusive ontologies from B1, …, Bn such that ontologies in a group share the same semantics for p. Resolve the conflict in A by adopting one of the following:
(a) redefine p in ontology A, /* not a good solution */ or
(b) Rename p in Gj to, say, p_Gj for j = 1, …, m to reflect their different semantics. To conform to the unique name assumption. Each p in the schema that has the same semantics as P_Gj must be renamed to p_Gj.
FOR each group Gj (j = 1, …, m) with 2 or more ontolgoies having property p_Gj DO
/* An conflict situation exists between ontology A and the ontologies in Gj;*/ /* p_Gj has the same semantics in the ontologies of Gj */
Resolve the conflict in ontology A using the method described in *** and ******.
ENDFOR ENDIF
ENDFOR
information can be processed and understood by a computer. The present e-business also requires that the semantic information can be automati-cally exchanged among different agents of the e-business partners.
When the concepts in different ontologies are GH¿QHGZLWKclear semantics andZLWKRXWFRQÀLFWV, the sharing concepts in ontologies can be used to annotate the Semantic Web pages or the agents of the e-business partners. If the information in two
different Semantic Web pages refers to the same concept from the same ontology, the information has the same semantics, otherwise the information is different in the two Semantic Web pages. This can be automatically recognized by the computer. It is similar for the semantic information process-ing in e-business.
We use an example to show how to achieve the automatically and semantically exchange of information.
2318
...
- tailieumienphi.vn
nguon tai.lieu . vn