TY - GEN
T1 - Concept Learning For Achieving Personalized Ontologies
T2 - An Active Learning Approach
AU - Sensoy, Murat
AU - Yolum, Pinar
PY - 2009/8/4
Y1 - 2009/8/4
N2 - In many multiagent approaches, it is usual to assume the existence of a common ontology among agents. However, in dynamic systems, the existence of such an ontology is unrealistic and its maintenance is cumbersome. Burden of maintaining a common ontology can be alleviated by enabling agents to evolve their ontologies personally. However, with different ontologies, agents are likely to run into communication problems since their vocabularies are different from each other. Therefore, to achieve personalized ontologies, agents must have a means to understand the concepts used by others. Consequently, this paper proposes an approach that enables agents to teach each other concepts from their ontologies using examples. Unlike other concept learning approaches, our approach enables the learner to elicit most informative examples interactively from the teacher. Hence, the learner participates to the learning process actively. We empirically compare the proposed approach with the previous concept learning approaches. Our experiments show that using the proposed approach, agents can learn new concepts successfully and with fewer examples.
AB - In many multiagent approaches, it is usual to assume the existence of a common ontology among agents. However, in dynamic systems, the existence of such an ontology is unrealistic and its maintenance is cumbersome. Burden of maintaining a common ontology can be alleviated by enabling agents to evolve their ontologies personally. However, with different ontologies, agents are likely to run into communication problems since their vocabularies are different from each other. Therefore, to achieve personalized ontologies, agents must have a means to understand the concepts used by others. Consequently, this paper proposes an approach that enables agents to teach each other concepts from their ontologies using examples. Unlike other concept learning approaches, our approach enables the learner to elicit most informative examples interactively from the teacher. Hence, the learner participates to the learning process actively. We empirically compare the proposed approach with the previous concept learning approaches. Our experiments show that using the proposed approach, agents can learn new concepts successfully and with fewer examples.
KW - concept learning
KW - data mining
U2 - 10.1007/978-3-642-03603-3_13
DO - 10.1007/978-3-642-03603-3_13
M3 - Published conference contribution
SN - 978-3-642-03602-6
VL - 5680
T3 - Lecture Notes in Computer Science
SP - 170
EP - 182
BT - Agents and Data Mining Interaction
A2 - Cao, Longbing
A2 - Gorodetsky, Vladimir
A2 - Liu, Jiming
A2 - Weiss, Gerhard
A2 - Yu, Philip S
PB - Springer Berlin / Heidelberg
ER -