- Tytuł:
- Handling the description noise using an attribute value ontology
- Autorzy:
-
Łukaszewski, T.
Józefowska, J.
Ławrynowicz, A.
Józefowski, Ł. - Powiązania:
- https://bibliotekanauki.pl/articles/206376.pdf
- Data publikacji:
- 2011
- Wydawca:
- Polska Akademia Nauk. Instytut Badań Systemowych PAN
- Tematy:
-
imprecise descriptions
attribute noise
ontology
naďve Bayesian classifier - Opis:
- The quality of any classifier depends on a number of factors, including the quality of training data. In real-world scenarios, data are often noisy. One reason for noisy data (erroneous values) is in the representation language, insufficient to model different levels of knowledge granularity. In this paper, to address the problem of such description noise, we propose a novel extension of the na've Bayesian classifier by an attribute value ontology (AVO). In the proposed approach, every attribute is a hierarchy of concepts from the domain knowledge base. In this way an example is described either very precisely (using a concept from the low-level of the hierarchy) or, when it is not possible, in a more general way (using a concept from higher levels of the hierarchy). Our general strategy is to classify a new example using training examples described in the same way or more precisely at lower levels of knowledge granularity. Hence, the hierarchy introduces a bias which in effect can contribute to improvement of a classification.
- Źródło:
-
Control and Cybernetics; 2011, 40, 2; 275-292
0324-8569 - Pojawia się w:
- Control and Cybernetics
- Dostawca treści:
- Biblioteka Nauki
Artykuł