Turney, P.D., The identification of context-sensitive features: A formal definition of context for concept learning. in ICML-96 Workshop on Learning in Context-Sensitive Domains, (Bari, Italy, 1996), 53-59.  Home/Search   Document Details and Download   Summary   Related Articles   Check

.... machine learning strategy with a meta level algorithm utilising this information (e.g. 27] Others look to augment strategies using implicit information about the context to adjust features of the learning such as the weightings [6] or normalisation [26] Turney discusses the problem in [28]. He surveys the various heuristics tried to mitigate the effects of context on machine learning techniques in [29] He maintains a bibliography on context sensitive learning at URL: http: extractor.iit.nrc.ca bibliographies context sensitive.html 2.4 Context in Natural Language It has been ....

Turney, P.D., The identification of context-sensitive features: A formal definition of context for concept learning. in ICML-96 Workshop on Learning in Context-Sensitive Domains, (Bari, Italy, 1996), 53-59.

....the contextual parameters. Turney [4] provides a formal definition of contextual features. This applies to learning classification rules, which is not the task that we are handling in this paper. Features are divided into primary, contextual, and relevant features. In an accompanying paper Turney [5] discusses five methods for managing the context. These methods are: normalization, expansion, classifier selection, classification adjustment, and weighting. According to this classification, our work belongs in the first and the last category. 3 The normalization approach The normalization ....

....domain, for instance, the exhaust gas temperature of the engine is a performance attribute since it is related to the quantity of energy lost, a fundamental performance measure. Note that formal definitions proposed by other researchers to categorize attributes (such as those proposed by Turney [5] and John et al. 1] are not applicable here since those required the use of a class attribute. The normalization approach is composed of two steps: the contextual analysis and the normalization. In the contextual analysis, we model the effects of each contextual attribute on the performance ....

Turney, P: The Identification of Context-sensitive Features: A Formal Definition of Context for Concept Learning. Procs. of the Workshop on Leaning in Contextsensitive Domains, 13th International Conf. on Machine Learning (1996) 53-59.

.... information can be discovered from this scientific domain [Fayyad et al. 1996a] However, the complexity of this application (diverse forms of data, time series relationships, high dimentionality, imbalance number of positive and negative examples [Kubat and Matwin 1997] presence of contexts [Turney 1996]) makes development of an appropriate knowledge discovery strategy difficult. In this paper, we discuss the specific issues to consider during the analysis of commercial aircraft data. We further introduce a knowledge discovery in databases (KDD) approach that we are developing to discover hidden ....

....should be normalized. The difficulty comes from the fact that we generally do not know what the required transformations are. Inferring appropriate normalization formulas from the data represent a difficult challenge that has been recently addressed by other researchers [Katz et al. 1990; Turney 1996]. Given the above characteristics, our overall goal is to develop a methodology that can be used to either explain component failures performance deviations or help in predicting the occurrence of future problems. We define component failures as problems in which a particular component or ....

[Article contains additional citation context not shown here]

Turney, P. (1996). The Identification of Context-Sensitive Features: A Formal Definition of Context for Concept Learning. Proceedings of the Workshop on Learning in Context-Sensitive Domains, at the 13th International Conference on Machine Learning. 53-59. Bari, Italy.

....in another context may fundamentally make it difficult to reuse the case. Related to the question of the context of a feature is the question of its relevance. 1] pointed out that the relevance of the feature is a context specific property. Traditional approaches for classification tasks, e.g. [14], have defined the relevance and context of a feature in terms of statistical information such as the distribution of their values relative to the values of other features. In contrast to classification tasks, in synthesis tasks such as planning, other elements affect the context and relevance of ....

P. D. Turney. The identification of context-sensitive features: A formal definition of context for concept learning. In Proceedings of the ECML-96 Workshop on Learning in Context-Sensitive Domains, 1996.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC