Harries M., Sammut C., Horn K., Extracting hidden context, Machine Learning, 32(2), 1998, 101-126.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....have been developed that employ Machine Learning methods in real life applications. They learn real life concepts that tend to change over time [8, 14, 15] An illustrative example comes from Text Mining when learning shifting human interests [3] The concept drift, whether abrupt or gradual [5, 6], occurs over time. The evidences for changes in a concept are represented by the training examples, which are distributed over time. Hence the old observation can become irrelevant to the current time period and thus the learned knowledge can be outdated. Several methods have been suggested to ....

....are distributed over time. Hence the old observation can become irrelevant to the current time period and thus the learned knowledge can be outdated. Several methods have been suggested to cope with this problem, either to forget outdated induced knowledge, or to forget outdated training examples [3, 5, 9, 15]. In the following sections we will focus on forgetting training examples (learning with partial memory) since this approach is more general and does not require significant changes in training algorithms. 2.2 Partial memory learning Partial memory learners are systems that select and maintain ....

M. B. Harries, C. Sammut, and K. Horn. Extracting hidden context. Machine Learning, 32:101--126, 1998.

....using only one (too little knowledge) or more than two commands (over tting) It also showed that C4.5 [7] a decision tree learner, performs better than other widely used batch learning algorithms on this task. To test if time is a relevant feature we use the approach taken in the Splice system [1]: if a decision tree learner uses the attribute in building a decision tree, it is a relevant feature. We use J48, a variant of C4.5 in Weka [9] a GNU data mining tool. The Greenberg dataset consists of data from 156 users with one data set per user, divided in four groups: novice users (54) ....

Michael Bonnell Harries, Claude Sammut, and Kim Horn. Extracting hidden context. Machine Learning, 32(2):101-126, 1998.

....applies a metalearning process to a basic incremental learning neural net; the meta algorithm adjusts the window over which the basic learning process works. Here it is an assumption that contexts are contiguous in time and so a time window is a sufficient representation of context. Harries et al. [19] employ a batch learner as a meta algorithm to identify stable contexts and their concepts; this makes the assumption that the contexts are contiguous in the environmental variables and can only be done off line. Aha describes an incremental instance basedlearning which uses a clustering ....

Harries, M.B., Sammut, C. and Horn, K. Extracting Hidden Contexts. Machine Learning, 32. 101-112.

....learning approaches (e.g. time window) mentioned above. Another approach to learning and tracking changing concepts employs two level learning algorithms to adjust to changing contexts by trying to detect (via metalearning) contextual clues and using this information to focus the learning process [5] and [17] The approach presented in this paper also employs a meta learning level, however it doesn t assume that the attributes represent current context explicitly. It assumes that the recent observations are able to provide information about current context. The recent relevant observations ....

....themes. User profiles must also adapt to changing interests of the users over time. This research shows that user s interests can be tracked over time by measuring the similarity of interests in a time period. An offline meta learning algorithm for identification of hidden context is presented in [5]. The approach assumes that concepts are likely to be stable for some period of time. It uses batch learning and contextual clustering to detect stable concepts and to extract hidden context. An intelligent agent called NewsDude that is able to adapt to changing user interests is presented in ....

Harries, M. and Sammut, C.: Extracting Hidden Context. Machine Learning 32, Kluwer Academic Publishers (1998), 101-126.

....changes in a concept are represented by the training examples, which are distributed over time. Hence the old observation can become irrelevant to the current period thus the learned knowledge can be out ofdate. The systems use different forgetting mechanisms to cope with this problem (e.g. 2] [3], 5] 8] 13] 14] etc. Usually, these methods forget abruptly. That means the examples that are irrelevant according to some time criteria (e.g. examples that are outdated) are deleted from the partial memory [6] Hence, these instances are totally forgotten. The examples that remain in the ....

....long term model. The purpose of the long term user model is to model the user s general preferences for news stories and compute predictions for stories that could not be classified by the short term model. An offline meta learning algorithm for identification of hidden context is presented in [3]. The approach assumes that concepts are likely to be stable for some period of time. It uses batch learning and contextual clustering to detect stable concepts and to extract hidden context. In [5] and [6] a method for selecting training examples for a partial memory learning system is ....

[Article contains additional citation context not shown here]

Harries M. B., Sammut C., Horn K. (1998). Extracting Hidden Context, Machine Learning 32, 101-126, Kluwer Academic Publishers.

.... (e.g. Dietterich and Michalski s work with Eleusis [DM86] work with temporal logics and their applications to recognising events, for example Kumar s work on temporal event conceptualisation [KM87] and the recent work in context detection and extraction for machine learning applications [Wid96, HHS98] While some of these areas bear some interesting relations to temporal classification, they differ in several regards. Sequence prediction is not about learning from labelled examples, nor, typically does it deal with multivariate data. The event conceptualisation work focuses on recognition of ....

Michael Harries, Kim Horn, and Claude Sammut. Extracting hidden context. Machine Learning, 32(2), August 1998.

.... prediction (e.g. Dietterich and Michalski s work with Eleusis [DM86] temporal logics and their applications to recognising events, for example Kumar s work on temporal event conceptualisation [KM87] and the recent work in context detection and extraction for machine learning applications [Wid96, HHS98] While some of these areas bear interesting relations to temporal classification, they di#er in several regards. Sequence prediction is not about learning from labelled examples. The event conceptualisation work focuses on recognition of temporal events, but not learning the events themselves. ....

Michael Harries, Kim Horn, and Claude Sammut. Extracting hidden context. Machine Learning, 32(2), August 1998.

....on the part of the data analyst. One of the promises of SPLICE 2 is tO reduce the level of additional work required of the analyst in such domains. We also find that a state of the art conceptual clustering method does not identify the hidden context. i INTRODUCTION 3 1 Introduction SPLICE 2 [2] is a meta learner designed for domains with hidden changes in context. SPLICE S job is to detect hidden contexts that are persistent over time l, this includes, in particular, the detection of contexts that recur after alternation with other contexts. SPLICE applies a state of the art ....

....window. A variety of training window sizes, k, were tested, ranging from one week (336 examples) to ten weeks (3360 examples) A variable sized training window including all past examples was also tested. The test window, r, was kept at one week (336 examples) for this experiment. SPLICE 2 [2] used C4.5 as the underlying learner (run with default parameters plus subsetting) A window size of 2000 items was used in order to focus SPLICE 2 on large scale contextual effects. The contextual clustering process was begun by randomly selecting a single instant of time upon which to split the ....

Michael B. Harries, Claude Sammut, and Kim Horn. Extracting hidden context. Machine Learning, 32:101-126, 1998.

No context found.

Harries M., Sammut C., Horn K., Extracting hidden context, Machine Learning, 32(2), 1998, 101-126.

No context found.

M.B. Harries, C. Sammut, and K. Horn, "Extracting hidden context," Machine Learning, vol. 32, pp. 101--126, 1998.

No context found.

Harries, M. and Sammut, C. 1998. Extracting Hidden Context. Machine Learning 32: 101-126.

No context found.

Harries M. B., Sammut C., Horn K. (1998). Extracting Hidden Context, Machine Learning 32, 101126.

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