In many machine learning domains, there is a large supply of unlabeled data but limited labeled data, which can be expensive to generate. Consequently, semi-supervised learning, learning from a combination of both labeled and unlabeled data, has become a topic of significant recent interest. Our research focus is on semi-supervised clustering, which uses a small amount of supervised data in the...

A number of approaches have been developed for combining wikis with semantic technologies. Many semantic wikis focus on enabling users to specify properties and relationships of individual elements. Complex schema information is typically not edited by the wiki user. Nevertheless, semantic wikis could benefit from taking existing schema information into account, and to allow users to specify ad...

One of the current bottlenecks for automating ontology evolution is resolving the right links between newly arising information and the existing knowledge in the ontology. Most of existing approaches mainly rely on the user when it comes to capturing and representing new knowledge. Our ontology evolution framework intends to reduce or even eliminate user input through the use of background know...

Prior work in automated scientific discovery has been successful in finding patterns in data, given that a reasonably small set of mostly relevant features is specified. The work described in this paper places data in the context of large bodies of background knowledge. Specifically, data items are connected to multiple databases of background knowledge represented as inheritance networks. The ...

It is a well-known fact that propositional learning algorithms require \good" features to perform well in practice. So a major step in data engineering for inductive learning is the construction of good features by domain experts. These features often represent properties of structured objects, where a property typically is the occurrence of a certain substruc-ture having certain properties. To...

In this contribution we present a method for constraining the learning of a Multi-Layer Perceptron network with background knowledge. The algorithms presented here can be used to train the partial derivatives of the network to match given numerical values or to minimize a given cost function. Thus the mapping produced by the network can be constrained according to known input-output models, mon...

Text document clustering plays an important role in providing intuitive navigation and browsing mechanisms by organizing large amounts of information into a small number of meaningful clusters. Standard partitional or agglomerative clustering methods efficiently compute results to this end. However, the bag of words representation used for these clustering methods is often unsatisfactory as it ...

The development of ontologies for various purposes is now a relatively commonplace process. A number of different approaches towards this aim are evident; empirical methodologies, giving rise to data-driven procedures or self-reflective (innate) methodologies, resulting in artifacts that are based on intellectual background understanding. In this paper, we compare and contrast these approaches ...

Texts are replete with gaps, information omitted since authors assume a certain amount of background knowledge. We describe the kind of information (the formalism and methods to derive the content) useful for automated filling of such gaps. We describe a stepwise procedure with a detailed example.

In this paper we present a method to enrich the hypothesis language which is used to construct the bottom clause. This approach, which is embedded in the Aleph system, is based on numerical fixed subintervals and categorical subsets. Each subinterval/subset contains values, for each attribute which is predefined, that are related with the example to saturate. The enriched language allows to red...