Exploiting the enormous amount of open source software (OSS) as a vehicle for reuse is a promising opportunity for software engineers. However, this task is far from trivial, since such projects are sometimes not easy to understand and adapt to target systems, whereas at the same time the reusable assets are not obvious to identify. In this study, we assess open source software projects, with r...

This paper argues that the sorting of metaphorical expressions according to source domains can be verified through using the WordNet lexical knowledgebase and SUMO ontology. This study uses the examples from RECESSION and demonstrates that expressions such as ‘painful recession’ and ‘the exacerbated recovery’ are related to the decreasing physical state of organism (i.e., ECONOMY IS A DISEASE)....

Lower, upper, sandwich, mixed, and convex power domains are isomorphic to domains of second order predicates mapping predicates on the ground domain to logical values in a semiring. The various power domains diier in the nature of the underlying semiring logic and in logical constraints on the second order predicates.

1. Aims of paper The general aim of this paper is to present the implications of a corpus-based study of metaphors for communication in British English for the extrapolation of metaphorical ‘source’ domains from linguistic data. Our specific aims are: 1) to provide more evidence for the relevance and centrality of the notion of ‘scene’ (or ‘scenario’) in the analysis of metaphorical patterns, a...

Domain adaptation algorithms aim at handling the shift between source and target domains. A classifier is trained on images from the source domain; and the classifier recognizes objects in images from the target domain. In this paper, we present a joint subspace and dictionary learning framework for domain adaptation. Our approach simultaneously exploits the low-dimensional structures in two do...

A novel upper power domain construction is deened by means of strongly compact sets. Its power domains contain less elements than the classical ones in terms of compact sets, but still admit all necessary operations, i.e. they contain less junk. The notion of strong compactness allows a proof of stronger properties than compactness would, e.g. an intrinsic universal property of the upper power ...

Adversarial learning has been successfully embedded into deep networks to learn transferable features, which reduce distribution discrepancy between the source and target domains. Existing domain adversarial networks assume fully shared label space across domains. In the presence of big data, there is strong motivation of transferring both classification and representation models from existing ...