Learning in imbalanced relational data

Traditional learning techniques learn from flat data files with the assumption that each class has a similar number of examples. However, the majority of real-world data are stored as relational systems with imbalanced data distribution, where one class of data is over-represented as compared with other classes. We propose to extend a relational learning technique called Probabilistic Relational Models (PRMs) to deal with the imbalanced class problem. We address learning from imbalanced relational data using an ensemble of PRMs and propose a new model: the PRMs-IM. We show the performance of PRMs-IM on a real university relational database to identify students at risk.

Modelling relational aspects of critical infrastructure systems

The relational aspects for critical infrastructure systems are not readily quantifiable as there are numerous variability’s and system dynamics that lack uniformity and are difficult to quantify. Notwithstanding this, there is a large body of existing research that is founded in the area of quantitative analysis of critical infrastructure networks, their system relationships and the resilience of these networks. However, the focus of this research is to investigate the aspect of taking a different, more generalised and holistic system perspective approach. This is to suggest that that through applying network theory and taking a ‘soft’ system-like modelling approach that this offers an alternative approach to viewing and modelling critical infrastructure system relational aspects that warrants further enquiry.

Elaborative relational structures in the research prose of sociology : an intercultural study

This paper investigates elaborative relational structures utilised by native English speaking and native Polish speaking scholars in sociology research articles written in English. The examined texts have been produced in American, Australian and Polish academic discourse communities. The study utilised the framework of the analysis of the rhetorical structure of tests (FARS) as an analytical tool (Golebiowski 2009, 2011). The following types of elaboration relations are discussed : amplification, extension, reformulation, explanation, instantiation and addition. Elaboration is analysed with respect to its textual function, frequency of employment, hierarchical location, recursiveness, discoursal prominence and explicitness. The elaborative systems in the examined texts are shown to be complex, with pervasive presence of multi-stage recursive structures. It is suggested that elaborativeness may be a general characteristic of the style of writing sociology, which, as a relatively new discipline, requires establishing of wide grounds for the proposed claims, where writers persuade their readers not only of the specific claims of their text, but also of frameworks of thought in which the claims are placed. It is hypothesized that the similarities in the elaborativeness across texts result from the shared stylistic conventions and traditions of the disciplinary research community of sociology, while differences in the mode of employment of elaboration relations are attributed to cultural norms and conventions as well as educational systems prevailing within the discourse communities constituting the social contexts of the studied texts.
Golebiowski, Z. (2011). Scholarly criticism across discourse communities. In Salager-Meyer, Françoise and Lewin, Beverly A. (eds), Crossed words : Criticism in scholarly writing, pp. 203-224, Peter Lang International Academic Publishers, Berlin, Germany.
Golebiowski Z. (2009). The use of contrastive strategies in a sociology research paper: A cross-cultural study. In Suomela-Salmi, Eija and Dervin, Fred (eds), Cross-linguistic and cross-cultural perspectives on academic discourse, pp. 165-186, John Benjamins Publishing Company, Philadelphia.

Acquisition of domain ontology from relational schemes based on reverse engineering

Acquisition of domain ontology from database has been of catholic concern. This paper, taking relational schemes as example, analyzes how to identify the information about the structure of relational schemes in legacy systems. Then, it presents twelve extraction rules, which facilitate the obtaining of terms and relations from the relational schemes. Finally, it uses the EER diagram to further obtain semantic information from relational schemes for refining ontology model. The development method of domain ontology based on reverse engineering is a supplement to forward engineering. The union of the two development methods is certainly beneficial for the designers of domain ontology. © 2009 IEEE.

Collaborating differently on different topics: a multi-relational approach to multi-task learning

Multi-task learning offers a way to benefit from synergy of multiple related prediction tasks via their joint modeling. Current multi-task techniques model related tasks jointly, assuming that the tasks share the same relationship across features uniformly. This assumption is seldom true as tasks may be related across some features but not others. Addressing this problem, we propose a new multi-task learning model that learns separate task relationships along different features. This added flexibility allows our model to have a finer and differential level of control in joint modeling of tasks along different features. We formulate the model as an optimization problem and provide an efficient, iterative solution. We illustrate the behavior of the proposed model using a synthetic dataset where we induce varied feature-dependent task relationships: positive relationship, negative relationship, no relationship. Using four real datasets, we evaluate the effectiveness of the proposed model for many multi-task regression and classification problems, and demonstrate its superiority over other state-of-the-art multi-task learning models

Mitigating malicious feedback attacks in trust management systems

In electronic commerce (e-commerce) environment, trust management has been identified as vital component for establishing and maintaining successful relational exchanges between the trading partners. As trust management systems depend on the feedbacks provided by the trading partners, they are fallible to strategic manipulation of the rating attacks. Therefore, in order to improve the reliability of the trust management systems, an approach that addresses feedback-related vulnerabilities is paramount. This paper proposes an approach for identifying and actioning of falsified feedbacks to make trust management systems robust against rating manipulation attacks. The viability of the proposed approach is studied experimentally and the results of various simulation experiments show that the proposed approach can be highly effective in identifying falsified feedbacks.