Improving reliability of unbalanced text mining by reducing performance bias

Class imbalance in textual data is one important factor that affects the reliability of text mining. For imbalanced textual data, conventional classifiers tend to have a strong performance bias, which results in high accuracy rate on the majority class but very low rate on the minorities. An extreme strategy for unbalanced learning is to discard the majority instances and apply one-class classification to the minority class. However, this could easily cause another type of bias, which increases the accuracy rate on minorities by sacrificing the majorities. This chapter aims to investigate approaches that reduce these two types of performance bias and improve the reliability of discovered classification rules. Experimental results show that the inexact field learning method and parameter optimized one class classifiers achieve more balanced performance than the standard approaches.

Unis accused of 'elite bias' in enrolments

Australia’s elite universities are biased against taking disadvantaged students for fear it will lower their prestigious image, an Adelaide academic says.

Anger at schools' Christian 'bias'

Buddhist community leader Dr Sue Smith has complained of the ''Christian bias'' in religious education in Victoria, saying if her group had access to government funding, they too could expand to hundreds of schools.

Are most samples of animals systematically biased? Consistent individual trait differences bias samples despite random sampling

Sampling animals from the wild for study is something nearly every biologist has done, but despite our best efforts to obtain random samples of animals, ‘hidden’ trait biases may still exist. For example, consistent behavioral traits can affect trappability/catchability, independent of obvious factors such as size and gender, and these traits are often correlated with other repeatable physiological and/or life history traits. If so, systematic sampling bias may exist for any of these traits. The extent to which this is a problem, of course, depends on the magnitude of bias, which is presently unknown because the underlying trait distributions in populations are usually unknown, or unknowable. Indeed, our present knowledge about sampling bias comes from samples (not complete population censuses), which can possess bias to begin with. I had the unique opportunity to create naturalized populations of fish by seeding each of four small fishless lakes with equal densities of slow-, intermediate-, and fast-growing fish. Using sampling methods that are not size-selective, I observed that fast-growing fish were up to two-times more likely to be sampled than slower-growing fish. This indicates substantial and systematic bias with respect to an important life history trait (growth rate). If correlations between behavioral, physiological and life-history traits are as widespread as the literature suggests, then many animal samples may be systematically biased with respect to these traits (e.g., when collecting animals for laboratory use), and affect our inferences about population structure and abundance. I conclude with a discussion on ways to minimize sampling bias for particular physiological/behavioral/life-history types within animal populations.

Estimating non-response bias in a web-based survey of technology acceptance : a case study of unit guide information systems

Surveys are mostly challenged by response rates. Among the various types of survey research, web-based (internet-based/electronic/online) surveys are commonly used for data collection for a geographically diverse population. In surveys with high/low response rates, non-response bias can be a major concern. While it is not always possible to measure the actual bias due to non-response there are different approaches and techniques that help to identify reasons of non-response bias. The aims of this paper are twofold. (1) To provide an appropriate, interesting and important non-response bias case study for future web-based surveys that will provide guidance to other Information Systems researchers. The case-study concerns an online-survey to evaluate a technology acceptance model for Unit Guide Information systems (UGIS). (2) To discuss how nonresponse bias in a web-based technology acceptance study of an information system (UGIS in this case) can be contained and managed.

The relationship between attachment style, anxiety sensitivity and interpretive bias among adolescent nonclinical panickers

Elevated anxiety sensitivity and the tendency to catastrophically misinterpret ambiguous bodily sensations has been demonstrated in people who experience nonclinical levels of panic (Richards, Austin, & Alvarenga, 2001), and anxiety sensitivity has been shown to be associated with insecure attachment in adolescents and young adults (Weems, Berman, Silverman, and Saavedra, 2001). This study investigated the relationship between attachment style, anxiety sensitivity and catastrophic misinterpretation among 11 nonclinical panickers and 58 nonanxious controls aged 18 to 19 years. Participants completed the Brief Bodily Sensations Interpretation Questionnaire (BBSIQ), Anxiety Sensitivity Index (ASI) and an attachment questionnaire. The hypothesis that insecurely attached individuals would demonstrate greater catastrophic misinterpretation and higher anxiety sensitivity than securely attached individuals was not supported; however, nonclinical panickers gave more anxiety-related interpretations of ambiguous internal stimuli than nonanxious controls. Results do not support the notion that attachment style is related to anxiety sensitivity or catastrophic misinterpretation (regardless of panic experience). Results do, however, support the notion that anxiety-related misinterpretation of ambiguous somatic sensations precedes the onset of panic disorder.

Dissemination of therapist-assisted internet cognitive behaviour therapy : development and open pilot study of a workshop

There is considerable research suggesting that therapist-assisted Internet cognitive behaviour therapy (ICBT) is efficacious in the treatment of depression and anxiety. Given this research, there is a growing interest in training students in therapist-assisted ICBT in order to assist with the dissemination of this emerging modality into routine clinical practice. In this study, we developed, delivered, and evaluated a therapist-assisted ICBT workshop for clinical psychology graduate students (n = 20). The workshop provided both research evidence and practical information related to the delivery of therapist-assisted ICBT. The workshop also incorporated an experiential component with students working on and discussing responses to client e-mails. Before and after the workshop, we measured knowledge of therapist-assisted ICBT research and professional practice issues, as well as attitudes towards and confidence in delivering therapist-assisted ICBT. Statistically significant changes were observed in all areas. Eighty-five per cent of students are now offering therapist-assisted ICBT under supervision. We conclude by discussing future research directions related to disseminating therapist-assisted ICBT.

Sex does not matter: gender bias and gender differences in peer assessments of contributions to group work

This paper considers the possibility of gender bias in peer ratings for contributions to team assignments, as measured by an online self-and-peer assessment tool. The research was conducted to determine whether peer assessment led to reliable and fair marking outcomes. The methodology of Falchikov and Magin was followed in order to test their finding that gender has no discernable impact on peer ratings. Data from over 1500 participants at two universities enrolled in four different degree programmes were analysed. The research indicates an absence of gender bias in six case studies. The research also found that women received significantly higher ratings than men.