class evaluation that classifies each statement as true, viewpoint or erroneous; and 2) 2-class evaluation that distinguishes between the true statements and all the others (i.e. viewpoint and erroneous statements were considered as one category). Interestingly, as shown in Table 2 the obtained results for the baselines are comparable for the two experiments, while for all the aggregation measures the second experiment's results were consistently higher (by up to 17%) than those of the first experiment. Therefore, we conclude that workers have quite a good ability to objectively assess the others' opinions, while their own opinions seem less reliable and consequently yield lower accuracy in classification. The accuracy of the individual worker judgment baseline is quite low for both experiments (0.7 and 0.72). Approximately 30,000 individual worker judgments were produced in each of the crowdsourcing experiments. Thus, every worker in isolation does not do any better than the "all true" baseline strategy (0.73), as could be expected. However, the workers' collective decisions (after aggregation) for each statement were much more accurate. We observe that the aggregation measure has a crucial influence on the results: a better aggregation measure can increase the accuracy by over 25% compared to the baselines. The best results were obtained by the Bayesian inference measure with alpha=0.5 and beta=0.5 for Jeffrey's prior. The AUC values are presented in Table 2 and the ROC curves are shown in Figure 3. This measure elicited 0.92 accuracy (by definition this measure could only be applied for the 2-class classification). CONCLUSION The main contribution of this research is that we show that crowdsourcing workers can quite accurately assess statements in a multi-viewpoint ontology to distinguish between true, viewpoint and erroneous statements for a given professional domain, and especially to differentiate true statements from the others. In addition, we found that a h

In order to support innovative scholarship, digital library development must be informed by an understanding of new modes of scholarly production. This paper considers an emergent genre of scholarly publication in the humanities: the digital thematic research collection. While thematic research collections have grown in number and academic significance over the past decade, and this is reflected in the literature on scholarly practice, we do not have a definitive grasp on this exciting phenomenon. What significant features distinguish thematic research collections as a mode of scholarly production? This paper describes a pilot study conducted as part of a project investigating the significant features of thematic research collections. A stronger understanding of thematic research collections will contribute to evolving scholarly evaluation processes for tenure and promotion and the advancement of digital libraries to support new modes of research. [ABSTRACT FROM AUTHOR]