PRINS: scalable model inference for component-based system logs

Behavioral software models play a key role in many software engineering tasks; unfortunately, these models either are not available during software development or, if available, quickly become outdated as implementations evolve. Model inference techniques have been proposed as a viable solution to extract finite state models from execution logs. However, existing techniques do not scale well when processing very large logs that can be commonly found in practice. In this paper, we address the scalability problem of inferring the model of a component-based system from large system logs, without requiring any extra information. Our model inference technique, called PRINS, follows a divide-and-conquer approach. The idea is to first infer a model of each system component from the corresponding logs; then, the individual component models are merged together taking into account the flow of events across components, as reflected in the logs. We evaluated PRINS in terms of scalability and accuracy, using nine datasets composed of logs extracted from publicly available benchmarks and a personal computer running desktop business applications. The results show that PRINS can process large logs much faster than a publicly available and well-known state-of-the-art tool, without significantly compromising the accuracy of inferred models.

Tree decay modulates the functional response of lichen communities in Patagonian temperate forests

Epiphytic and epixylic lichens respond negatively to forest degradation, climate change and pollution, but those effects may depend on functional traits or interact with the stage of tree decay. Disentangling the main drivers of lichen communities remains a challenge in regions where lichens are diverse and poorly known, as the case of Patagonian temperate forests. We used a multi-scale approach to evaluate the relationship between environmental variables, tree decay stage and lichens. We sampled lichens across three increasing scales (tree ≪ site ≪ landscape) by selecting 19 landscape units, where trees in four decay stages (snags, logs, cavity trees and healthy trees) were selected within sampling plots. A total of 35 predictors were measured over different scales, including 25 remote sensing indices of forest conditions, climate and air pollutants. Structural Equation Models were used to test the causal linkages of predictors with lichens, distinguishing functional categories (size, growth and reproductive strategy). A total of 69 lichen species were recorded. Cavity trees and logs supported the largest diversity, while snags and healthy trees had the lowest diversity. Functional lichen groups responded differently to fine-scale variables, including the diameter, height, density and pH of trees. Air pollutants affected species with sexual and mixed strategies. Lichens were sensitive to precipitation, temperature and wind speed, with foliose and sexual species responding positively to the latter. The abundance of all species and macrolichens increased with tree senescence and decreased with canopy continuity. Lichens occupying snags and logs responded negatively to primary productivity and tree senescence, but positively to soil organic matter. Our findings suggest: i) the functional structure of lichen communities varies non-linearly with the wood decay process; ii) the reproductive strategy influences the sensitivity to air pollutants, iii) climate variables influence dispersal and colonization of woody substrates; and iv) forest structure/succession interacts with tree decay.

Feedback learning opportunities from medical student logs of paediatric patients

BACKGROUND

Feedback can alter medical student logging practices, although most learners feel feedback is inadequate. A varied case mix in rural and urban contexts offers diverse clinical encounters. Logs are an indicator of these clinical experiences, and contain opportunities for feedback, which can greatly influence learning: we labelled these 'feedback learning opportunities' (FLOs). We asked: How often do FLOs occur? What are the case complexities of rural compared to urban paediatric logs? Do more complex cases result in more FLOs?

METHODS

In Western Australia, 25% of medical students are dispersed in a Rural Clinical School (RCSWA) up to 2175 miles (3500 km) from the city. Urban students logged 20 written cases; rural students logged a minimum of 25 paediatric cases electronically. These were reviewed to identify FLOs, using a coding convention. FLO categories provided a structure for feedback: medical, professionalism, insufficient, clinical reasoning, student wellbeing, quality and safety, and sociocultural. Each log was assigned an overall primary, secondary or tertiary case complexity.

RESULTS

There were 76 consenting students in each urban and rural group, providing 3034 logs for analysis after exclusions. FLOs occurred in more than half the logs, with significantly more rural (OR 1.35 95% CI 1.17, 1.56; p < 0.0001). Major FLOs occurred in over a third of logs, but with no significant difference between rural and urban (OR 1.10 95% CI 0.94, 1.28; p = 0.24). Medical FLOs were the most common, accounting for 64.0% of rural and 75.2% of urban FLOs (OR 1.71 95% CI 1.37, 2.12; p < 0.0001). Students logged cases with a variety of complexities. Most cases logged by urban students in a tertiary healthcare setting were of primary and secondary complexity. Major medical FLOs increased with increasing patient complexity, occurring in 32.1% of tertiary complexity cases logged by urban students (p < 0.001).

CONCLUSIONS

Case logs are a valuable resource for medical educators to enhance students' learning by providing meaningful feedback. FLOs occurred often, particularly in paediatric cases with multiple medical problems. This study strengthens recommendations for regular review and timely feedback on student logs. We recommend the FLOs categories as a framework for medical educators to identify FLOs.