Sirma Group (Bulgaria)

Abstract The use of various alternative sources is a critical component of environmental protection. This article presents a theoretical and experimental study on the possibilities for using the kinetic energy of rivers. A theoretical calculation has been performed on several major rivers in Bulgaria, along with an experimental study in laboratory conditions of 12 different types of hydrokinetic turbines at three distinct flow velocities—0.1, 0.2, and 0.3 m/s. An analysis of the results has been conducted.

The digitalization of clinical workflows and the increasing performance of deep learning algorithms are paving the way towards new methods for tackling cancer diagnosis. However, the availability of medical specialists to annotate digitized images and free-text diagnostic reports does not scale with the need for large datasets required to train robust computer-aided diagnosis methods that can target the high variability of clinical cases and data produced. This work proposes and evaluates an approach to eliminate the need for manual annotations to train computer-aided diagnosis tools in digital pathology. The approach includes two components, to automatically extract semantically meaningful concepts from diagnostic reports and use them as weak labels to train convolutional neural networks (CNNs) for histopathology diagnosis. The approach is trained (through 10-fold cross-validation) on 3’769 clinical images and reports, provided by two hospitals and tested on over 11’000 images from private and publicly available datasets. The CNN, trained with automatically generated labels, is compared with the same architecture trained with manual labels. Results show that combining text analysis and end-to-end deep neural networks allows building computer-aided diagnosis tools that reach solid performance (micro-accuracy = 0.908 at image-level) based only on existing clinical data without the need for manual annotations.

n/a

This chapter focuses on semantic searching at web scale. The solution presented takes advantage of the specific strengths of semantic repositories and the raw power of relational databases, the latter having been developed over decades and capable of handling efficiently large volumes of data with fixed structure (which is the case with the occurrence statistics) and the former allowing for inference and querying on top of formal knowledge. The interactive faceted search capability described is a demonstration how an approach based on these two technologies is more powerful and efficient for certain tasks as compared to traditional search engines. The contributions of the chapter can be summarized as follows: ● novel indexing schema for semantic search, based on entity occurrence; ● description of a scalable implementation of such indexing; ● advanced faceted search interface, based on co-occurrence.

This article presents the evidence base for 'operator models' - that is, how to deliver a sustainable service through the interaction of the 'client', 'revenue collector' and 'operator' functions - for municipal solid waste management in emerging and developing countries. The companion article addresses a selection of locally appropriate operator models. The evidence shows that no 'standard' operator model is effective in all developing countries and circumstances. Each city uses a mix of different operator models; 134 cases showed on average 2.5 models per city, each applying to different elements of municipal solid waste management - that is, street sweeping, primary collection, secondary collection, transfer, recycling, resource recovery and disposal or a combination. Operator models were analysed in detail for 28 case studies; the article summarises evidence across all elements and in more detail for waste collection. Operators fall into three main groups: The public sector, formal private sector, and micro-service providers including micro-, community-based and informal enterprises. Micro-service providers emerge as a common group; they are effective in expanding primary collection service coverage into poor- or peri-urban neighbourhoods and in delivering recycling. Both public and private sector operators can deliver effective services in the appropriate situation; what matters more is a strong client organisation responsible for municipal solid waste management within the municipality, with stable political and financial backing and capacity to manage service delivery. Revenue collection is also integral to operator models: Generally the municipality pays the operator from direct charges and/or indirect taxes, rather than the operator collecting fees directly from the service user.

Abstract The paper describes a methodology for bootstrapping relation extraction from unstructured text in the context of GATE, but also applied to the KIM semantic annotation platform. The focus is on identifying a set of relations between entities previously found by named entity recognizer. The methodology is developed and applied to three kinds of relations and evaluated both with the ANNIE system and the default information extraction module of KIM. The methodology covers the problem of identifying the task, the target domain, the development of training and testing corpora, and useful lexical resources, the choice of a particular relation extraction approach. The application of information extraction for the Semantic Web also brings a new interesting dimension of not merely recognizing the entity type, but going into instantiation of entity references and linking them to an entity instance in a semantic repository.KeywordsRelation Information ExtractionSemantic WebMethodology

A separation to the baseline isocratic technique has been developed for the evaluation of the purity of aprotinin by internal normalization. The column used was a 30 nm pore diameter butyl-bonded silica stationary phase. An exact relationship was found to give an adequate description of the retention of aprotinin using acetonitrile concentrations of 17-22% and NaClO4 concentrations of 5-50 mmol. It can be used to optimize the mobile phase content for the particular user column and to compensate for possible variations in the retention time of aprotinin, analysed on different batches of butyl-bonded stationary phases.