3D or not 3D? Evaluation of the effectiveness of 3D-enhanced warning messages for communication in crisis situations

Grzegorz Banerski, Katarzyna Abramczuk, Cezary Biele

2020 SAFETY SCIENCE, T. 132, s. 104961

A common problem in crisis situations is the unwillingness of the local communities to undertake individual safety measures. This study examined whether the motivation of the local population can be increased if 3D animation are used for visualization of the potential disaster scenarios. Furthermore, the study aimed at finding out how the use of such material affects the ability of the audience to remember what preventive actions should be taken.
As part of the experiment, three versions of a video message about flood threat were prepared: a video limited to the presenter with no additional elements in the background, a classic TV-like video message format, and a video featuring 3D visualisations of the consequences of flooding. The videos were presented to groups of subjects who live in flood-prone areas. Protection Motivation Theory was used as the experiment’s framework.
We found that the type of warning message did not influence the level of acquired knowledge of protective actions. Important differences were found in the structure of the determinants of self-protective motivation across the groups. The group that watched the 3D enhanced video was the only one in which the more emotional cognitive path of threat appraisal had a significant impact on the level of motivation. As regards the level of motivation to take protective actions, the observed effects were more complex than expected. The paper discusses the potential impact of the results on efficient risk communication design and the usefulness of the use of 3D animation in warning messages.


A Bidirectional Iterative Algorithm for Nested Named Entity Recognition

Sławomir Dadas, Jarosław Protasiewicz

2020 IEEE Access, T. 8, s. 135091 - 135102

Nested named entity recognition (NER) is a special case of structured prediction in which annotated sequences can be contained inside each other. It is a challenging and significant problem in natural language processing. In this paper, we propose a novel framework for nested named entity recognition tasks. Our approach is based on a deep learning model which can be called in an iterative way, expanding the set of predicted entity mentions with each subsequent iteration. The proposed framework combines two such models trained to identify named entities in different directions: from general to specific ( outside-in ), and from specific to general ( inside-out ). The predictions of both models are then aggregated by a selection policy. We propose and evaluate several selection policies which can be used with our algorithm. Our method does not impose any restrictions on the length of entity mentions, number of entity classes, depth, or structure of the predicted output. The framework has been validated experimentally on four well-known nested named entity recognition datasets: GENIA, NNE, PolEval, and GermEval. The datasets differ in terms of domain (biomedical, news, mixed), language (English, Polish, German), and the structure of nesting (simple, complex). Through extensive tests, we prove that the approach we have proposed outperforms existing methods for nested named entity recognition.


A Neural Network Toolbox for Electricity Consumption Forecasting

Jarosław Protasiewicz

2020 W: Prediction of Customer Status in Corporate Banking Using Neural Networks / Stanisław Osowski, Łukasz Siereński ; Piscataway: IEEE, s. 1-6

Prediction of Customer Status in Corporate Banking Using Neural Networks. Glasgow, 2020-07-19 - 2020-07-24

The aim of this study is to propose a neural toolbox for electricity consumption prediction. The toolkit covers the implementation of three artificial neural networks, namely: (i) a multi-layered perceptron network; (ii) a radial basis function network; and (iii) a self-organising map. Moreover, the toolbox includes tools known as metamodels, which allow easy use of these networks for forecasting. There are two prediction systems, namely: (i) serially connected models; and (ii) a two-levelled structure containing a classifier and a set of parallel models. They have been validated experimentally in the task of electricity consumption prediction. The results and flexibility of use suggest that the neural toolbox should help users to develop prediction systems of electricity consumption more conveniently, as it is designed for that particular purpose. The toolbox has been developed as open source – no commercial software is required to use it.


Academic entrepreneurship and the research productivity in Poland

Marta Bojko, Anna Knapińska, Aldona Tomczyńska

2020 Industry and Innovation, APR 2020, 1-21

Is it possible to reconcile academic entrepreneurship with the internationalisation of scientists’ productivity? We provide an answer to this question through the statistical analysis of a representative survey of 811 scientists in Poland. Based on Robert K. Merton’s distinction between two types of scholars: local (oriented towards the national science system) and cosmopolitan (aiming at international achievements), we found that research productivity positively influences academic entrepreneurship, but that scientists who are productive locally engage in academic entrepreneurship more often than the cosmopolitan ones. This suggests that the internationalisation of one’s scientific activity, at least in countries transitioning from a local to a cosmopolitan mode of research, is an absorbing endeavour that only the most productive researchers can reconcile with academic entrepreneurship. Such countries should balance their policies regarding the career development of their scientists also to include the promotion of science-business cooperation and not just the internationalisation of research.


Biologically Plausible Learning of Text Representation with Spiking Neural Networks

Marcin Białas, Marcin Mirończuk, Jacek Mańdziuk

2020 W: Parallel Problem Solving from Nature – PPSN XVI / Thomas Bäck, Mike Preuss, André Deutz, Hao Wang, Carola Doerr, Michael Emmerich, Heike Trautmann; Cham: Springer, s. 433-447

International Conference on Parallel Problem Solving from Nature. Leiden, 2020-09-05 - 2020-09-09

This study proposes a novel biologically plausible mechanism for generating low-dimensional spike-based text representation. First, we demonstrate how to transform documents into series of spikes (spike trains) which are subsequently used as input in the training process of a spiking neural network (SNN). The network is composed of biologically plausible elements, and trained according to the unsupervised Hebbian learning rule, Spike-Timing-Dependent Plasticity (STDP). After training, the SNN can be used to generate low-dimensional spike-based text representation suitable for text/document classification. Empirical results demonstrate that the generated text representation may be effectively used in text classification leading to an accuracy of 80.19% on the bydate version of the 20 newsgroups data set, which is a leading result amongst approaches that rely on low-dimensional text representations.


Dorobek piśmienniczy jednostek naukowych z grupy nauk ekonomicznych w świetle oceny parametrycznej z roku 2017

Jacek Drogosz

2020 Studia Prawno-Ekonomiczne, T. 115, s. 203-226

Background: The process of appraising scientific institutions provides the possibility to review and rank their scientific publications, and makes it possible to assess the current publications in the aspect of the next appraisal. The article analyzes the scientific publications of scientific insti-tutions classified as economic.

Research purpose: The purpose of the research is to indicate the most popular journals, publish-ers, and scientific conferences in the economic disciplines, defining their significance in the next appraisal in 2021. It makes it possible to draw preliminary conclusions regarding the implemen-tation of scientific policy in the field of economic sciences.

Methods: The material was developed on the basis of data from scientific institutions that were assessed in the parametric appraisal in 2017. Using analytical methods, information on the literary output of researchers from the economic disciplines was extracted. Data on journals, publishers, and conferences were ranked. Various data records by the reporting agents were taken into ac-count, which were unified for this work.

Conclusions: The analysis shows that alarge part of the most valuable literary output in the studied disciplines will lose its relevance in the next evaluation. There is aneed to adapt to new appraisal principles in order to increase the importance of scientific achievements. It was pointed out that the publications under study in the field of economics are not fully representative.


Dorobek piśmienniczy pracowników naukowych z dyscypliny bibliologia i informatologia w świetle oceny parametrycznej działalności naukowej polskich jednostek naukowych z roku 2017

Jacek Drogosz

2020 Przegląd Biblioteczny, T. 88, z. 1, s. 65-80

This article presents informations about scientific literature of researchers from the discipline of bibliology and information science from 2013-2016. The purpose/thesis – The purpose of the research is indicating the most popular journals, publishers and scientific conferences in the discipline defining their significance in the next evaluation process (2021). It allows drawing preliminary conclusions regarding the implementation of scientific policy in the field of bibliology and information science. The methods – The material was developed on the basis of data from scientific institutions that were assessed in the parameterization process in 2017. Using analytical methods, information on the literary output of researchers representing the discipline was extracted. Data on journals, publishers and conferences have been sorted by ranking. Various data records by the reporting agents were taken into account, which were unified for the purposes of this work. The results/conclusions – The analysis shows that a large part of the most valuable literary output in the studied discipline will lose its relevance in the next evaluation. Attention was drawn to the need to adapt to new evaluation principles in order to increase the importance of scientific achievements. It was pointed out that the representativeness of the researched publications in the scientific achievements of the studied discipline is not full.


Evaluation of Sentence Representations in Polish

Sławomir Dadas, Michał Perełkiewicz, Rafał Poświata

2020 W: TWELFTH INTERNATIONAL CONFERENCE ON LANGUAGE RESOURCES AND EVALUATION, May 11-16 , 2020, PALAIS DU PHARO, Marseille, France : CONFERENCE PROCEEDINGS / Nicoletta Calzolari, Frédéric Béchet, Philippe Blache, Khalid Choukri, Christopher Cieri, Thierry Declerck, Sara Goggi, Hitoshi Isahara, Bente Maegaard, Joseph Mariani, Hélène Mazo, Asuncion Moreno, Jan Odijk, Stelios Piperidis; Marseille: European Language Resources Association, s. 1674-1680

12th Language Resources and Evaluation Conference. Marsylia, 2020-05-11 - 2020-05-16

Methods for learning sentence representations have been actively developed in recent years. However, the lack of pre-trained models and datasets annotated at the sentence level has been a problem for low-resource languages such as Polish which led to less interest in applying these methods to language-specific tasks. In this study, we introduce two new Polish datasets for evaluating sentence embeddings and provide a comprehensive evaluation of eight sentence representation methods including Polish and multilingual models. We consider classic word embedding models, recently developed contextual embeddings and multilingual sentence encoders, showing strengths and weaknesses of specific approaches. We also examine different methods of aggregating word vectors into a single sentence vector.


High-resolution net load forecasting for micro-neighbourhoods with high penetration of renewable energy sources

Paweł Kobyliński, Mariusz Wierzbowski, Krzysztof Piotrowski

2020 International Journal of Electrical Power and Energy Systems, T. 117, May 2020, Article number 105635

Though extensive, the literature on electrical load forecasting lacks reports on studies focused on existing re- sidential micro-neighbourhoods comprising small numbers of single-family houses equipped with solar panels. This paper provides a full description of an ANN-based model designed to predict short-term high-resolution (15- min intervals) micro-scale residential net load profiles. Since it seems especially relevant due to the specificity of local autocorrelations in load signal, in this paper we put stress on the systematic approach to feature selection in the context of lagged signal. We performed a case study of a real micro-neighbourhood comprising only 75 single-family houses. The obtained average prediction error was equivalent to 5.4 per cent of the maximal measured net load. The issues, i.e.: (1) the feasibility of micro-scale residential load forecasting taking into account renewable energy penetration, (2) the feasibility to predict net load with dense temporal resolution of 15 min, (3) the feature selection problem, (4) the proposed prosumption- and comparison-oriented prediction model key performance measure, could be of interest to engineers designing energy balancing systems for local smart grids.


Information Extraction System for Transforming Unstructured Text Data in Fire Reports into Structured Forms: A Polish Case Study

Marcin Mirończuk

2020 Fire Technology, T. 56, s. 545-581

In this paper, the author presents a novel information extraction systemthat analyses fire service reports. Although the reports contain valuable informationconcerning fire and rescue incidents, the narrative information in these reports hasreceived little attention as a source of data. This is because of the challenges associ-ated with processing these data and making sense of the contents through the use ofmachines. Therefore, a new issue has emerged: How can we bring to light valuableinformation from the narrative portions of reports that currently escape the attentionof analysts? The idea of information extraction and the relevant system for analysingdata that lies outside existing hierarchical coding schemes can be challenging forresearchers and practitioners. Furthermore, comprehensive discussion and proposi-tions of such systems in rescue service areas are insufficient. Therefore, the authorcomprehensively and systematically describes the ways in which information extrac-tion systems transform unstructured text data from fire reports into structured forms.Each step of the process has been verified and evaluated on real cases, including datacollected from the Polish Fire Service. The realisation of the system has illustratedthat we must analyse not only text data from the reports but also consider the dataacquisition process. Consequently, we can create suitable analytical requirements.Moreover, the quantitative analysis and experimental results verify that we can (1)obtain good results of the text segmentation (F-measure 95.5%) and classificationprocesses (F-measure 90%) and (2) implement the information extraction process andperform useful analysis