Production of Gaseous Tracer I2 from the Sodiumiodide Salt

Found in the nature in form different, the iodine has been used in diverse works in the area of the industry and health. The iodine is very unstable and volatile in the ambient temperature and the I2 is one of the diverse gaseous forms found. In this work was developed methodology for production of gaseous tracer from the sodium iodide (NaI) 0,1 M marked with 123I. The synthesis was processed with in chlorine acid (HCl) 1M and sodium iodate salt (NaIO3). The production of gas I2 initially was carried through in unit of glass with the inert material and the purpose was to study the kinetic of reaction. The synthesis occurs in the reaction bottle and the produced gas is stored in the collect bottle that contains a starch solution (5 g/100 mL water). To determine the efficiency of production of gas I2, analytic tests had been carried through, where the consumption of iodide ions of the bottle of reaction is measured. The optimization of production of the gaseous tracer was studied varying parameters as: concentration of iodide and iodate, concentration of acid and temperature. Then, the synthesis of the radiotracer was realized in the compact unit, being utilized as main reagent the salt radiated of sodium iodide, Na123I. The transportation of elementary iodine was studied by a scintillation detector NaI (2 x 2)” placed in the reaction bottle. To acquire the data, the detector use a set of electronic modules for the acquisition of signals generated.

The Metaethics of Nursing Codes of Ethics and Conduct

Nursing codes of ethics and conduct are features of professional practice across the world, and in the UK, the regulator has recently consulted on and published a new code. Initially part of a professionalising agenda, nursing codes have recently come to represent a managerialist and disciplinary agenda and nursing can no longer be regarded as a self-regulating profession.This paper argues that codes of ethics and codes of conduct are significantly different in form and function similar to the difference between ethics and law in everyday life. Some codes successfully integrate these two functions within the same document, while others, principally the UK Code, conflate them resulting in an ambiguous document unable to fulfil its functions effectively. The paper analyses the differences between ethical- codes and conduct-codes by discussing titles, authorship, level, scope for disagreement, consequences of transgression, language and finally and possibly most importantly agent-centeredness. It is argued that conduct codes cannot require nurses to be compassionate because compassion involves an emotional response. The concept of kindness provides a plausible alternative for conduct-codes as it is possible to understand it solely in terms of acts. But if kindness is required in conduct-codes, investigation and possible censure follows from its absence. Using examples it is argued that there are at last five possible accounts of the absence of kindness. As well as being potentially problematic for disciplinary panels, difficulty in understanding the features of blameworthy absence of kindness may challenge UK nurses who, following a recently introduced revalidation procedure, are required to reflect on their practice in relation to The Code. It is concluded that closer attention to metaethical concerns by code writers will better support the functions of their issuing organisations.

Grading mit Grappa - Ein Werkstattbericht

„Grappa“ ist eine Middleware, die auf die Anbindung verschiedener Autobewerter an verschiedene E-Learning-Frontends respektive Lernmanagementsysteme (LMS) spezialisiert ist. Ein Prototyp befindet sich seit mehreren Semestern an der Hochschule Hannover mit dem LMS „moodle“ und dem Backend „aSQLg“ im Einsatz und wird regelmäßig evaluiert. Dieser Beitrag stellt den aktuellen Entwicklungsstand von Grappa nach diversen Neu- und Weiterentwicklungen vor. Nach einem Bericht über zuletzt gesammelte Erfahrungen mit der genannten Kombination von Systemen stellen wir wesentliche Neuerungen der moodle-Plugins, welche der Steuerung von Grappa aus moodle heraus dienen, vor. Anschließend stellen wir eine Erweiterung der bisherigen Architektur in Form eines neuentwickelten Grappa-php-Clients zur effizienteren Anbindung von LMS vor. Weiterhin berichten wir über die Anbindung eines weiteren Autobewerters „Graja“ für Programmieraufgaben in Java. Der Bericht zeigt, dass bereits wichtige Schritte für eine einheitliche Darstellung automatisierter Programmbewertung in LMS mit unterschiedlichen Autobewertern für die Studierenden absolviert sind. Die praktischen Erfahrungen zeigen aber auch, dass sowohl bei jeder der Systemkomponenten individuell, wie auch in deren Zusammenspiel via Grappa noch weitere Entwicklungsarbeiten erforderlich sind, um die Akzeptanz und Nutzung bei Studierenden sowie Lehrenden weiter zu steigern.

Dokumentation der Befragungsinstrumente der Interventionsstudie im Projekt “Conditions and Consequences of Classroom Assessment” (Co²CA)

Das im Rahmen des DFG-Schwerpunktprogramms „Kompetenzmodelle“ durchgeführte Projekt „Conditions and Consequences of Classroom Assessment“ (Co²CA) geht in vier Teilstudien der Frage nach, wie formatives Assessment im Unterricht gestaltet werden kann, um sowohl eine präzise Leistungsmessung zu ermöglichen als auch positive Wirkungen auf den Lehr-Lernprozess zu erreichen. Das Project Co²CA leistet damit einen wichtigen Beitrag zur Erforschung zweier Kernelemente formativen Assessments – der detaillierten Diagnose von Schülerleistungen und der Nutzung der gewonnenen Informationen in Form lernförderlichen Feedbacks. Zentrale Idee von formativem Assessment (Lernbegleitende Leistungsbeurteilung und –rückmeldung) ist es mit Hilfe von Leistungsmessungen Informationen über den Lernstand der Schülerinnen und Schüler zu gewinnen und diese Informationen für die Gestaltung des weiteren Lehr- und Lernprozess zu nutzen. Den Lernenden kann auf Basis der Leistungsbeurteilung lernförderliches Feedback gegeben werden, um so die Diskrepanz zwischen Lernstand und Lernziel zu verringern. Die Kernelemente formativem Assessments bestehen also aus einer detaillierten Diagnose des Lernstandes und der Nutzung der gewonnen Informationen – z.B. in Form von Feedback. [...] Das vorliegende Skalenhandbuch dokumentiert die in der Unterrichtsstudie eingesetzten Befragungsinstrumente für Schülerinnen und Schüler sowie für Lehrkräfte. (DIPF/Autor)

Medidas cautelares de embargo de bienes estatales en procesos ejecutivos ante la jurisdicción contencioso-administrativa

A partir del principio general de inembargabilidad de bienes del Estado, el contenido de este documento incursiona en el tema de las excepciones al mismo legalmente contempladas, a efectos de verificar la aplicabilidad de la medida cautelar de embargo sobre tales bienes en los casos procedentes, cuando los particulares adelantan procesos ejecutivos ante la jurisdicción contenciosoadministrativa, en procura de defender sus intereses y del pago de dineros a cargo del Estado como consecuencia de contratos, sentencias, acreencias laborales, laudos arbitrales u otro tipo de documento eficaz. El propósito fue establecer claramente la normatividad aplicable en este tipo de eventos, aportando de este modo respuesta objetiva y confiable frente a la problemática generada tanto por las distintas posiciones en torno al tema adoptadas por el legislador, como igualmente por dificultades para clasificar los bienes y rentas estatales en orden a determinar su carácter y la posibilidad real de su embargo. Todo ello, además, con el referente de pronunciamientos jurisprudenciales adoptados como soporte conceptual básico puesto que desde esta perspectiva dan cuenta de una línea coincidente de pensamiento por parte de los altos organismos pertinentes.

Kies- / Sandkörper im Wesertal zwischen Rinteln und Porta Westfalica

In this article the sand-/gravelbodies from Hausberge-Veltheim and Krankenhagen- Möllenbeck in the Wesertal are described and compared considering their depositional and architectural regime. A scenario is developed to explain the genetic sequence of the deposits. The sand-/gravel-body exposed at Krankenhagen-Möllenbeck was deposited first in form of a marginal käme. Subsequently during the Drenthe-stade of the Saale ice age, the sedimentation of the sand- /gravelbody at Hausberge-Veltheim took place under the depositional environment of an end- moraine.

A comunicação oral formal: a exposição oral

Relatório de Estágio apresentado à Escola Superior de Educação de Lisboa para obtenção de grau de mestre em Ensino do 1.º e 2.º Ciclos de Ensino Básico

Fluviatile und äolische Prozesse im Gebiet der unteren Elbe bei Neuhaus (Niedersachsen) seit dem Weichselspätglazial

The Late Glacial and Holocene landscape development in the vicinity of the River Elbe near Neuhaus, Lower Saxony, was studied during geological mapping of the area. The geological and geobotanical methods used in these investigations were chosen to cope with the difficulties which arise during research on Quaternary flood plains in low country. Paleochannel fill and areas of flood-plain sediments were drilled, the lithology examined, and the sediments dated on the basis of their pollen content. No evidence was found for the existence before the Middle Ages of paleo- channels the size of the present River Elbe. Before the first measures were made to regulate the Elbe River, it was an anastomosing river system with numerous small branches. The lower parts of the flood-plain profiles are predominantly sand and the upper parts silty-clayey loam. With the construction of effective levees over the last several centuries, the flow velocity of the Elbe has increased considerably during high water periods and instead of the deposition of meadow loam, sand was deposited as natural levees. The main belt of sand dunes on the east bank of the Elbe overlies Preboreal to Boreal lake mud and is, therefore, of Holocene age.

Gewebeansatz-Strukturen auf pyritisierten Steinkernen von Ammonoideen

3400 pyritized internal moulds of Upper Devonian, Triassic, Jurassic and Lower Cretaceous ammonoids show various soft tissue attachment structures. They are preserved as regularly distributed black patterns on the moulds. All structures can be interpreted as attachment areas of muscles, ligaments and intracameral membranes. Paired structures are developed along the umbilicus and on the flanks of the moulds, unpaired ones appear on the middle of their dorsal and ventral sides. Strong lateral muscles cause paired twin lines on the flanks of the phragmocone and of the body chamber. A ventral muscle is deduced from small rounded or crescent shaped spots in front of each septum on the ventral side. These spots are often connected, forming a band-like structure. Broad dark external bands on the ventral side of the phragmocone, ventral preseptal areas in the posterior part of the living chamber, small twin lines or oval shaped areas on the ventral side of the living chamber represent paired or unpaired attachment areas of the hyponome muscle. A middorsal muscle is documented by small roughened areas in front of each dorsal lobe. Dark spots along the umbilicus, often connected and thus forming a band-like structure (tracking band), are remains of a pair of small dorsolateral muscles at the posterior end of the soft body. Dark bands, lines and rows of small crescent shaped structures behind the tips of sutural lobes are due to spotlike fixation places of the posterior part of the mantle and their translocation before subsequent septal secretion. Devonian goniatites had a paired system of lateral and ventrolateral muscles preserved on the moulds as black or incised lines on the flanks of the living chamber and as dark preseptal areas, ventrally indented. These structures represent the attachment areas of paired lateral cephalic and paired ventral hyponome retractors. Fine black lines on the phragmocone situated parallel to the sutures (pseudosutures) represent a rhythmical secretion of camera! membranes during softbody translocation. Goniatites had a paired system of lateral and ventrolateral muscles, whilst Neoammonoids have a paired lateral and dorsolateral system, and, additionally, an unpaired system on the ventral and on the dorsal side. Mesoammonoids show only a paired lateral and an unpaired dorsal one. Fine black lines situated parallel to the saddles and behind the lobes of the suture line can be interpreted as structures left during softbody translocation and a temporary attachment of rhythmical secreted cameral membranes. Cameral membranes had supported the efficiency of the phragmocone. Only some of the observed structures are also present in recent Nautilus. Differences in the form and position of attachment sites between ammonoids and recent Nautilus indicate different soft body organizations between ammonoids and nautiloids. The attachment structures of goniatites especially of tornoceratids can be compared with those of Nautilus which indicates Richter - Gewebeansatz-Strukturen bei Ammonoideen 3 a comparable mode of life. Differences in the form and position of attachment structures between goniatites and ammonites may indicate an increasing differentiation of the muscular system in the phylogeny of this group. Different soft body organization may depend on shell morphology and on a different mode of life. On the modification or reduction of distinct muscle systems ammonoids can be assigned to different ecotypes. Based on shell morphology and the attachment areas of cephalic and hyponome retractor muscles two groups can be subdivided: - Depressed, evolute morphotypes with longidome body-chambers show only small ventral hyponome retractor muscles. Lateral cephalic retractors are not developed. These morphotypes are adapted to a demersal mode of life. Without strong cephalic retractor muscles no efficient jet propulsion can be produced. These groups represent vertical migrants with efficient phragmocone properties (multilobate sutures, cameral membranes, narrow septal spacing). - Compressed, involute moiphotypes with brevidome body-chambers show strong cephalic and hyponome retractor muscles and represent a group of active swimmers. These morphotypes were able to live at different depths, in the free water column or/and near the seafloor. They are not confined only to one habitat. Most of the examined genera and species belong to this group. Changes of the attachment structures in the course of ontogeny confirm that juveniles of Amaltheus and Quenstedtoceras lived as passive planche drifters in upper and intermediate parts of the free water column after hatching. At the end of the juvenile stage with a shell diameter of 0,3 - 0,5 cm cephalic retractor muscles developed. With the beginning of an active swimming mode of life (neanic stage) the subadult animals left the free water column and moved into shallow water habitats. Fuciniceras showed no marked changes in the attachment structures during ontogeny. This indicates that there occur no differences in the mode of life between juvenile and adult growth stages. Based on attachment structures and shell morphology of Devonian goniatites their relation to the systematic position permits statements about probable phylogenetic relationships between the Cheiloceratidae and Tornoceratidae. In some cases attachment structures of ammonites permit statements about phylogenetic relationships on family and genus level.

Desencontros legais e morais na pesca artesanal: a Educação Ambiental política para a transformação socioambiental em Rio Grande/RS e São José do Norte/RS

A dissertação intitulada “Desencontros legais e morais na pesca artesanal: a Educação Ambiental política para a transformação socioambiental em Rio Grande/RS e São José do Norte/RS” busca analisar a compreensão das comunidades de pescadores artesanais do Rio Grande e de São José do Norte acerca de aspectos do desencontro entre as normas legais pesqueiras e as normas morais daqueles atores sociais, e os reflexos na conservação do meio ambiente e no exercício da cidadania ambiental, a fim de colaborar, por intermédio da Educação Ambiental, com o fortalecimento dos pescadores para a reversão do atual quadro de crise social e ambiental vivenciada na pesca artesanal da região, bem como para refletir sobre a ineficiência das regras legais aplicáveis no estuário da Lagoa dos Patos. Para tanto os objetivos foram estruturados de forma a investigar os aspectos do desencontro mencionado; analisar seus reflexos no declínio da conservação dos recursos renováveis e no exercício da cidadania pelos pescadores artesanais; e apresentar alternativas possíveis pela e para a Educação Ambiental, enquanto educação política, visando a proporcionar a transformação socioambiental dos pescadores para a garantia de sua participação cidadã nas esferas de decisão relativas à sua vida social e profissional. A pesquisa teve uma abordagem qualitativa, com enfoque teórico-metodológico nas representações coletivas dos pescadores. O trabalho foi composto de levantamento bibliográfico e documental, que serviu de apoio para a obtenção das representações mediante a realização de grupos focais junto a quatro comunidades de pescadores nas cidades pesquisadas. Esse corpus de pesquisa foi analisado com base na metodologia Análise Textual Discursiva. Os resultados obtidos indicaram que a crise instaurada na pesca ultrapassa a dicotomia existente entre as leis ambientais e as normas morais dos pescadores artesanais, emergindo o aspecto da necessidade financeira e de sobrevivência de pescadores e de suas famílias. Verificou-se, outrossim, que o desencontro pesquisado refletiu negativamente na conservação dos recursos naturais, na medida em que gerou a falta de efetividade de leis relacionadas àquela atividade profissional, repercutindo desfavoravelmente no exercício da cidadania ambiental pelos pescadores. Concluiu-se que a Educação Ambiental política possui um papel fundamental para o empoderamento dos pescadores artesanais em busca de sua participação ativa nos processos de gestão dos recursos pesqueiros estuarinos, numa verdadeira governança participativa, voltada à conservação ambiental e edificação do bem coletivo.

'Explicit Learning: an Effort towards Human Scheduling Algorithms'

Abstract Scheduling problems are generally NP-hard combinatorial problems, and a lot of research has been done to solve these problems heuristically. However, most of the previous approaches are problem-specific and research into the development of a general scheduling algorithm is still in its infancy. Mimicking the natural evolutionary process of the survival of the fittest, Genetic Algorithms (GAs) have attracted much attention in solving difficult scheduling problems in recent years. Some obstacles exist when using GAs: there is no canonical mechanism to deal with constraints, which are commonly met in most real-world scheduling problems, and small changes to a solution are difficult. To overcome both difficulties, indirect approaches have been presented (in [1] and [2]) for nurse scheduling and driver scheduling, where GAs are used by mapping the solution space, and separate decoding routines then build solutions to the original problem. In our previous indirect GAs, learning is implicit and is restricted to the efficient adjustment of weights for a set of rules that are used to construct schedules. The major limitation of those approaches is that they learn in a non-human way: like most existing construction algorithms, once the best weight combination is found, the rules used in the construction process are fixed at each iteration. However, normally a long sequence of moves is needed to construct a schedule and using fixed rules at each move is thus unreasonable and not coherent with human learning processes. When a human scheduler is working, he normally builds a schedule step by step following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not completed yet, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this research we intend to design more human-like scheduling algorithms, by using ideas derived from Bayesian Optimization Algorithms (BOA) and Learning Classifier Systems (LCS) to implement explicit learning from past solutions. BOA can be applied to learn to identify good partial solutions and to complete them by building a Bayesian network of the joint distribution of solutions [3]. A Bayesian network is a directed acyclic graph with each node corresponding to one variable, and each variable corresponding to individual rule by which a schedule will be constructed step by step. The conditional probabilities are computed according to an initial set of promising solutions. Subsequently, each new instance for each node is generated by using the corresponding conditional probabilities, until values for all nodes have been generated. Another set of rule strings will be generated in this way, some of which will replace previous strings based on fitness selection. If stopping conditions are not met, the Bayesian network is updated again using the current set of good rule strings. The algorithm thereby tries to explicitly identify and mix promising building blocks. It should be noted that for most scheduling problems the structure of the network model is known and all the variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus learning can amount to 'counting' in the case of multinomial distributions. In the LCS approach, each rule has its strength showing its current usefulness in the system, and this strength is constantly assessed [4]. To implement sophisticated learning based on previous solutions, an improved LCS-based algorithm is designed, which consists of the following three steps. The initialization step is to assign each rule at each stage a constant initial strength. Then rules are selected by using the Roulette Wheel strategy. The next step is to reinforce the strengths of the rules used in the previous solution, keeping the strength of unused rules unchanged. The selection step is to select fitter rules for the next generation. It is envisaged that the LCS part of the algorithm will be used as a hill climber to the BOA algorithm. This is exciting and ambitious research, which might provide the stepping-stone for a new class of scheduling algorithms. Data sets from nurse scheduling and mall problems will be used as test-beds. It is envisaged that once the concept has been proven successful, it will be implemented into general scheduling algorithms. It is also hoped that this research will give some preliminary answers about how to include human-like learning into scheduling algorithms and may therefore be of interest to researchers and practitioners in areas of scheduling and evolutionary computation. References 1. Aickelin, U. and Dowsland, K. (2003) 'Indirect Genetic Algorithm for a Nurse Scheduling Problem', Computer & Operational Research (in print). 2. Li, J. and Kwan, R.S.K. (2003), 'Fuzzy Genetic Algorithm for Driver Scheduling', European Journal of Operational Research 147(2): 334-344. 3. Pelikan, M., Goldberg, D. and Cantu-Paz, E. (1999) 'BOA: The Bayesian Optimization Algorithm', IlliGAL Report No 99003, University of Illinois. 4. Wilson, S. (1994) 'ZCS: A Zeroth-level Classifier System', Evolutionary Computation 2(1), pp 1-18.

'Bayesian Optimisation for Nurse Scheduling'

A Bayesian optimisation algorithm for a nurse scheduling problem is presented, which involves choosing a suitable scheduling rule from a set for each nurse's assignment. When a human scheduler works, he normally builds a schedule systematically following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not yet completed, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this paper, we design a more human-like scheduling algorithm, by using a Bayesian optimisation algorithm to implement explicit learning from past solutions. A nurse scheduling problem from a UK hospital is used for testing. Unlike our previous work that used Genetic Algorithms to implement implicit learning [1], the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The Bayesian optimisation algorithm is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, new rule strings have been obtained. Sets of rule strings are generated in this way, some of which will replace previous strings based on fitness. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. For clarity, consider the following toy example of scheduling five nurses with two rules (1: random allocation, 2: allocate nurse to low-cost shifts). In the beginning of the search, the probabilities of choosing rule 1 or 2 for each nurse is equal, i.e. 50%. After a few iterations, due to the selection pressure and reinforcement learning, we experience two solution pathways: Because pure low-cost or random allocation produces low quality solutions, either rule 1 is used for the first 2-3 nurses and rule 2 on remainder or vice versa. In essence, Bayesian network learns 'use rule 2 after 2-3x using rule 1' or vice versa. It should be noted that for our and most other scheduling problems, the structure of the network model is known and all variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus, learning can amount to 'counting' in the case of multinomial distributions. For our problem, we use our rules: Random, Cheapest Cost, Best Cover and Balance of Cost and Cover. In more detail, the steps of our Bayesian optimisation algorithm for nurse scheduling are: 1. Set t = 0, and generate an initial population P(0) at random; 2. Use roulette-wheel selection to choose a set of promising rule strings S(t) from P(t); 3. Compute conditional probabilities of each node according to this set of promising solutions; 4. Assign each nurse using roulette-wheel selection based on the rules' conditional probabilities. A set of new rule strings O(t) will be generated in this way; 5. Create a new population P(t+1) by replacing some rule strings from P(t) with O(t), and set t = t+1; 6. If the termination conditions are not met (we use 2000 generations), go to step 2. Computational results from 52 real data instances demonstrate the success of this approach. They also suggest that the learning mechanism in the proposed approach might be suitable for other scheduling problems. Another direction for further research is to see if there is a good constructing sequence for individual data instances, given a fixed nurse scheduling order. If so, the good patterns could be recognized and then extracted as new domain knowledge. Thus, by using this extracted knowledge, we can assign specific rules to the corresponding nurses beforehand, and only schedule the remaining nurses with all available rules, making it possible to reduce the solution space. Acknowledgements The work was funded by the UK Government's major funding agency, Engineering and Physical Sciences Research Council (EPSRC), under grand GR/R92899/01. References [1] Aickelin U, "An Indirect Genetic Algorithm for Set Covering Problems", Journal of the Operational Research Society, 53(10): 1118-1126,

Organisatorische Umsetzung eines E-Learning-Konzepts einer Hochschule am Beispiel des Zentrums virtUOS der Universität Osnabrück

Die ELAN-Initiative des Landes Niedersachsen treibt zurzeit den vermehrten Einsatz moderner E-Learning-Konzepte in Form dreier Pilotprojekte mit dem Ziel voran, landesweit ein akademisches Kompetenznetzwerk aufzubauen. Sämtliche Hochschulen des Landes sollen von den Kompetenzen der jeweils anderen Universitäten und Fachhochschulen profitieren und die Lehre damit effizienter gestalten. Die an einem dieser Piloten beteiligte Universität Osnabrück nahm die Initiative zum Anlass, ein neues Dienstleistungszentrum zu gründen, dessen Aufgabe sich durch die Unterstützung der multi- und telemedialen Lehre definiert. Dieser Beitrag gibt einen Einblick in die Organisation dieses Zentrums, dessen Forschungs- und Dienstleistungen und zeigt die daraus resultierenden Vorteile auf. (DIPF/Orig.)

Software para apoio à gestão de empresas recorrendo a Cloud Computing

Dissertação apresentada ao Instituto Politécnico de Castelo Branco para cumprimento dos requisitos necessários à obtenção do grau de Mestre em Desenvolvimento de Software e Sistemas Interativos, realizada sob a orientação científica do Doutor Fernando Reinaldo Silva Garcia Ribeiro e do Doutor José Carlos Meireles Monteiro Metrôlho, Professores Adjuntos da Unidade Técnico-Científica de Informática da Escola Superior de Tecnologia do Instituto Politécnico de Castelo Branco.

'Bayesian Optimisation for Nurse Scheduling'

A Bayesian optimisation algorithm for a nurse scheduling problem is presented, which involves choosing a suitable scheduling rule from a set for each nurse's assignment. When a human scheduler works, he normally builds a schedule systematically following a set of rules. After much practice, the scheduler gradually masters the knowledge of which solution parts go well with others. He can identify good parts and is aware of the solution quality even if the scheduling process is not yet completed, thus having the ability to finish a schedule by using flexible, rather than fixed, rules. In this paper, we design a more human-like scheduling algorithm, by using a Bayesian optimisation algorithm to implement explicit learning from past solutions. A nurse scheduling problem from a UK hospital is used for testing. Unlike our previous work that used Genetic Algorithms to implement implicit learning [1], the learning in the proposed algorithm is explicit, i.e. we identify and mix building blocks directly. The Bayesian optimisation algorithm is applied to implement such explicit learning by building a Bayesian network of the joint distribution of solutions. The conditional probability of each variable in the network is computed according to an initial set of promising solutions. Subsequently, each new instance for each variable is generated by using the corresponding conditional probabilities, until all variables have been generated, i.e. in our case, new rule strings have been obtained. Sets of rule strings are generated in this way, some of which will replace previous strings based on fitness. If stopping conditions are not met, the conditional probabilities for all nodes in the Bayesian network are updated again using the current set of promising rule strings. For clarity, consider the following toy example of scheduling five nurses with two rules (1: random allocation, 2: allocate nurse to low-cost shifts). In the beginning of the search, the probabilities of choosing rule 1 or 2 for each nurse is equal, i.e. 50%. After a few iterations, due to the selection pressure and reinforcement learning, we experience two solution pathways: Because pure low-cost or random allocation produces low quality solutions, either rule 1 is used for the first 2-3 nurses and rule 2 on remainder or vice versa. In essence, Bayesian network learns 'use rule 2 after 2-3x using rule 1' or vice versa. It should be noted that for our and most other scheduling problems, the structure of the network model is known and all variables are fully observed. In this case, the goal of learning is to find the rule values that maximize the likelihood of the training data. Thus, learning can amount to 'counting' in the case of multinomial distributions. For our problem, we use our rules: Random, Cheapest Cost, Best Cover and Balance of Cost and Cover. In more detail, the steps of our Bayesian optimisation algorithm for nurse scheduling are: 1. Set t = 0, and generate an initial population P(0) at random; 2. Use roulette-wheel selection to choose a set of promising rule strings S(t) from P(t); 3. Compute conditional probabilities of each node according to this set of promising solutions; 4. Assign each nurse using roulette-wheel selection based on the rules' conditional probabilities. A set of new rule strings O(t) will be generated in this way; 5. Create a new population P(t+1) by replacing some rule strings from P(t) with O(t), and set t = t+1; 6. If the termination conditions are not met (we use 2000 generations), go to step 2. Computational results from 52 real data instances demonstrate the success of this approach. They also suggest that the learning mechanism in the proposed approach might be suitable for other scheduling problems. Another direction for further research is to see if there is a good constructing sequence for individual data instances, given a fixed nurse scheduling order. If so, the good patterns could be recognized and then extracted as new domain knowledge. Thus, by using this extracted knowledge, we can assign specific rules to the corresponding nurses beforehand, and only schedule the remaining nurses with all available rules, making it possible to reduce the solution space. Acknowledgements The work was funded by the UK Government's major funding agency, Engineering and Physical Sciences Research Council (EPSRC), under grand GR/R92899/01. References [1] Aickelin U, "An Indirect Genetic Algorithm for Set Covering Problems", Journal of the Operational Research Society, 53(10): 1118-1126,