Licentiate dissertation, written in Swedish with an English abstract. The educational goal of “a school for all” creates many challenges. Issues of socialjustice and equity are central tenets of the concept of inclusive education. Despitethe goal of comprehensive education for all children, for many pupils with physicaldisabilities in an inclusive school system PE (Physical Education) teaching canresult in experiences of injustice. In the struggle to achieve social equity and createeducational experiences where disability does not matter, it seams to matter verymuch.The aim of this study is to illustrate a dilemma and its potential outcomeswhere demands for justice for pupils with physical disabilities are raised in inclusivephysical education. In the study, based on theories of redistribution andrecognition, two empirical examples are presented.The first example, in the context of the decentralisation of Swedish compulsoryschools, illustrates degrees of awareness about numbers of pupils with physicaldisabilities attending compulsory schools located in a case municipality. Foursources, with varying responsibilities for disabled children, all reported differentnumbers of pupils. Thus, with no congruent data at municipality level, distribu-tion and redistribution of necessary resources becomes difficult.In the second example of a ten year old boy with a physical disability,experiences of participation in inclusive physical education are described, fromthe perspective of five different actors (the boy himself, his PE teachers (2), hisparents, classmates, and his personal assistant).Data for this case study was gathered through interviews and systematic ob-servation. In this particular case, the outcome of inclusive PE was judged to besuccessful. The example illustrates the importance of recognition needing to befulfilled at several distributive levels. The positive outcome is discussed in terms ofthe combination of identification of particular special needs, sensitive adaptation,and general respect for the child with physical disability.Dilemmas of justice for disabled pupils in physical education are best studiedwithin a multi-level context. Recognition and redistribution demands need to besimultaneously addressed in order to fulfil the goal of equitable education forpupils with physical disabilities attending PE within the compulsory school sys-tem. An attempt to combine these different ideological approaches is discussedfrom the perspective of social status.The outcome of the two empirical examples presented in this study illustrategeneral dilemmas reaching beyond the educational challenges facing pupils withphysical disabilities and the responses of the compulsory school system to them.

The natural world provides sensory systems with noisy and ambiguous information, which is often transformed into a more stable categorical percept. This thesis aims to investigate the nature of the neural representations in the visual system that support this transformation. To do so, we will employ a behavioural task that requires participants to average several independent sources of perceptual information. This task allows for the dissociation of two theoretically orthogonal sources of decision uncertainty: the mean distance of the perceptual information from a category boundary and the variability of the evidence under consideration. Behaviourally, both decreasing the mean distance to bound of information and increasing information variability are associated with increased errors and prolonged response times. We will present a computational model that can account for the independent behavioural effects of these two sources of uncertainty by assuming that categorical decisions are made on the basis of a probabilistic transformation of perceptual evidence. BOLD measurements demonstrate that these effects of mean and variability are supported by a partially dissociable network of brain regions. Electroencephalography demonstrates the differential influence of mean and variance in the pre- and post-decision period. Furthermore, we show that there is adaptation at the level of the perceptual representation to the information variance. Not only does this show that the visual system must represent information at the summary level, in addition to individual feature-based representation, but it also suggests that the costs associated with this form of perceptual uncertainty can be largely mitigated by the adoption of a more suitable representational range.

To obtain valid and reliable research findings it is important to follow the process to validate measuring instruments. This entails determining the psychometric properties of a measure to eliminate or decrease the presence of measurement errors. Measurement errors have a negative impact on the validity of research findings. The aim of this study was to perform a confirmatory factor analytic study on the Aspects of Identity Questionnaire (AIQ-IV) to assess the model fit of the data. Confirmatory factor analysis was employed as part of the process to assess construct validity of a questionnaire to discover the misrepresenting influences of these measurement errors and to provide a foundation for further research. The AIQ-IV was administered to a sample of 157 participants in the South African context including different race, gender, age and occupation groups, drawn by means of convenient sampling. The research results and fit indices indicated that the data reflected a reasonable model fit. Copyright 2010, University of Pretoria. All rights reserved. The copyright in this work vests in the University of Pretoria. No part of this work may be reproduced or transmitted in any form or by any means, without the prior written permission of the University of Pretoria. Please cite as follows: Els, L 2010, A confirmatory factor analytic study of the aspects of identity questionnaire (AIQ-IV), MCom dissertation, University of Pretoria, Pretoria, viewed yymmdd < http://upetd.up.ac.za/thesis/available/etd-02092012-122330 / > C12/4/67/gm / Dissertation (MCom)--University of Pretoria, 2010. / Human Resource Management / unrestricted

De nos jours, l’identification automatique des espèces de plantes par l’analyse d’images, devient incontournable pour faire perdurer, standardiser voire approfondir les connaissances relatives à la communauté végétale. Cette thèse aborde le problème d’identification automatique des espèces de plantes en utilisant les images de feuilles. Elle s’attaque à deux principaux challenges: Le premier challenge est le grand nombre et la large variabilité de la morphologie foliaire des espèces et le deuxième challenge est la variabilité intra-espèces qui se manifeste localement au niveau de régions particulières des feuilles. Pour pallier à ces deux problèmes, un retour à la botanique et notamment aux concepts botaniques foliaires a été établi pour définir une structuration automatique des feuilles à deux niveaux: Le premier niveau concerne un schéma de catégorisation selon les deux concepts botaniques “arrangement” et “lobation”. Le deuxième niveau consiste à définir les parties sémantiques qui composent la feuille. L’approche de la thèse s’articule autour de deux principaux volets: Dans le premier volet, nous nous intéressons à mettre en place cette structuration guidée par la sémantique botanique en définissant des propriétés géométriques simples corrélées avec les définitions et les observations botaniques. Dans le deuxième volet, nous étudions la faisabilité et la pertinence d’intégrer cette structuration dans la chaîne d’identification. Particulièrement, nous établissons des recherches ciblées dans les catégories et nous définissons des modèles de parties à significations botaniques. Nous établissons notre évaluation sur les deux bases d’images de Scans de feuilles ImageCLEF 2011 et ImageCLEF 2012. Nous comparons notre approche par rapport à un schéma d’identification de référence, appliqué sur la totalité de la base et en utilisant l’image entière, et par rapport à plusieurs méthodes référencées dans la littérature. / Nowadays, automatic identification of plant species, by image analysis, has become crucial to maintain, standardize or deepen knowledge about the plant community. This thesis focus on the problem of automatic identification of plant species using leaf images. It addresses two main challenges: The first challenge is the large number and the high variability in foliar morphology across species. The second challenge is the intra-species variability which occurs locally at particular regions of leaves. To overcome these two problems, a return to botany and especially to leaf botanical concepts is established in order to define an automatic structuring of leaves at two levels: The first level concerns a categorisation scheme according to the botanical concepts “arrangement” and “lobation". The second level consists in decomposing leaves into semantic parts. The approach of the thesis is based on two key parts: In the first part, we focus on establishing this botanical-based structuring process by defining simple geometric properties correlated with botanical definitions and observations. In the second part, we investigate the feasibility and opportunities to integrate this structuring process in the identification scheme. Particularly, we make targeted researches in categories and we define specific part-based models.Experiments are conducted using the ImageCLEF 2011 and 2012 Scan images leaf databases. We compare our approach with respect to the reference identification scheme, applied on the whole databaseand using the entire images, and with respect to several methods referenced in the literature.

The languages spoken on northern Halmahera and surrounding small islands constitute a group of related ‘Papuan’ languages called North Halmahera. They are also, together with other Papuan and Austronesian languages, included in a proposed sprachbund which is called East Nusantara. Neuter gender and numeral classifiers have both been proposed to characterize the sprachbund. Consequently,an investigation of the noun categorisation systems in the North Halmahera languages, which is the subject of this study, can be of interest for the characterization of the sprachbund. The method for the investigation is to search for information about seven languages in existing grammatical descriptions, complemented with information which can be culled from published texts in the languages. There are mainly two categorisation systems in all the investigated languages: genders and numeral classifiers. The numerals often contain fossilized prefixes. Among the numeral classifiers, the human classifiers are special because of their origin from pronominal undergoer prefixes and the limitations of its use in some languages. Except in West Makian, there is a default classifier and a classifier for trees, and secondarily for houses, in all languages. A classifier for two-dimensional objects is also quite common. The other classifiers are used with a very limited number of nouns. / Språken som talas på norra Halmahera och omkringliggande småöar utgör en grupp av besläktade ’papuanska’ språk som kallas Nord-Halmahera-språk. De ingår också, ihop med andra papuanska och austronesiska språk, i ett antaget sprachbund som kallas för Östra Nusantara. Neutrum-genus ochnumeriska klassificerare har båda föreslagits karakterisera sprachbundet. Således kan en undersökning av substantivklassificering från ett historiskt och typologiskt perspektiv i Nord-Halmahera-språken, som är ämnet för den här studien, vara av intresse för karakteriseringen av sprachbundet. Metoden för undersökningen är att söka efter information för sju språk i existerande grammatiska beskrivningar, kompletterat med information som kan fås från publicerade texter på språken. Det förekommer huvudsakligen två klassificeringssystem i alla de undersökta språken: genus och numeriska klassificerare. Räkneorden innehåller ofta fossiliserade prefix. Bland de numeriska klassificerarna ärmännisko-klassificerarna speciella genom sitt ursprung i pronominella undergoer-prefix och den begränsade användnings-möjligheten i vissa språk. Utom i västmakianska, förekommer en allmän klassificerare och en klassificerare för träd, och sekundärt för hus, i alla språk. En klassificerare för två-dimensionella objekt är också ganska vanlig. Övriga klassificerare används oftast med ett mycket begränsat antal substantiv.

Natural language processing and Machine learning are subjects undergoing intense study nowadays. These fields are continually spreading, and are more interrelated than ever before. A case in point is text classification which is an instance of Machine learning(ML) application in Natural Language processing(NLP).Although these subjects have evolved over the recent years, they still have some problems that have to be considered. Some are related to the computing power techniques from these subjects require, whereas the others to how much training data they require.The research problem addressed in this thesis regards lack of knowledge on whether Machine learning techniques such as Word2Vec, Bidirectional encoder representations from transformers (BERT) and Support vector machine(SVM) classifier can be used for text classification, provided only a small training set. Furthermore, it is not known whether these techniques can be run on regular laptops.To solve the research problem, the main purpose of this thesis was to develop two separate conversational interfaces utilizing text classification techniques. These interfaces, provided with user input, can recognise the intent behind it, viz. classify the input sentence within a small set of pre-defined categories. Firstly, a conversational interface utilizing Word2Vec, and SVM classifier was developed. Secondly, an interface utilizing BERT and SVM classifier was developed. The goal of the thesis was to determine whether a small dataset can be used for intent classification and with what accuracy, and if it can be run on regular laptops.The research reported in this thesis followed a standard applied research method. The main purpose was achieved and the two conversational interfaces were developed. Regarding the conversational interface utilizing Word2Vec pre-trained dataset, and SVM classifier, the main results showed that it can be used for intent classification with the accuracy of 60%, and that it can be run on regular computers. Concerning the conversational interface utilizing BERT and SVM Classifier, the results showed that this interface cannot be trained and run on regular laptops. The training ran over 24 hours and then crashed.The results showed that it is possible to make a conversational interface which is able to classify intents provided only a small training set. However, due to the small training set, and consequently low accuracy, this conversational interface is not a suitable option for important tasks, but can be used for some non-critical classification tasks. / Natural language processing och maskininlärning är ämnen som forskas mycket om idag. Dessa områden fortsätter växa och blir allt mer sammanvävda, nu mer än någonsin. Ett område är textklassifikation som är en gren av maskininlärningsapplikationer (ML) inom Natural language processing (NLP).Även om dessa ämnen har utvecklats de senaste åren, finns det fortfarande problem att ha i å tanke. Vissa är relaterade till rå datakraft som krävs för dessa tekniker medans andra problem handlar om mängden data som krävs.Forskningsfrågan i denna avhandling handlar om kunskapsbrist inom maskininlärningtekniker som Word2vec, Bidirectional encoder representations from transformers (BERT) och Support vector machine(SVM) klassificierare kan användas som klassification, givet endast små träningsset. Fortsättningsvis, vet man inte om dessa metoder fungerar på vanliga datorer.För att lösa forskningsproblemet, huvudsyftet för denna avhandling var att utveckla två separata konversationsgränssnitt som använder textklassifikationstekniker. Dessa gränssnitt, give med data, kan känna igen syftet bakom det, med andra ord, klassificera given datamening inom ett litet set av fördefinierade kategorier. Först, utvecklades ett konversationsgränssnitt som använder Word2vec och SVM klassificerare. För det andra, utvecklades ett gränssnitt som använder BERT och SVM klassificerare. Målet med denna avhandling var att avgöra om ett litet dataset kan användas för syftesklassifikation och med vad för träffsäkerhet, och om det kan användas på vanliga datorer.Forskningen i denna avhandling följde en standard tillämpad forskningsmetod. Huvudsyftet uppnåddes och de två konversationsgränssnitten utvecklades. Angående konversationsgränssnittet som använde Word2vec förtränat dataset och SVM klassificerar, visade resultatet att det kan användas för syftesklassifikation till en träffsäkerhet på 60%, och fungerar på vanliga datorer. Angående konversationsgränssnittet som använde BERT och SVM klassificerare, visade resultatet att det inte går att köra det på vanliga datorer. Träningen kördes i över 24 timmar och kraschade efter det.Resultatet visade att det är möjligt att skapa ett konversationsgränssnitt som kan klassificera syften, givet endast ett litet träningsset. Däremot, på grund av det begränsade träningssetet, och konsekvent låg träffsäkerhet, är denna konversationsgränssnitt inte lämplig för viktiga uppgifter, men kan användas för icke kritiska klassifikationsuppdrag.

Machine categorisation of the emotional content of music is an ongoing research area. Feature description and extraction for such a vague and subjective field as emotion presents a difficulty for human-designed audioprocessing. Research into machine categorisation of music based on genrehas expanded as media companies have increased their recommendation and automation efforts, but work into categorising music based on sentiment remains lacking. We took an informed experimental method towards finding a workable solution for a multimedia company, Ichigoichie, who wished to develop a generalizable classifier on musical qualities. This consisted of first orienting ourselves within the academic literature relevant on the subject, which suggested applying spectrographic pre-processing to the sound samples, and then analyzing these visually with a convolutional neural network. To verify this method, we prototyped the model in a high level framework utilizing Python which pre-processes 10 second audio files into spectrographs and then provides these as learning data to a convolutional neural network. This network is assessed on both its categorization accuracy and its generalizability to other data sets. Our results show that the method is justifiable as a technique for providing machine categorization of music based on genre, and even provides evidence that such a method is technically feasible for commercial applications today. / Maskinkategorisering av känsloprofilen i musik är ett pågående forskningsområde. Traditionellt sett görs detta med algoritmer som är skräddarsydda för en visstyp av musik och kategoriseringsområde. En nackdel med detta är att det inte går att applicera sådana algoritmer på flera användningsområden, och att det krävs både god musikkunnighet och även tekniskt vetande för att lyckas utveckla sådana algoritmer. På grund av dessa anledningar ökar stadigt mängden av forskning runt huruvida samma ändamål går att åstadkommas med hjälp av maskininlärningstekniker, och speciellt artificiella neuronnät, en delgrupp av maskininlärning. I detta forskningsprojekt ämnade vi att fortsätta med detta forskningsområde,och i slutändan hoppas kunna besvara frågan om huruvida det går att klassificera och kategorisera musik utifrån känsloprofilen inom musiken, med hjälp av artificiella neuronnät. Vi fann genom experimentell forskning att artificiella neuronnät är en mycket lovande teknik för klassificering av musik, och uppnådde goda resultat. Metoden som användes bestådde av spektrografisk ljudprocessering, och sedan analys av dessa spektrogram med konvolutionella neuronnät, en sorts artificiella neuronnät ämnade för visuell analys.

In recent years, artificial intelligence (AI) has been used frequently in the Human Resources industry in order to streamline recruitment processes and exclude human biases about candidates. The shift from human recruitment to recruitment using digital tools is affecting how workplaces are shaped and which candidates are hired. The purpose of this study is to investigate how artificial intelligence affects the recruitment process in Human Resources and what advantages and disadvantages can be generated by the increased use of AI technology in HR. By studying this, it is also possible to visualize how individuals and societal structures may be affected by AI recruitment. The study’s theoretical frame of reference is based on Anthony Giddens' (1991) theory of modern society followed by Pierre Bourdieu's (1986) theories of capital and social space. This study is a literature review which is a study that summarizes several scientific studies. The results of the study show, among other things, that the use of artificial intelligence in recruitment processes can have a negative impact on individuals and social structures in workplaces, it can contribute to homogeneous workplaces and the reproduction of society’s power structures. Furthermore, the study shows that there is a gap in knowledge regarding research on individuals' subjective experiences of AI recruitment. / Den artificiella intelligensen (AI) har under de senaste åren använts inom Human Resources branschen i syfte att effektivisera rekryteringsprocesser och exkludera mänskliga fördomar om kandidater. Skiftet från mänsklig rekrytering till rekrytering med hjälp av digitala verktyg påverkar hur arbetsplatser formas och vilka kandidater som anställs. Syftet med studien är att undersöka hur den artificiella intelligensen påverkar rekryteringsprocessen inom Human Resources och vilka för- och nackdelar som kan genereras av AI-teknologins ökade användning inom HR. Genom att studera detta synliggörs även hur individer och samhälleliga strukturer kan tänkas påverkas av AI-rekrytering. Studiens teoretiska ramverk utgår från Anthony Giddens (1991) teori om det moderna samhället följt av Pierre Bourdieus (1986) teorier om kapital och det sociala rummet. Denna studie är en litteraturstudie vilket är en studie som sammanfattar flera vetenskapliga studier. Studiens resultat visar bland annat att användandet av artificiell intelligens i rekryteringsprocesser kan ha en negativ inverkan på individer och strukturer på arbetsplatser, det kan bidra till homogena arbetsplatser och reproducering av samhälleliga maktstrukturer. Vidare visar studien att det finns ett kunskapsglapp vad gäller forskning kring individers subjektiva upplevelser av AI-rekrytering.

Before entering into a democratic dispensation, South African military and defence systems were constituted by seven disparate armed forces. The transformation of South Africa from a separatist state introduced renewed efforts and challenges to integrate what was once a divided military corps and society. In 1994, the formation of the South African National Defence Force (SANDF) was born out of the effort to integrate various statutory and non-statutory armed forces in South Africa, including forces from former TBVC states. Through a phenomenological inquiry, integration experiences of former TBVC Armed Force members into the new SANDF structure are investigated. The study aimed to find out from the former TBVC force members how they were affected by the integration process and what their perception with regards to the effectiveness of the integration process on enhancing representation on all rank levels in the new SANDF was. The findings from in-depth semi-structured interviews with 16 such members indicate that transformation of South Africa’s military outfit into an integrated system is not exactly an epitome of a new and different yet cohesive and unified structure. The notion of ‘integration’ is as such cast as essentially a problematic one where inequalities still reflect in how former statutory and non-statutory force members are treated particularly with regards to promotion opportunities. In addition, there are perceived lingering vestiges of a previous separatist system such as the use of Afrikaans language as a medium of instruction and communication, previous SADF policies which helps produce the idea of integration as more a process of absorption instead. Despite this problematisation of integration processes in shaping how the new SANDF outfit is currently experienced by members, there are perceived benefits from the change brought about by ‘integration’ of forces. Some benefits are as tangible as individual career advancement, while others tend to reflect impacts at systemic level of family where members indirectly profit from the reorganisation and call for adjustment to new settings and structures that followed integration / Research / M.A. (Research Consultation)

Following the 22 February 2011, MW 6.2 earthquake located on a fault beneath the Port Hills of Christchurch, fissuring of up to several hundred metres in length was observed in the loess and loess-colluvium of foot-slope positions in north-facing valleys of the Port Hills. The fissuring was observed in all major valleys, occurred at similar low altitudes, showing a contour-parallel orientation and often accompanied by both lateral compression/extension features and spring formation in the valley floor below. Fissuring locations studied in depth included Bowenvale Valley, Hillsborough Valley, Huntlywood Terrace–Lucas Lane, Bridle Path Road, and Maffeys Road–La Costa Lane. Investigations into loess soil, its properties and mannerisms, as well as international examples of its failure were undertaken, including study of the Loess Plateau of China, the Teton Dam, and palaeo-fissuring on Banks Peninsula. These investigations lead to the conclusion that loess has the propensity to fail, often due to the infiltration of water, the presence of which can lead to its instantaneous disaggregation. Literature study and laboratory analysis of Port Hills loess concluded that is has the ability to be stable in steep, sub-vertical escarpments, and often has a sub-vertically jointed internal structure and has a peak shear strength when dry. Values for cohesion, c (kPa) and the internal friction angle, ϕ (degrees) of Port Hills loess were established. The c values for the 40 Rapaki Road, 3 Glenview Terrace loess samples were 13.4 kPa and 19.7 kPa, respectively. The corresponding ϕ values were thought unusually high, at 42.0° and 43.4°.The analysed loess behaved very plastically, with little or no peak strength visible in the plots as the test went almost directly to residual strength. A geophysics resistivity survey showed an area of low resistivity which likely corresponds to a zone of saturated clayey loess/loess colluvium, indicating a high water table in the area. This is consistent with the appearances of local springs which are located towards the northern end of each distinct section of fissure trace and chemical analysis shows that they are sourced from the Port Hills volcanics. Port Hills fissuring may be sub-divided into three categories, Category A, Category B, and Category C, each characterised by distinctive features of the fissures. Category A includes fissures which display evidence of, spring formation, tunnel-gullying, and lateral spreading-like behaviour or quasi-toppling. These fissures are several metres down-slope of the loess-bedrock interface, and are in valleys containing a loess-colluvium fill. Category B fissures are in wider valleys than those in Category A, and the valleys contain estuarine silty sediments which liquefied during the earthquake. Category C fissures occurred at higher elevations than the fissures in the preceding categories, being almost coincident with bedrock outcropping. It is believed that the mechanism responsible for causing the fissuring is a complex combination of three mechanisms: the trampoline effect, bedrock fracturing, and lateral spreading. These three mechanisms can be applied in varying degrees to each of the fissuring sites in categories A, B, and C, in order to provide explanation for the observations made at each. Toppling failure can describe the soil movement as a consequence of the a three causative mechanisms, and provides insight into the movement of the loess. Intra-loess water coursing and tunnel gullying is thought to have encouraged and exacerbated the fissuring, while not being the driving force per se. Incipient landsliding is considered to be the least likely of the possible fissuring interpretations.