To tackle climate change, caused by increasing levels of greenhouse gases in the atmosphere, legislation to curb CO2 emissions have become more stringent. The steel industry is responsible for 6.5% of anthropogenic CO2 emissions. Consequently, the steel industry has been endeavouring to develop green steel production technologies with HIsarna being one of the promising alternative ironmaking technologies. HIsarna technology offers the flexibility to use reductants other than solid carbon materials. Natural gas has the potential to be used either partially or fully as an alternative to solid carbon materials in the ironmaking process. Its major component, methane, contains two reductants: elemental carbon (Carbon Black) and hydrogen. The by-product of using hydrogen in the ironmaking process is steam, resulting in reduced CO2 emissions from the overall process. This PhD research studies the reaction mechanisms of methane in the smelting reduction vessel (SRV) based on five main objectives with the aim to provide the fundamental knowledge needed to (partially) replace coal with natural gas in the HIsarna process.

A static heat and mass balance model was developed for the HIsarna process to predict CO2 emissions and coal consumption. The results were validated against pilot plant data. The model was used to predict the CO2 emissions for two cases: (1) methane used as a carrier gas of coal (2) methane fully replacing coal. The research also investigated the gasification reactivity of carbon black (CB) along with its chemical-physical properties, which could potentially effect its reactivity. The reduction behaviour of the individual reductants, carbon and hydrogen, with molten HIsarna slag in conditions similar to the smelting reduction vessel (SRV) was examined. Finally, the reaction mechanism between methane and the HIsarna slag was investigated. This research has been conducted systematically and provides a fundamental understanding of the methane gas reduction behaviour in HIsarna thermal conditions.

Efficient and effective similarity evaluation is the fundamental cornerstone of various applications, including social networks, search engines and recommender systems. This thesis presents novel strategies for link-based similarity search on network data, aimed at strengthening computational efficiency and enhancing the accuracy of similarity assessment. The contributions of this thesis are summarised as follows:
(1) Two novel computation-sharing methods are proposed, specifically for efficient all-pairs queries in the RoleSim algorithm. Both methods strategically aim at minimising redundant calculations and effectively reusing previously computed maximum weighted matching information. The first method utilises a Radix Trie, arranging nodes in incremental order of their in-neighbour sets. The second method employs a Steiner tree for optimising topological ordering.
(2) The second contribution introduces efficient and scalable algorithms for partial-pairs and single-source query types. These algorithms are specifically designed for selective similarity search, effectively extracting and caching relevant nodes to avoid unnecessary computational costs.
(3) The third significant contribution is the conceptualisation of RoleSim*, a novel similarity model that provides a more comprehensive and precise evaluation of pairwise role similarities. RoleSim* establishes unique properties of existence, uniqueness, and axiomatic properties. A distance metric based on RoleSim* similarity is also derived, confirming its adherence to the triangle inequality, which implies the sum-transitivity of similarity scores. Furthermore, the threshold-based RoleSim* model efficiently reduces computation time while preserving provable accuracy, showcasing the practical advantages of this approach.
(4) The final contribution is a novel graph embedding solution for recommendation tasks, designed to simultaneously capture local proximity and structural role information. This solution blends the advantages of node2vec and struc2vec, offering a regulated transition between proximity-based and role-based representations.
Comprehensive real-world experiments demonstrate the superior performance of the proposed methods, providing compelling evidence for the effectiveness of the methodologies presented in this thesis.

Sonic hedgehog (SHH) is a secreted signalling protein playing key roles during embryogenesis and tissue homeostasis. SHH interacts with its main receptor PTCH1, the coreceptors CDON, BOC and GAS1, and the negative regulator HHIP to ensure optimal pathway regulation in vivo. Understanding these interactions both at the molecular and physiological level is crucial to find new ways to prevent pathway mis-regulation. Complex structures between SHH and most binding partners have been solved by Cryo-EM or X-ray crystallography methods, while the structure of the coreceptor GAS1 remains elusive, hindering any understanding of the molecular mechanism in which SHH and GAS1 engage to regulate the pathway. Genetically encoded photo-crosslinkers have been extensively used as a successful tool to map interacting protein interfaces in the native environment of live cells. In this study, we used the unnatural amino acid p-benzoyl-L-phenylalanine (BPA) to firstly analyse SHH-HHIP and SHH-HHIP interface, and later to study novel interfaces between SHH and the coreceptor GAS1. Lastly, zebrafish was used as a model system to analyse the physiological relevance of gas1, by analysing gas1 expression pattern and by generating gas1a-/- and gas1b-/- mutant lines for phenotypic characterization.

Interest in non-religion (the "nones") has grown significantly in recent years, but scholars have yet to devote sufficient energy to exploring the phenomenon of deconversion ("nonversion") in non-Christian contexts. This is a particularly noteworthy lacuna where Islamic origin nonverts ("ex-Muslims") are concerned, given that Muslims are the UK's largest non-Christian religious group. Further, existing interest is primarily sociological, neglecting political dimensions of the ex-Muslim phenomenon.

This study has therefore set out to understand the political dimensions of British ex-Muslims and their movement, and in so doing to make a substantial contribution to the burgeoning but still young study of the "nones". To accomplish this, I carried out 25 interviews with British ex-Muslims, using a narrative interview methodology, to explore questions of identity (both individual and group), and the means by which isolated individuals across the UK and the world can coalesce, organise, and become active.

Relying on Subcultural Identity Theory, and elements of Social Movement Theory, I posit that ex-Muslims are a primarily online subcultural movement, capable of thriving not in spite of the immense hostility they face, but because of it. Indeed, this adversity has created a coherent and unique identity.

This online subculture is a dynamic space for identity formation both individual and group, recruitment, mobilisation, the sharing of narratives and activism. But it carries risks, including growing online censorship by states and social media corporations, and in the offline world ex-Muslims are often "closeted", ostracised and unable to gather. Ex-Muslim organisations fill some of this gap, but the barriers for ex-Muslims remain high, and include the failure of the state to understand their perspectives, despite 15 years of political activism.