No description available.

No description available.

No description available.

No description available.

No description available.

Though several commercial wine recommender systems exist, they are largely tailored to consumers outside of South Africa (SA). Consequently, these systems are of limited use to novice wine consumers in SA. To address this, the aim of this research is to develop a system for South African consumers that yields high-quality wine recommendations, maximises the accuracy of predicted ratings for those recommendations and provides insights into why those suggestions were made. To achieve this, a hybrid system “soMLier” (pronounced “sommelier”) is built in this thesis that makes use of two datasets. Firstly, a database containing several attributes of South African wines such as the chemical composition, style, aroma, price and description was supplied by wine.co.za (a SA wine retailer). Secondly, for each wine in that database, the numeric 5-star ratings and textual reviews made by users worldwide were further scraped from Vivino.com to serve as a dataset of user preferences. Together, these are used to develop and compare several systems, the most optimal of which are combined in the final system. Item-based collaborative filtering methods are investigated first along with model-based techniques (such as matrix factorisation and neural networks) when applied to the user rating dataset to generate wine recommendations through the ranking of rating predictions. Respectively, these methods are determined to excel at generating lists of relevant wine recommendations and producing accurate corresponding predicted ratings. Next, the wine attribute data is used to explore the efficacy of content-based systems. Numeric features (such as price) are compared along with categorical features (such as style) using various distance measures and the relationships between the textual descriptions of the wines are determined using natural language processing methods. These methods are found to be most appropriate for explaining wine recommendations. Hence, the final hybrid system makes use of collaborative filtering to generate recommendations, matrix factorisation to predict user ratings, and content-based techniques to rationalise the wine suggestions made. This thesis contributes the “soMLier” system that is of specific use to SA wine consumers as it bridges the gap between the technologies used by highly-developed existing systems and the SA wine market. Though this final system would benefit from more explicit user data to establish a richer model of user preferences, it can ultimately assist consumers in exploring unfamiliar wines, discovering wines they will likely enjoy, and understanding their preferences of SA wine.

No description available.

Most complex systems in the world are time-dependent and dynamic in nature, many of which are suitable to be modeled as dynamic networks that evolve over time. From the analysis of time-varying social networks to the analysis of functional brain networks in longitudinal study designs, new statistical methods are needed for a better understanding of network dynamics and the underlying complex systems. Our work revolves around statistical modeling, sampling and inference for dynamic networks driven by various applications. Specifically, we develop a class of random graph hidden Markov models (RGHMM) for percolation in noisy dynamic networks to infer the type of phase transitions undergone in epileptic seizures. We also develop a broadly applicable class of coevolving latent space network with attractors (CLSNA) models for characterizing coevolutionary phenomenon in social behaviors, such as flocking and polarization, and use it under the context of American politics to disentangle positive and negative partisanship in affective polarization. Finally, we provide uncertainty quantification in conjunction with estimation of the frequency of motifs in dynamic networks under a certain sampling model, by studying the asymptotics for streaming data applications.

No description available.

The graph matching (GM) problem seeks to find an alignment between the vertex sets of two graphs and has applications in social network analysis, bioinformatics, and pattern recognition. In this dissertation, we propose to (a) develop the iGraphMatch R package for common GM algorithms and steps for analysis of GM, (b) develop the mutual nearest neighbor algorithm for matching node pairs with high precision in polynomial time, and (c) analyze a user de-anonymization problem based on their Venmo transaction networks along with various information associated with users and their transactions. The iGraphMatch package enables seamless matching of generalized graphs with versatile options for the form of input graphs and the specification of available prior information, as well as evaluation of matching performance, and visualization. For the methodology component, we empirically demonstrate the effectiveness of the mutual nearest neighbor algorithm for finding a high precision sub-match. In addition, we develop a mathematical framework for the analysis of the proposed method and provide conditions for achieving high matching precision under a correlated random graph model. Lastly, for the challenging GM problem on Venmo transaction networks, we introduce a similarity-based learning method for integrating multiple features into a single similarity score that optimizes the expected number of correctly matched nodes.