Targeting a small portion of customers for advertising has long been recognized by businesses. In this thesis we proposed a novel approach to promoting products with no prior transaction records. This approach starts with discovering the GP-association rules between customer types and product genres that had occurred frequently in transaction records. Customers are characterized by demographic attributes, some of these attributes have concept hierarchies and products can be generalized through some product taxonomy. Based on GP-association rules set, we developed a comprehensive algorithm to locating a short list of prospective customers for a given promotion product. The new approach was evaluated using the patron¡¦s circulation data from OPAC system of our university library. We measured the accuracy of estimated method and the effectiveness of targeted advertising in different parameters. The result shows that our approach achieved higher accuracy and effectiveness than other methods.

Στην παρούσα διπλωματική εργασία προτείνουμε και υλοποιούμε ένα σύστημα, το οποίο κινείται στα πλαίσια των τεχνολογιών κινητού υπολογισμού και σχεδιάστηκε για να χρησιμοποιείται από τους καταναλωτές κατά την διαδικασία αγοράς αγαθών από τα ράφια των super market. / In this postgraduate project a proposed system is built within mobile computing technology and is designed for use from consumers while they go for shopping to the super market.

Effects of marketing mix tools, such as price promotions and digital advertising, are dynamic and long-term. Yet marketing literature has mainly focused on their immediate and short-term effects. Marketing Mix changes consumers' state, such as expectations and beliefs, and the changed state affects consumers' subsequent behaviors and responses to marketing stimuli and treatment, which may cause unintended consequences and generate important marketing opportunities as well. In Essay 1, we uncover the perils of tensile discounts (e.g., maximum discounts such as "up to 60% off") by examining their effects on store visits and purchases. Through a series of studies and using a multi-method approach including empirical data analysis, quasi-field experiment, numerical simulations, and controlled lab studies, we find that despite the initial efficacy in shaping consumer discount expectations and stimulating store visits, a maximum discount (vs. a minimum discount such as "starting at 20% off" or a range discount such as "20% off to 60% off") could reduce consumer purchases in many conditions. There is thus a need to balance the outcomes of the two stages and choose the optimal tensile discount under different discount distributions to maximize store sales. In Essay 2, we investigate how digital platforms leverage curation source (algorithm or human) advertising to promote depth or breadth selling and best sequence these two types of advertising to further improve advertising effects. Through a survey study and a field experiment, we find that consumers have a lay belief about the competitive advantages of algorithms and humans that algorithms (humans) generate personalized (novel) recommendations, and importantly, they consider and purchase products similar (new) to their previous preferences when receiving such advertising. Moreover, the complementarity between depth and breadth selling can be best leveraged in sequential advertising by the initial breadth selling amplifying the subsequent depth selling effectiveness. Thus, a carefully-sequenced hybrid advertising can help cultivate consumers’ ever-renewing interests and consumption. / Business Administration/Marketing

Este trabalho aborda as áreas de teoria dos grafos, sistemas de recomendação, e comércio eletrônico, que já foram tema de diversas publicações ao longo das últimas décadas. Entretanto, o estudo da importância da utilização de medidas de centralidade de redes como atributos preditivos de modelos de aprendizado de máquina é um assunto que ainda não foi explorado pela literatura. Neste trabalho, além de relatarmos resultados que sugerem que essas medidas de centralidade podem aumentar a precisão dos modelos preditivos, também apresentamos os principais conceitos teóricos de redes complexas, como tipos de redes, caracterização, métricas de distância, além de propriedades de redes reais. Também apresentamos as ferramentas e metodologia utilizadas para o desenvolvimento de um webcrawler próprio, software necessário para a construção da rede de produtos comprados em conjunto no comércio eletrônico. Modelos de aprendizado de máquina foram treinados utilizando a base de produtos obtida pelo webcrawler, possibilitando a obtenção de modelos preditivos de estimativa de preços de produtos, e de previsão de probabilidade de ligação entre produtos da rede. A performance dos modelos preditivos obtidos são apresentadas. / This work approaches areas such as graph theory, recommendation systems, and electronic commerce, which have been chosen as topics for several publications over the last decades. Although, studying the importance of using network centrality measures as predictive features within machine learning models is a topic which was not yet explored on literature. In this work, besides reporting results which suggest that those centrality measures can increase the precision of predictive models, we also present the main theoretical concepts of complex networks, such as network types, characterization, distance metrics, besides some properties of real networks. We also present the tools and methodology used on the development of our own webcrawler, a software required for the generation of the network of products bought together in the electronic commerce. Machine learning models were trained using the product database obtained using the webcrawler, allowing the achievement of predictive models for product price estimation, and also link prediction between products of the network. The performance of the predictive models are also presented.

The goal of this project is to explore the problem of product recommendations in the area of e-commerce and to evaluate known techniques, design product recommendation system for an existing e-commerce site, implement it and test it. This report introduces the problem, briefly examines current state of affairs in this area and defines requirements for a product recommendation module. The concept of data mining in general is introduced. The report proceeds to present detailed design corresponding to defined requirements and summarizes data gathered during testing phase. It concludes with evaluation and with discussion of the remaining goals for this thesis.