The advent of social media platforms has caused many changes in humans’ daily lifestyle. One of the most significant changes is the way in which people participate in social and cultural events. Users' participation in social media platforms is continuously increasing. This has provided brands with new opportunities such as enhancing brand influence and understanding online users' reactions through user-generated content (UGC) analysis. We provide and describe a large-scale hashtag-based dataset of social media posts published on Instagram about the Big Four international fashion weeks in New York, Paris, Milan, and London. The dataset provides the data of the 2018 events and has a periodic and well-established structure. Moreover, we designed a two-stage platform for collecting such large-scale datasets related to long-running events based on relevant hashtags: In the first stage, the platform extracts all the posts, and in the second stage, it extracts the information about the authors of the posts.

Social media platforms offer their audience the possibility to reply to posts through comments and reactions. This allows social media users to express their ideas and opinions on shared content, thus opening virtual discussions. Most studies on social networks have focused only on user relationships or on the shared content, while ignoring the valuable information hidden in the digital conversations, in terms of structure of the discussion and relation between contents, which is essential for understanding online communication behavior. This work proposes a graph-based framework to assess the shape and structure of online conversations. The analysis was composed of two main stages: intent analysis and network generation. Users’ intention was detected using keyword-based classification, followed by the implementation of machine learning-based classification algorithms for uncategorized comments. Afterwards, human-in-the-loop was involved in improving the keyword-based classification. To extract essential information on social media communication patterns among the users, we built conversation graphs using a directed multigraph network and we show our model at work in two real-life experiments. The first experiment used data from a real social media challenge and it was able to categorize 90% of comments with 98% accuracy. The second experiment focused on COVID vaccine-related discussions in online forums and investigated the stance and sentiment to understand how the comments are affected by their parent discussion. Finally, the most popular online discussion patterns were mined and interpreted. We see that the dynamics obtained from conversation graphs are similar to traditional communication activities.

One of the travelers’ main challenges is that they have to spend a great effort to find and choose the most desired travel offer(s) among a vast list of non-categorized and non-personalized items. Recommendation systems provide an effective way to solve the problem of information overload. In this work, we design and implement “The Hybrid Offer Ranker” (THOR), a hybrid, personalized recommender system for the transportation domain. THOR assigns every traveler a unique contextual preference model built using solely their personal data, which makes the model sensitive to the user’s choices. This model is used to rank travel offers presented to each user according to their personal preferences. We reduce the recommendation problem to one of binary classification that predicts the probability with which the traveler will buy each available travel offer. Travel offers are ranked according to the computed probabilities, hence to the user’s personal preference model. Moreover, to tackle the cold start problem for new users, we apply clustering algorithms to identify groups of travelers with similar profiles and build a preference model for each group. To test the system’s performance, we generate a dataset according to some carefully designed rules. The results of the experiments show that the THOR tool is capable of learning the contextual preferences of each traveler and ranks offers starting from those that have the higher probability of being selected.

In the last few years, social media has dominated various aspects of people’s life including social events. Users participate more and more in long-running periodical events in social media, by sharing their experiences and preferences. This information provides unprecedented opportunities allowing businesses to promote their brands coverage by using word-of-mouth (WOM), that is enabled by the user generated contents (UGCs). Studying social media content popularity by considering the societies’ behavioral patterns is, therefore, paramount. In this thesis, we inspect users’ engagement motives in long-running events by means of a comprehensive statistical analysis of fashion week events on Instagram. Additionally, we develop a multi-modal approach to solve the problem of post popularity prediction that exploits potentially influential factors and apply it on fashion week events. We employ two metrics for implementing a filter feature selection technique, together with an automated grid search for optimizing hyper-parameters in four regression methods: ridge, support vector regressor, gradient tree boosting and neural networks.