User Profiling Across Multiple Online Social Platform

Online social platforms (OSPs), such as Facebook, Twitter, and Instagram, have grown monumentally over recent years. It was reported that as of August 2017, Facebook has over 2 billion monthly active users, while Instagram and Twitter have over 700 million and 300 million monthly active user accounts, respectively. The vast amount of user-generated content and social data gathered in these behemoth platforms have made them rich data sources for academic and industrial research. However, most of the existing research work has focused on analyzing and modeling user behaviors in a single platform setting, neglecting the inter-dependencies of user behaviors across multiple OSPs. In this project, we design novel techniques that enable the analysis and modeling of user behaviors in multiple OSPs. In particular, we have developed algorithms that (i) link users' profiles across multiple social platforms, (ii) analyze users' topical interests and platform preferences across multiple OSPs, and (iii) model influential users in multiple OSPs.

Linky: A visual analytical tool that extracts the results from different user identity linkage methods performed on multiple online social networks and visualizes the user profiles, content and ego networks of the linked user identities.

Online Misbehavior and Disinformation Mining

Misbehaviors in Social Media

Misbehaviors in social media as such as spreading hate speech are important issues that break the cohesiveness of online social communities and even raises public safety concerns in our societies. Motivated by this rising issue, researchers have developed many traditional machine learning and deep learning methods to detect hate speech in online social platforms automatically. This project aims to design new techniques that enable the analysis and detection of online misbehaviors in online social platforms. Specifically, we are innovating new methods that detect and monitor multilingual and multimodal hateful content on multiple social platforms. For a preliminary understanding of online hate speech detection, please refer to our ASONAM tutorial. We also partner with the Saskatchewan Human Rights Commission in a sub-project to investigate the online xenophobia against Asian communities during the COVID-19 pandemic. This sub-project is funded by the Social Science and Humanities Research Council of Canada Partnership Engage Grants.

Disinformation in Social Media

Due to the prevalence of online social platforms, social media is increasingly being used to influence people in various contexts, from viral marketing to disinformation campaigns. Social disinformation campaigns are especially damaging and can exist in various forms such as hoaxes, propaganda, fake news, etc. This project aims to analyze the disinformation campaigns in social media and develop algorithms that (a) detect disinformation campaigns and (b) track the sources that propagate disinformation. This project is funded by Temasek Laboratories.

Analysis of online xenophobic behaviors amidst the COVID-19 Pandemic in Twitter. We have collected and annotated a dataset with over 40 million COVID-19 related tweets.

Disentangling Hate in Online Memes. We designed DisMultiHate, which is a multimodal model that disentangles the targets in hateful memes for hateful meme classification.

Social Natural Langauge Generation

Natural language generation (NLG) is the task of generating text with the goal of appearing indistinguishable from human-written text, and it is one of the long-term goals of AI research. NLG has a wide range of applications, including text summarization, text simplification, paraphrase, automated dialogue (e.g., chatbots), etc. In this project, we explore the sociolinguistic aspects of text generation. For instance, we design algorithms that can modify the communication styles of generated text in different social settings. For a further look at the related works in this field, please refer to our survey, Text Style Transfer: A Review and Experimental Evaluation. Preliminary research in this area is funded by Living Sky Technologies.

AutoChart: A multimodal text generation model that recognizes charts and generates a text analysis. The above analysis is generated by AutoChart automatically.

Social Recommender Systems

The proliferation of online social platforms (OSPs) has changed the way we lead our lives and conduct businesses. Popular OSPs such as Facebook already has more than 2 billion active users worldwide. This motivates businesses and governments to engage this massive number of users online. However, the wide adoption of OSPs has overwhelmed users with information. To mitigate this burden for users, OSPs have designed personalized social recommender systems to filter content, activities, and relationships based on the users’ needs and interests. In this project, we aim to explore the factors in creating the next generation of social recommender systems that improve user experience and trust in online social platforms. Working towards this goal, we design new social recommender systems with the following objectives: 1) improving personalization, 2) providing explanations, and 3) strengthening privacy preservation in social recommender systems. The project is funded through the National Sciences and Engineering Research Council of Canada (NSERC) Discovery Grant (title: "Next-Generation Social Recommender Systems: Personalized, Explainable, and Privacy-Preserving").