A fundamental challenge in developing high-impact machine learning technologies is balancing the ability to model rich, structured domains with the ability to scale to big data. Many important problem areas are both richly structured and large scale, from social and biological networks, to knowledge graphs and the Web, to images, video, and natural language. In this paper, we introduce two new formalisms for modeling structured data, distinguished from previous approaches by their ability to both capture rich structure and scale to big data. The first, hinge-loss Markov random fields (HL-MRFs), is a new kind of probabilistic graphical model that generalizes different approaches to convex inference. We unite three approaches from the randomized algorithms, probabilistic graphical models, and fuzzy logic communities, showing that all three lead to the same inference objective. We then derive HL-MRFs by generalizing this unified objective. The second new formalism, probabilistic soft logic (PSL), is a probabilistic programming language that makes HL-MRFs easy to define using a syntax based on first-order logic. We next introduce an algorithm for inferring most-probable variable assignments (MAP inference) that is much more scalable than general-purpose convex optimization software, because it uses message passing to take advantage of sparse dependency structures. We then show how to learn the parameters of HL-MRFs. The learned HL-MRFs are as accurate as analogous discrete models, but much more scalable. Together, these algorithms enable HL-MRFs and PSL to model rich, structured data at scales not previously possible.

Abstract In this work, we build an entity/event-level sentiment analysis system, which is able to recognize and infer both explicit and implicit sentiments toward entities and events in the text. We design Probabilistic Soft Logic models that integrate explicit sentiments, inference rules, and +/-effect event information (events that positively or negatively affect entities). The experiments show that the method is able to greatly improve over baseline accuracies in recognizing entity/event-level sentiments.

In contrast to traditional social networks, signed ones encode both relations of affinity and disagreement. Community discovery in this kind of networks has been successfully addressed using the Potts model, originated in statistical me-chanics to explain the magnetic dipole moments of atomic spins. However, due to the computational complexity of finding an exact solution, it has not been ap-plied to many real-world networks yet. We propose a novel approach to compute an approximated solution to the Potts model applied to the context of community discovering, which is based on a continuous convex relaxation of the original prob-lem using hinge-loss functions. We show empirically the benefits of the proposed method in comparison with loopy belief propagation in terms of the communities discovered. We illustrate the scalability and effectiveness of our approach by ap-plying it to the network of voters of the European Parliament that we have crawled for this study. This large-scale and dense network comprises about 300 votings pe-riods on the actual term involving a total of more than 730 voters. Remarkably, the two major communities are those created by the european-antieuropean antag-onism, rather than the classical right-left antagonism. 1