Building machines that can reason about physical events and their causal relationships is crucial for flexible interaction with the physical world. However, most existing physical and causal reasoning benchmarks are exclusively based on synthetically generated events and synthetic natural language descriptions of the causal relationships. This design brings up two issues. First, there is a lack of diversity in both event types and natural language descriptions; second, causal relationships based on manually-defined heuristics are different from human judgments. To address both shortcomings, we present the CLEVRER-Humans benchmark, a video reasoning dataset for causal judgment of physical events with human labels. We employ two techniques to improve data collection efficiency: first, a novel iterative event cloze task to elicit a new representation of events in videos, which we term Causal Event Graphs (CEGs); second, a data augmentation technique based on neural language generative models. We convert the collected CEGs into questions and answers to be consistent with prior work. Finally, we study a collection of baseline approaches for CLEVRER-Humans question-answering, highlighting great challenges set forth by our benchmark.

Pretraining a language model (LM) on text has been shown to help various downstream NLP tasks. Recent works show that a knowledge graph (KG) can complement text data, offering structured background knowledge that provides a useful scaffold for reasoning. However, these works are not pretrained to learn a deep fusion of the two modalities at scale, limiting the potential to acquire fully joint representations of text and KG. Here we propose DRAGON (Deep Bidirectional Language-Knowledge Graph Pretraining), a self-supervised approach to pretraining a deeply joint language-knowledge foundation model from text and KG at scale. Specifically, our model takes pairs of text segments and relevant KG subgraphs as input and bidirectionally fuses information from both modalities. We pretrain this model by unifying two self-supervised reasoning tasks, masked language modeling and KG link prediction. DRAGON outperforms existing LM and LM+KG models on diverse downstream tasks including question answering across general and biomedical domains, with +5% absolute gain on average. In particular, DRAGON achieves notable performance on complex reasoning about language and knowledge (+10% on questions involving long contexts or multi-step reasoning) and low-resource QA (+8% on OBQA and RiddleSense), and new state-of-the-art results on various BioNLP tasks. Our code and trained models are available at https://github.com/michiyasunaga/dragon.

Modeling sequential interactions between users and items/products is crucial in domains such as e-commerce, social networking, and education. Representation learning presents an attractive opportunity to model the dynamic evolution of users and items, where each user/item can be embedded in a Euclidean space and its evolution can be modeled by an embedding trajectory in this space. However, existing dynamic embedding methods generate embeddings only when users take actions and do not explicitly model the future trajectory of the user/item in the embedding space. Here we propose JODIE, a coupled recurrent neural network model that learns the embedding trajectories of users and items. JODIE employs two recurrent neural networks to update the embedding of a user and an item at every interaction. Crucially, JODIE also models the future embedding trajectory of a user/item. To this end, it introduces a novel projection operator that learns to estimate the embedding of the user at any time in the future. These estimated embeddings are then used to predict future user-item interactions. To make the method scalable, we develop a t-Batch algorithm that creates time-consistent batches and leads to 9x faster training. We conduct six experiments to validate JODIE on two prediction tasks---future interaction prediction and state change prediction---using four real-world datasets. We show that JODIE outperforms six state-of-the-art algorithms in these tasks by at least 20% in predicting future interactions and 12% in state change prediction.

Pattern-based methods have been successful in information extraction and NLP research. Previous approaches learn the quality of a textual pattern as relatedness to a certain task based on statistics of its individual content (e.g., length, frequency) and hundreds of carefully-annotated labels. However, patterns of good content-quality may generate heavily conflicting information due to the big gap between relatedness and correctness. Evaluating the correctness of information is critical in (entity, attribute, value)-tuple extraction. In this work, we propose a novel method, called TruePIE, that finds reliable patterns which can extract not only related but also correct information. TruePIE adopts the self-training framework and repeats the training-predicting-extracting process to gradually discover more and more reliable patterns. To better represent the textual patterns, pattern embeddings are formulated so that patterns with similar semantic meanings are embedded closely to each other. The embeddings jointly consider the local pattern information and the distributional information of the extractions. To conquer the challenge of lacking supervision on patterns' reliability, TruePIE can automatically generate high quality training patterns based on a couple of seed patterns by applying the arity-constraints to distinguish highly reliable patterns (i.e., positive patterns) and highly unreliable patterns (i.e., negative patterns). Experiments on a huge news dataset (over 25GB) demonstrate that the proposed TruePIE significantly outperforms baseline methods on each of the three tasks: reliable tuple extraction, reliable pattern extraction, and negative pattern extraction.