Luke Zettlemoyer

Open question answering over curated and extracted knowledge bases

We consider the problem of open-domain question answering (Open QA) over massive knowledge bases (KBs). Existing approaches use either manually curated KBs like Freebase or KBs automatically extracted from unstructured text. In this paper, we present OQA, the first approach to leverage both curated and extracted KBs. A key technical challenge is designing systems that are robust to the high variability in both natural language questions and massive KBs. OQA achieves robustness by decomposing the full Open QA problem into smaller sub-problems including question paraphrasing and query reformulation. OQA solves these sub-problems by mining millions of rules from an unlabeled question corpus and across multiple KBs. OQA then learns to integrate these rules by performing discriminative training on question-answer pairs using a latent-variable structured perceptron algorithm. We evaluate OQA on three benchmark question sets and demonstrate that it achieves up to twice the precision and recall of a state-of-the-art Open QA system.

Reinforcement Learning for Mapping Instructions to Actions

In this paper, we present a reinforcement learning approach for mapping natural language instructions to sequences of executable actions. We assume access to a reward function that defines the quality of the executed actions. During training, the learner repeatedly constructs action sequences for a set of documents, executes those actions, and observes the resulting reward. We use a policy gradient algorithm to estimate the parameters of a log-linear model for action selection. We apply our method to interpret instructions in two domains --- Windows troubleshooting guides and game tutorials. Our results demonstrate that this method can rival supervised learning techniques while requiring few or no annotated training examples.

Learning to Parse Natural Language Commands to a Robot Control System

As robots become more ubiquitous and capable of performing complex tasks, the importance of enabling untrained users to interact with them has increased. In response, unconstrained natural-language interaction with robots has emerged as a significant research area. We discuss the problem of parsing natural language commands to actions and control structures that can be readily implemented in a robot execution system. Our approach learns a parser based on example pairs of English commands and corresponding control language expressions. We evaluate this approach in the context of following route instructions through an indoor environment, and demonstrate that our system can learn to translate English commands into sequences of desired actions, while correctly capturing the semantic intent of statements involving complex control structures. The procedural nature of our formal representation allows a robot to interpret route instructions online while moving through a previously unknown environment.

Learning symbolic models of stochastic domains

In this article, we work towards the goal of developing agents that can learn to act in complex worlds. We develop a probabilistic, relational planning rule representation that compactly models noisy, nondeterministic action effects, and show how such rules can be effectively learned. Through experiments in simple planning domains and a 3D simulated blocks world with realistic physics, we demonstrate that this learning algorithm allows agents to effectively model world dynamics.

Learning Context-Dependent Mappings from Sentences to Logical Form

We consider the problem of learning context-dependent mappings from sentences to logical form. The training examples are sequences of sentences annotated with lambda-calculus meaning representations. We develop an algorithm that maintains explicit, lambda-calculus representations of salient discourse entities and uses a context-dependent analysis pipeline to recover logical forms. The method uses a hidden-variable variant of the perception algorithm to learn a linear model used to select the best analysis. Experiments on context-dependent utterances from the ATIS corpus show that the method recovers fully correct logical forms with 83.7% accuracy.

A Generative Model for Parsing Natural Language to Meaning Representations

In this paper, we present an algorithm for learning a generative model of natural language sentences together with their formal meaning representations with hierarchical structures. The model is applied to the task of mapping sentences to hierarchical representations of their underlying meaning. We introduce dynamic programming techniques for efficient training and decoding. In experiments, we demonstrate that the model, when coupled with a discriminative reranking technique, achieves state-of-the-art performance when tested on two publicly available corpora. The generative model degrades robustly when presented with instances that are different from those seen in training. This allows a notable improvement in recall compared to previous models.

Learning to Automatically Solve Algebra Word Problems

We present an approach for automatically learning to solve algebra word problems. Our algorithm reasons across sentence boundaries to construct and solve a system of linear equations, while simultaneously recovering an alignment of the variables and numbers in these equations to the problem text. The learning algorithm uses varied supervision, including either full equations or just the final answers. We evaluate performance on a newly gathered corpus of algebra word problems, demonstrating that the system can correctly answer almost 70% of the questions in the dataset. This is, to our knowledge, the first learning result for this task.

TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehension

We present TriviaQA, a challenging reading comprehension dataset containing over 650K question-answer-evidence triples. TriviaQA includes 95K question-answer pairs authored by trivia enthusiasts and independently gathered evidence documents, six per question on average, that provide high quality distant supervision for answering the questions. We show that, in comparison to other recently introduced large-scale datasets, TriviaQA (1) has relatively complex, compositional questions, (2) has considerable syntactic and lexical variability between questions and corresponding answer-evidence sentences, and (3) requires more cross sentence reasoning to find answers. We also present two baseline algorithms: a feature-based classifier and a state-of-the-art neural network, that performs well on SQuAD reading comprehension. Neither approach comes close to human performance (23% and 40% vs. 80%), suggesting that TriviaQA is a challenging testbed that is worth significant future study. Data and code available at -- this http URL

Broad-coverage CCG Semantic Parsing with AMR

We propose a grammar induction technique for AMR semantic parsing. While previous grammar induction techniques were designed to re-learn a new parser for each target application, the recently annotated AMR Bank provides a unique opportunity to induce a single model for understanding broad-coverage newswire text and support a wide range of applications. We present a new model that combines CCG parsing to recover compositional aspects of meaning and a factor graph to model non-compositional phenomena, such as anaphoric dependencies. Our approach achieves 66.2 Smatch F1 score on the AMR bank, significantly outperforming the previous state of the art.

Learning from unscripted deictic gesture and language for human-robot interactions

As robots become more ubiquitous, it is increasingly important for untrained users to be able to interact with them intuitively. In this work, we investigate how people refer to objects in the world during relatively unstructured communication with robots. We collect a corpus of deictic interactions from users describing objects, which we use to train language and gesture models that allow our robot to determine what objects are being indicated. We introduce a temporal extension to state-of-the-art hierarchical matching pursuit features to support gesture understanding, and demonstrate that combining multiple communication modalities more effectively capture user intent than relying on a single type of input. Finally, we present initial interactions with a robot that uses the learned models to follow commands.