Compositional question answering begins by mapping questions to logical forms, but training a semantic parser to perform this mapping typically requires the costly annotation of the target logical forms. In this paper, we learn to map questions to answers via latent logical forms, which are induced automatically from question-answer pairs. In tackling this challenging learning problem, we introduce a new semantic representation which highlights a parallel between dependency syntax and efficient evaluation of logical forms. On two standard semantic parsing benchmarks (GEO and JOBS), our system obtains the highest published accuracies, despite requiring no annotated logical forms. 1

We consider the problem of learning factored probabilistic CCG grammars for semantic parsing from data containing sentences paired with logical-form meaning representations. Traditional CCG lexicons list lexical items that pair words and phrases with syntactic and semantic content. Such lexicons can be inefficient when words appear repeatedly with closely related lexical content. In this paper, we introduce factored lexicons, which include both lexemes to model word meaning and templates to model systematic variation in word usage. We also present an algorithm for learning factored CCG lexicons, along with a probabilistic parse-selection model. Evaluations on benchmark datasets demonstrate that the approach learns highly accurate parsers, whose generalization performance benefits greatly from the lexical factoring. 1

This is a response to Regina Barzilay's talk. Although it is hard for computer to do what human do in the physical world (e.g. play soccer, or cook dinners) because of the problem of perception and control, it is much easier for them to accomplish tasks in the cyber space. Here we will propose a personal assistant that helps users accomplish tasks they can do with a computer. Specifically, we will focus on information search and management tasks. In order to allow assistants to exhibit complex behavior, we further introduce a type of reinforcement learner called relation machines. 1.

Conversations provide rich opportunities for interactive, continuous learning. When something goes wrong, a system can ask for clarification, rewording, or otherwise redirect the interaction to achieve its goals. In this paper, we present an approach for using conversational interactions of this type to induce semantic parsers. We demonstrate learning without any explicit annotation of the meanings of user utterances. Instead, we model meaning with latent variables, and introduce a loss function to measure how well potential meanings match the conversation. This loss drives the overall learning approach, which induces a weighted CCG grammar that could be used to automatically bootstrap the semantic analysis component in a complete dialog system. Experiments on DARPA Communicator conversational logs demonstrate effective learning, despite requiring no explicit meaning annotations. 1

Automated understanding of natural language is a challenging problem, which has remained open for decades. We have investigated its special case, focused on identifying relevant concepts in natural-language text in the context of a specific given task. We have developed a set of general-purpose language interpretation techniques and applied them to the task of detecting malicious websites by analyzing comments of website visitors. In this context, concepts are related to behavior or contents of websites, such as presence of pop-ups and false testimonials. The developed algorithms are based on probabilistic topic models and other dimensionality reduction techniques applied to a special case of multi-label text classification, where concepts are output labels. We integrate information about the target task with other relevant information, including relations among concepts and external knowledge sources using a concept graph. The system iterates between training a topic model on the partially labeled data and optimizing the parameters and the label assignments. We analyze several alternative versions of this mechanism, such as one that measures the quality of separation among topics and eliminates words that are not discriminative. For the task of detecting malicious websites, we have developed an approach that applies

Many first-order probabilistic models can be represented much more compactly using aggregation operations such as counting. While traditional statistical relational representations share factors across sets of interchangeable random variables, representations that explicitly model aggregations also exploit interchangeability of random variables within factors. This is especially useful in decision making settings, where an agent might need to reason about counts of the different types of objects it interacts with. Previous work on counting formulas in statistical relational representations has mostly focused on the problem of exact inference on an existing model. The problem of learning such models is largely unexplored. In this pa-per,weintroduceCountingMarkovLogicNetworks(C-MLNs),anextensionofMarkovlogicnetworksthatcan compactly represent complex counting formulas. We present a structure learning algorithm for C-MLNs; we apply this algorithm to the novel problem of generalizingnaturallanguageinstructions,andtorelationalreinforcement learning in the Crossblock domain, in which standard MLN learning algorithms fail to find any useful structure. The C-MLN policies learned from natural language instructions are compact and intuitive, and,despiterequiringnoinstructionsontestgames,win 20 % more Crossblock games than a state-of-the-art algorithmfor followingnatural language instructions.