We propose Grounded Adaptation for Zero-shot Executable Semantic Parsing (GAZP) to adapt an existing semantic parser to new environments (e.g. new database schemas). GAZP combines a forward semantic parser with a backward utterance generator to synthesize data (e.g. utterances and SQL queries) in the new environment, then selects cycle-consistent examples to adapt the parser. Unlike data-augmentation, which typically synthesizes unverified examples in the training environment, GAZP synthesizes examples in the new environment whose input-output consistency are verified. On the Spider, Sparc, and CoSQL zero-shot semantic parsing tasks, GAZP improves logical form and execution accuracy of the baseline parser. Our analyses show that GAZP outperforms data-augmentation in the training environment, performance increases with the amount of GAZP-synthesized data, and cycle-consistency is central to successful adaptation.

We investigate the problem of generating natural language summaries from knowledge base triples. Our approach is based on a pointer-generator network, which, in addition to generating regular words from a fixed target vocabulary, is able to verbalise triples in several ways. We undertake an automatic and a human evaluation on single and open-domain summaries generation tasks. Both show that our approach significantly outperforms other data-driven baselines.

Dependency parsing is a longstanding natural language processing task, with its outputs crucial to various downstream tasks. Recently, neural network based (NN-based) dependency parsing has achieved significant progress and obtained the state-of-the-art results. As we all know, NN-based approaches require massive amounts of labeled training data, which is very expensive because it requires human annotation by experts. Thus few industrial-oriented dependency parser tools are publicly available. In this report, we present Baidu Dependency Parser (DDParser), a new Chinese dependency parser trained on a large-scale manually labeled dataset called Baidu Chinese Treebank (DuCTB). DuCTB consists of about one million annotated sentences from multiple sources including search logs, Chinese newswire, various forum discourses, and conversation programs. DDParser is extended on the graph-based biaffine parser to accommodate to the characteristics of Chinese dataset. We conduct experiments on two test sets: the standard test set with the same distribution as the training set and the random test set sampled from other sources, and the labeled attachment scores (LAS) of them are 92.9% and 86.9% respectively. DDParser achieves the state-of-the-art results, and is released at https://github.com/baidu/DDParser.

A common approach to solve complex tasks is by breaking them down into simple sub-problems that can then be solved by simpler modules. However, these approaches often need to be designed and trained specifically for each complex task. We propose a general approach, Text Modular Networks(TMNs), where the system learns to decompose any complex task into the language of existing models. Specifically, we focus on Question Answering (QA) and learn to decompose complex questions into sub-questions answerable by existing QA models. TMNs treat these models as blackboxes and learn their textual input-output behavior (i.e., their language) through their task datasets. Our next-question generator then learns to sequentially produce sub-questions that help answer a given complex question. These sub-questions are posed to different existing QA models and, together with their answers, provide a natural language explanation of the exact reasoning used by the model. We present the first system, incorporating a neural factoid QA model and a symbolic calculator, that uses decomposition for the DROP dataset, while also generalizing to the multi-hop HotpotQA dataset. Our system, ModularQA, outperforms a cross-task baseline by 10-60 F1 points and performs comparable to task-specific systems, while also providing an easy-to-read explanation of its reasoning.

As the number of robots in our daily surroundings like home, office, restaurants, factory floors, etc. are increasing rapidly, the development of natural human-robot interaction mechanism becomes more vital as it dictates the usability and acceptability of the robots. One of the valued features of such a cohabitant robot is that it performs tasks that are instructed in natural language. However, it is not trivial to execute the human intended tasks as natural language expressions can have large linguistic variations. Existing works assume either single task instruction is given to the robot at a time or there are multiple independent tasks in an instruction. However, complex task instructions composed of multiple inter-dependent tasks are not handled efficiently in the literature. There can be ordering dependency among the tasks, i.e., the tasks have to be executed in a certain order or there can be execution dependency, i.e., input parameter or execution of a task depends on the outcome of another task. Understanding such dependencies in a complex instruction is not trivial if an unconstrained natural language is allowed. In this work, we propose a method to find the intended order of execution of multiple inter-dependent tasks given in natural language instruction. Based on our experiment, we show that our system is very accurate in generating a viable execution plan from a complex instruction.

An institutional statement describes expected actions for actors within the presence or absence of particular constraints, or parameterizes features of an institutional system. Institutional statements convey information that contextualizes their applicability. They vary in prescriptiveness and force, as reflected by the presence of information that more or less strongly compels behavior and by the presence of information that specifies payoffs for compliance, or noncompliance, with statements instructions. Varying in the inclusion of these various kinds of information, institutional statements typically take two functional forms: constitutive and regulative. Constitutive statements constitute features of a system (e.g., actor positions and roles, processes, venues, etc). Regulative statements describe actions linked to specific actors within certain contextual parameters. According to the IG 2.0, institutional statements are commonly comprised of a set of syntactic components, with individual components associating with unique information, and which combine to convey a statement's institutional meaning. Regulative statements are composed of some or all of the following components with the corresponding syntactic labels: (i) an Actor, referred to as an Attribute; (ii) action associated with actor, referred to as an Aim; (iii) action context, referred to as Context; (iv) a receiver of action, referred to as an Object; (v) a prescriptive operator that describes how strongly an action is compelled or restrained, referred to as a Deontic; and (vi) an incentive linked to action, referred to as an Or else. Constitutive statements are composed of some or all of the following components with the corresponding syntactic labels: (i) the entity that is being constituted within a statement, referred to as a Constituted Entity; (ii) an action that constitutes the Constituted Entity, called the Constitutive Function; (iii) the constitution context, referred to as Context; (iv) properties that serve as input to the Constitutive Function, called Constituting Properties; (iv) A prescriptive operator that defines to what extent the action of an institutional statement is compelled, restrained, or discretionary, referred to as a Deontic; and (vi) an incentive linked to action, referred to as an Or else.

Despite achieving tremendous success, existing deep learning models have exposed limitations in compositional generalization, the capability to learn compositional rules and apply them to unseen cases in a systematic manner. To tackle this issue, we propose the Neural-Symbolic Stack Machine (NeSS). It contains a neural network to generate traces, which are then executed by a symbolic stack machine enhanced with sequence manipulation operations. NeSS combines the expressive power of neural sequence models with the recursion supported by the symbolic stack machine. Without training supervision on execution traces, NeSS achieves 100% generalization performance in three domains: the SCAN benchmark of language-driven navigation tasks, the compositional machine translation benchmark, and context-free grammar parsing tasks.

The machinery of the human brain - analog, probabilistic, embodied - can be characterized computationally, but what machinery confers what computational powers? Any such system can be abstractly cast in terms of two computational components: a finite state machine carrying out computational steps, whether via currents, chemistry, or mechanics; plus a set of allowable memory operations, typically formulated in terms of an information store that can be read from and written to, whether via synaptic change, state transition, or recurrent activity. Probing these mechanisms for their information content, we can capture the difference in computational power that various systems are capable of. Most human cognitive abilities, from perception to action to memory, are shared with other species; we seek to characterize those (few) capabilities that are ubiquitously present among humans and absent from other species. Three realms of formidable constraints --- a) measurable human cognitive abilities, b) measurable allometric anatomic brain characteristics, and c) measurable features of specific automata and formal grammars --- illustrate remarkably sharp restrictions on human abilities, unexpectedly confining human cognition to a specific class of automata ("nested stack"), which are markedly below Turing machines.

Humans are remarkably flexible when understanding new sentences that include combinations of concepts they have never encountered before. Recent work has shown that while deep networks can mimic some human language abilities when presented with novel sentences, systematic variation uncovers the limitations in the language-understanding abilities of neural networks. We demonstrate that these limitations can be overcome by addressing the generalization challenges in a recently-released dataset, gSCAN, which explicitly measures how well a robotic agent is able to interpret novel ideas grounded in vision, e.g., novel pairings of adjectives and nouns. The key principle we employ is compositionality: that the compositional structure of networks should reflect the compositional structure of the problem domain they address, while allowing all other parameters and properties to be learned end-to-end with weak supervision. We build a general-purpose mechanism that enables robots to generalize their language understanding to compositional domains. Crucially, our base network has the same state-of-the-art performance as prior work, 97% execution accuracy, while at the same time generalizing its knowledge when prior work does not; for example, achieving 95% accuracy on novel adjective-noun compositions where previous work has 55% average accuracy. Robust language understanding without dramatic failures and without corner causes is critical to building safe and fair robots; we demonstrate the significant role that compositionality can play in achieving that goal.

Knowledge of languages is the doorway to wisdom. I was amazed that Roger Bacon gave the above quote in the 13th century, and it still holds, Isn't it? I am sure that you all will agree with me. Today, the way of understanding languages has changed a lot from the 13th century. We now refer to it as linguistics and natural language processing.