On Learning Interpreted Languages with Recurrent Models

Can recurrent neural nets, inspired by human sequential data processing, learn to understand language? We construct simplified datasets reflecting core properties of natural language as modeled in formal syntax and semantics: recursive syntactic structure and compositionality. We find LSTM and GRU networks to generalise to compositional interpretation well, but only in the most favorable learning settings, with a well-paced curriculum, extensive training data, and left-to-right (but not right-to-left) composition.

Standpoint Logic: Multi-Perspective Knowledge Representation

Ontologies and knowledge bases encode, to a certain extent, the standpoints or perspectives of their creators. As differences and conflicts between standpoints should be expected in multi-agent scenarios, this will pose challenges for shared creation and usage of knowledge sources. Our work pursues the idea that, in some cases, a framework that can handle diverse and possibly conflicting standpoints is more useful and versatile than forcing their unification, and avoids common compromises required for their merge. Moreover, in analogy to the notion of family resemblance concepts, we propose that a collection of standpoints can provide a simpler yet more faithful and nuanced representation of some domains. To this end, we present standpoint logic, a multi-modal framework that is suitable for expressing information with semantically heterogeneous vocabularies, where a standpoint is a partial and acceptable interpretation of the domain. Standpoints can be organised hierarchically and combined, and complex correspondences can be established between them. We provide a formal syntax and semantics, outline the complexity for the propositional case, and explore the representational capacities of the framework in relation to standard techniques in ontology integration, with some examples in the Bio-Ontology domain.

Formal Syntax and Deep History

We show that, contrary to long-standing assumptions, syntactic traits, modeled here within the generative biolinguistic framework, provide insights into deep-time language history. To support this claim, we have encoded the diversity of nominal structures using 94 universally definable binary parameters, set in 69 languages spanning across up to 13 traditionally irreducible Eurasian families. We found a phylogenetic signal that distinguishes all such families and matches the family-internal tree topologies that are safely established through classical etymological methods and datasets. We have retrieved “near-perfect” phylogenies, which are essentially immune to homoplastic disruption and only moderately influenced by horizontal convergence, two factors that instead severely affect more externalized linguistic features, like sound inventories. This result allows us to draw some preliminary inferences about plausible/implausible cross-family classifications; it also provides a new source of evidence for testing the representation of diversity in syntactic theories.

Lolisa: Formal Syntax and Semantics for a Subset of the Solidity Programming Language in Mathematical Tool Coq

The security of blockchain smart contracts is one of the most emerging issues of the greatest interest for researchers. This article presents an intermediate specification language for the formal verification of Ethereum-based smart contract in Coq, denoted as Lolisa. The formal syntax and semantics of Lolisa contain a large subset of the Solidity programming language developed for the Ethereum blockchain platform. To enhance type safety, the formal syntax of Lolisa adopts a stronger static type system than Solidity. In addition, Lolisa includes a large subset of Solidity syntax components as well as general-purpose programming language features. Therefore, Solidity programs can be directly translated into Lolisa with line-by-line correspondence. Lolisa is inherently generalizable and can be extended to express other programming languages. Finally, the syntax and semantics of Lolisa have been encapsulated as an interpreter in mathematical tool Coq. Hence, smart contracts written in Lolisa can be symbolically executed and verified in Coq.

Recursion in SPARQL

The need for recursive queries in the Semantic Web setting is becoming more and more apparent with the emergence of datasets where different pieces of information are connected by complicated patterns. This was acknowledged by the W3C committee by the inclusion of property paths in the SPARQL standard. However, as more data becomes available, it is becoming clear that property paths alone are not enough to capture all recursive queries that the users are interested in, and the literature has already proposed several extensions to allow searching for more complex patterns. We propose a rather different, but simpler approach: add a general purpose recursion operator directly to SPARQL. In this paper we provide a formal syntax and semantics for this proposal, study its theoretical properties, and develop algorithms for evaluating it in practical scenarios. We also show how to implement this extension as a plug-in on top of existing systems, and test its performance on several synthetic and real world datasets, ranging from small graphs, up to the entire Wikidata database.

Deep Reinforcement Learning for Syntactic Error Repair in Student Programs

Novice programmers often struggle with the formal syntax of programming languages. In the traditional classroom setting, they can make progress with the help of real time feedback from their instructors which is often impossible to get in the massive open online course (MOOC) setting. Syntactic error repair techniques have huge potential to assist them at scale. Towards this, we design a novel programming language correction framework amenable to reinforcement learning. The framework allows an agent to mimic human actions for text navigation and editing. We demonstrate that the agent can be trained through self-exploration directly from the raw input, that is, program text itself, without either supervision or any prior knowledge of the formal syntax of the programming language. We evaluate our technique on a publicly available dataset containing 6975 erroneous C programs with typographic errors, written by students during an introductory programming course. Our technique fixes 1699 (24.4%) programs completely and 1310 (18.8%) program partially, outperforming DeepFix, a state-of-the-art syntactic error repair technique, which uses a fully supervised neural machine translation approach.

Strange yet Familiar Sounds in Zosha Di Castri’s String Quartet No. 1

This paper presents a detailed analysis of Zosha Di Castri’s String Quartet No. 1 (2016), situating the work in relation to the composer’s still-nascent oeuvre and showing how it contains a diverse mix of stylistic impulses within its relatively compact form. The concept of defamiliarization, borrowed from literary criticism, provides a useful theoretical basis for understanding the work. By presenting familiar musical figures within strange new contexts, Di Castri effectively bypasses habitual modes of reception and encourages the listener to engage more actively with the work’s multi-coded discourse. This revivification of perceptual awareness is accomplished using a number of compositional techniques, including the juxtaposition or superimposition of contrasting materials, the transformation of motives, and the distortion of formal syntax. Taken together, these defamiliarizing devices mount an aesthetic intervention that playfully subverts compositional norms, allowing Di Castri to take a decisive step forward in the development of her own musical style.

Interview with Noam Chomsky

<p>Avram Noam Chomsky is a world-renowned linguist, philosopher and political activist. He is Professor Emeritus in the Department of Linguistics at Massachusetts Institute of Technology and recently became a laureate Professor in the Department of Linguistics at University of Arizona. He has been the leader of the generative enterprise in linguistic theory since its beginning, in the late fifties, and is considered by many the father of modern Linguistics. In fact, his views have influenced the whole field and established points of departure for research in formal syntax, phonology and even semantics.</p><hr /><p><strong>ENTREVISTA COM NOAM CHOMSKY</strong></p><p>Avram Noam Chomsky é um renomado linguista, filósofo e ativista político. Ele é professor emérito do Departamento de Linguística do Instituto de Tecnologia de Massachusetts e recentemente se tornou professor laureado no Departamento de Linguística da Universidade do Arizona. Ele tem sido o líder do empreendimento gerativo na teoria linguística desde o seu início, no final dos anos 50, e é considerado por muitos o pai da linguística moderna. De fato, suas visões influenciaram todo o campo e estabeleceram pontos de partida para a pesquisa em sintaxe formal, fonologia e até mesmo semântica.</p><hr />