Human conversations naturally evolve around different topics and fluently move between them. In research on dialog systems, the ability to actively and smoothly transition to new topics is often ignored. In this paper we introduce TIAGE, a new topic-shift aware dialog benchmark constructed utilizing human annotations on topic shifts. Based on TIAGE, we introduce three tasks to investigate different scenarios of topic-shift modeling in dialog settings: topic-shift detection, topic-shift triggered response generation and topic-aware dialog generation. Experiments on these tasks show that the topic-shift signals in TIAGE are useful for topic-shift response generation. On the other hand, dialog systems still struggle to decide when to change topic. This indicates further research is needed in topic-shift aware dialog modeling.

We propose a novel morphologically aware probability model for bilingual lexicon induction, which jointly models lexeme translation and inflectional morphology in a structured way. Our model exploits the basic linguistic intuition that the lexeme is the key lexical unit of meaning, while inflectional morphology provides additional syntactic information. This approach leads to substantial performance improvements—19% average improvement in accuracy across 6 language pairs over the state of the art in the supervised setting and 16% in the weakly supervised setting. As another contribution, we highlight issues associated with modern BLI that stem from ignoring inflectional morphology, and propose three suggestions for improving the task.

Distributional Semantic Models (DSMs) construct vector representations of word meanings based on their contexts. Typically, the contexts of a word are defined as its closest neighbours, but they can also be retrieved from its syntactic dependency relations. In this work, we propose a new dependency-based DSM. The novelty of our model lies in associating an independent meaning representation, a matrix, with each dependency-label. This allows it to capture specifics of the relations between words and contexts, leading to good performance on both intrinsic and extrinsic evaluation tasks. In addition to that, our model has an inherent ability to represent dependency chains as products of matrices which provides a straightforward way of handling further contexts of a word.

The correct interpretation of quantifier statements in the context of a visual scene requires non-trivial inference mechanisms. For the example of “most”, we discuss two strategies which rely on fundamentally different cognitive concepts. Our aim is to identify what strategy deep learning models for visual question answering learn when trained on such questions. To this end, we carefully design data to replicate experiments from psycholinguistics where the same question was investigated for humans. Focusing on the FiLM visual question answering model, our experiments indicate that a form of approximate number system emerges whose performance declines with more difficult scenes as predicted by Weber’s law. Moreover, we identify confounding factors, like spatial arrangement of the scene, which impede the effectiveness of this system.

Human translators routinely have to translate rare inflections of words – due to the Zipfian distribution of words in a language. When translating from Spanish, a good translator would have no problem identifying the proper translation of a statistically rare inflection such as habláramos. Note the lexeme itself, hablar, is relatively common. In this work, we investigate whether state-of-the-art bilingual lexicon inducers are capable of learning this kind of generalization. We introduce 40 morphologically complete dictionaries in 10 languages and evaluate three of the best performing models on the task of translation of less frequent morphological forms. We demonstrate that the performance of state-of-the-art models drops considerably when evaluated on infrequent morphological inflections and then show that adding a simple morphological constraint at training time improves the performance, proving that the bilingual lexicon inducers can benefit from better encoding of morphology.

We discuss problems with the standard approaches to evaluation for tasks like visual question answering, and argue that artificial data can be used to address these as a complement to current practice. We demonstrate that with the help of existing ‘deep’ linguistic processing technology we are able to create challenging abstract datasets, which enable us to investigate the language understanding abilities of multimodal deep learning models in detail, as compared to a single performance value on a static and monolithic dataset.

We propose sentence chunking as a way to reduce the time and memory costs of realization of long sentences. During chunking we divide the semantic representation of a sentence into smaller components which can be processed and recombined without loss of information. Our meaning representation of choice is the Dependency Minimal Recursion Semantics (DMRS). We show that realizing chunks of a sentence and combining the results of such realizations increases the coverage for long sentences, significantly reduces the resources required and does not affect the quality of the realization.

We describe resources aimed at increasing the usability of the semantic representations utilized within the DELPH-IN (Deep Linguistic Processing with HPSG) consortium. We concentrate in particular on the Dependency Minimal Recursion Semantics (DMRS) formalism, a graph-based representation designed for compositional semantic representation with deep grammars. Our main focus is on English, and specifically English Resource Semantics (ERS) as used in the English Resource Grammar. We first give an introduction to ERS and DMRS and a brief overview of some existing resources and then describe in detail a new repository which has been developed to simplify the use of ERS/DMRS. We explain a number of operations on DMRS graphs which our repository supports, with sketches of the algorithms, and illustrate how these operations can be exploited in application building. We believe that this work will aid researchers to exploit the rich and effective but complex DELPH-IN resources.

When producing textual descriptions, humans express propositions regarding an object; but what do they express when annotating a document with simple tags? To answer this question, we have studied what users of tagging systems would have said if they were to describe a resource with fully fledged text. In particular, our work attempts to answer the following questions: if users were to use full descriptions, would their current tags be words present in these hypothetical sentences? If yes, what kind of language would connect these words? Such questions, although central to the problem of extracting binary relations between tags, have been sidestepped in the existing literature, which has focused on a small subset of possible inter-tag relations, namely hierarchical ones (e.g. “car” –is-a– “vehicle”), as opposed to non-taxonomical relations (e.g. “woman” –wears– “hat”). TagNText is the first attempt to construct a parallel corpus of tags and textual descriptions with respect to particular resources. The corpus provides enough data for the researcher to gain an insight into the nature of underlying relations, as well as the tools and methodology for constructing larger-scale parallel corpora that can aid non-taxonomical relation extraction.

The relevance of automatically identifying rhetorical moves in scientific texts has been widely acknowledged in the literature. This study focuses on abstracts of standard research papers written in English and aims to tackle a fundamental limitation of current machine-learning classifiers: they are mono-labeled, that is, a sentence can only be assigned one single label. However, such approach does not adequately reflect actual language use since a move can be realized by a clause, a sentence, or even several sentences. Here, we present MAZEA (Multi-label Argumentative Zoning for English Abstracts), a multi-label classifier which automatically identifies rhetorical moves in abstracts but allows for a given sentence to be assigned as many labels as appropriate. We have resorted to various other NLP tools and used two large training corpora: (i) one corpus consists of 645 abstracts from physical sciences and engineering (PE) and (ii) the other corpus is made up of 690 from life and health sciences (LH). This paper presents our preliminary results and also discusses the various challenges involved in multi-label tagging and works towards satisfactory solutions. In addition, we also make our two training corpora publicly available so that they may serve as benchmark for this new task.

Chemistry research papers are a primary source of information about chemistry, as in any scientific field. The presentation of the data is, predominantly, unstructured information, and so not immediately susceptible to processes developed within chemical informatics for carrying out chemistry research by information processing techniques. At one level, extracting the relevant information from research papers is a text mining task, requiring both extensive language resources and specialised knowledge of the subject domain. However, the papers also encode information about the way the research is conducted and the structure of the field itself. Applying language technology to research papers in chemistry can facilitate eScience on several different levels. The SciBorg project sets out to provide an extensive, analysed corpus of published chemistry research. This relies on the cooperation of several journal publishers to provide papers in an appropriate form. The work is carried out as a collaboration involving the Computer Laboratory, Chemistry Department and eScience Centre at Cambridge University, and is funded under the UK eScience programme.

We discuss preprocessing and tokenisation standards within DELPH-IN, a large scale open-source collaboration providing multiple independent multilingual shallow and deep processors. We discuss (i) a component-specific XML interface format which has been used for some time to interface preprocessor results to the PET parser, and (ii) our implementation of a more generic XML interface format influenced heavily by the (ISO working draft) Morphosyntactic Annotation Framework (MAF). Our generic format encapsulates the information which may be passed from the preprocessing stage to a parser: it uses standoff-annotation, a lattice for the representation of structural ambiguity, intra-annotation dependencies and allows for highly structured annotation content. This work builds on the existing Heart of Gold middleware system, and previous work on Robust Minimal Recursion Semantics (RMRS) as part of an inter-component interface. We give examples of usage with a number of the DELPH-IN processing components and deep grammars.