Discourse Relation Classification Research Articles

Discriminative explicit instance selection for implicit discourse relation classification

Discriminative explicit instance selection for implicit discourse relation classification

Enhanced semantic representation learning for implicit discourse relation classification

Enhanced semantic representation learning for implicit discourse relation classification

Learning Context-Aware Convolutional Filters for Implicit Discourse Relation Classification

Implicit discourse relation classification (IDRC) is considered the most difficult component of shallow discourse parsing as the relation prediction in the absence of necessary clues requires a deep understanding of the context information of the sentences. Convolutional neural networks (CNNs) have emerged as an important encoding block for sentences in natural language processing (NLP). CNNs use a specific set of filters for the inputs which may lead to the partial coverage of contextual clues. Furthermore, conventional CNNs may not allow the initial communication between the sentences which is a crucial step for IDRC. We present an adaptive convolution approach for IDRC that utilizes context aware filters for the convolution operation. The goal is to abstract the context of sentences in the filters and let them interact with sentence representations, i.e. learning the representations through learned filters. Our model acts as a cross questioning agent by generating filters from one argument and convolving them with the other for the IDRC task. This process is analogous to the attention mechanism because both methods aim at abstracting contextual information. Different from the attention mechanism, our approach directly encodes the contextual representations in the form of filters and allows the initial communication between arguments during encoding. Furthermore, the adaptive convolution can also work alongside the attention mechanism to enhance the representational ability of the adaptive CNN encoder. Experiments on PDTB 2.0 and CDTB datasets show that our approach outperforms all the baselines by a fair margin and achieves excellent results.

A Neural Approach to Discourse Relation Signal Detection

Previous data-driven work investigating the types and distributions of discourse relation signals, including discourse markers such as 'however' or phrases such as 'as a result' has focused on the relative frequencies of signal words within and outside text from each discourse relation. Such approaches do not allow us to quantify the signaling strength of individual instances of a signal on a scale (e.g. more or less discourse-relevant instances of 'and'), to assess the distribution of ambiguity for signals, or to identify words that hinder discourse relation identification in context ('anti-signals' or 'distractors'). In this paper we present a data-driven approach to signal detection using a distantly supervised neural network and develop a metric, Δs (or 'delta-softmax'), to quantify signaling strength. Ranging between -1 and 1 and relying on recent advances in contextualized words embeddings, the metric represents each word's positive or negative contribution to the identifiability of a relation in specific instances in context. Based on an English corpus annotated for discourse relations using Rhetorical Structure Theory and signal type annotations anchored to specific tokens, our analysis examines the reliability of the metric, the places where it overlaps with and differs from human judgments, and the implications for identifying features that neural models may need in order to perform better on automatic discourse relation classification.

Implicit Discourse Relation Classification via Multi-Task Neural Networks

Without discourse connectives, classifying implicit discourse relations is a challenging task and a bottleneck for building a practical discourse parser. Previous research usually makes use of one kind of discourse framework such as PDTB or RST to improve the classification performance on discourse relations. Actually, under different discourse annotation frameworks, there exist multiple corpora which have internal connections. To exploit the combination of different discourse corpora, we design related discourse classification tasks specific to a corpus, and propose a novel Convolutional Neural Network embedded multi-task learning system to synthesize these tasks by learning both unique and shared representations for each task. The experimental results on the PDTB implicit discourse relation classification task demonstrate that our model achieves significant gains over baseline systems.

Manually vs. Automatically Labelled Data in Discourse Relation Classification: Effects of Example and Feature Selection

Wordnets are lexical reference systems that follow the design principles of the Princeton WordNet project (Fellbaum, ). Domain ontologies (or domain-specific ontologies such as GOLD, or the GENE Ontology) represent knowledge about a specific domain in a format that supports automated reasoning about the objects in that domain and the relations between them (Erdmann, ). In this paper, we will discuss how the Web Ontology Language OWL can be used to represent and interrelate the entities and relations in both types of resources. Our special focus will be on the question, whether synsets should be modelled as individuals (we use individual and instance as synonyms and will refer to this option as instance model) or as classes (we will refer to this option as class model). We will present three OWL models, each of which offers different solutions to this question. These models were developed in the context of the research group “Text-technological Modelling of Information” as a collaboration of the projects SemDok and HyTex. Since these projects are mainly concerned with German documents and with corpora that contain documents of a special technical or scientific domain, we used subsets of the German wordnet GermaNet (Kunze and Lemnitzer, ), henceforth referred to as GN, and the German domain ontology TermNet (Beiswenger et al., ), henceforth referred to as TN, to develop and evaluate the three models. To relate the general vocabulary of GN with the domain specific terms in TN, we developed an approach that was inspired by the plug-in model proposed by Magnini and Speranza (). In this approach, which has been developed in cooperation with the GermaNet research group (see Kunze et al. () for details), we adapted the OWL model for the English Princeton WordNet suggested by van Assem et al. () to GN, i.e. we modelled German synsets as instances of word-class-specific synset classes. For the reasons explained in section , we wanted to experiment with alternative models that implement the class model. In section  we will present three alternative OWL representations for GN and TN and discuss their benefits and drawbacks.