Discourse Tree Research Articles

Sentiment Analysis by Hierarchical Deep Neural Networks for Audience Opinion Mining

The prominent applications of sentiment analysis encompass various fields, including marketing, customer service, and communication. The conventional bag-of-words approach for measuring sentiment only counts term frequencies, while neglecting the position of the terms within the discourse. As a remedy, this research aims to build a discourse-aware approach upon the discourse structure of documents. For this purpose, rhetorical structure theory (RST) is utilized to label (sub-) clauses according to their hierarchical relationships, and then polarity scores are assigned to individual leaves. To learn from the resulting rhetorical structure, a hierarchical category structure-based deep recurrent neural network is proposed to infer underlying tensors of salient passages of narrative materials to process the complete discourse tree. The significance of this study lies in enhancing the structure of exploding and vanishing gradients in deep recurrent neural networks and also improving evaluation criteria in text analysis using the structure of opinion mining. The proposed approach is RST-HRNN. Exploding gradients is a process in the backpropagation stage aimed at continuously sampling the gradient of the model parameter in the opposite direction based on the weight (w), which is continuously updated until reaching the minimum global function.

Adjunct Ellipsis and Licensing Conditions

This study aims to show that adjunct ellipsis is allowed, albeit fairly restrictively, in Korean and how it is licensed and recovered. First, I closely examine the various environments reported in the literature, where adjuncts can be elided. After that, I claim that not only syntactic and semantic identity conditions but also a discourse-pragmatic condition needs to be satisfied for an adjunct to be deleted. In particular, I suggest, as the discourse-pragmatic condition, that the ellipsis clause should be congruent to ‘Question under Discussions’ (QuDs, Roberts 1998/2012), which are generated by the antecedent clause and the elliptical clause, by providing a proper answer to the QuDs, and that the QuDs form a discourse tree (d-tree) that starts at a more general superquestion and ends at answers to specific questions relevant to the discourse context (Büring 2003).

<em>Or not</em> Alternative Questions, Focus and Discourse Structure

Or not alternative questions like Are you coming or not? give rise to so-called ‘cornering effects’ (Biezma 2009), consisting of two parts: (i) they cannot appear discourse-initially, and (ii) they do not allow for follow-up questions. Building on recent experimental data (Beltrama, Meertens & Romero 2020), the present paper raises problems for current analyses (Biezma 2009, Biezma & Rawlins 2012, 2018), reframes the second part of cornering as not specific to NAQs but as a general constraint on questions in general, and develops a novel proposal for the first part of cornering. The key ingredients of the new proposal are the intrinsic focus structure of or not questions and its effects on discourse trees.

Unsupervised Learning of Discourse Structures using a Tree Autoencoder

Discourse information, as postulated by popular discourse theories, such as RST and PDTB, has been shown to improve an increasing number of downstream NLP tasks, showing positive effects and synergies of discourse with important real-world applications. While methods for incorporating discourse become more and more sophisticated, the growing need for robust and general discourse structures has not been sufficiently met by current discourse parsers, usually trained on small scale datasets in a strictly limited number of domains. This makes the prediction for arbitrary tasks noisy and unreliable. The overall resulting lack of high-quality, high-quantity discourse trees poses a severe limitation to further progress. In order the alleviate this shortcoming, we propose a new strategy to generate tree structures in a task-agnostic, unsupervised fashion by extending a latent tree induction framework with an auto-encoding objective. The proposed approach can be applied to any tree-structured objective, such as syntactic parsing, discourse parsing and others. However, due to the especially difficult annotation process to generate discourse trees, we initially develop a method to generate larger and more diverse discourse treebanks. In this paper we are inferring general tree structures of natural text in multiple domains, showing promising results on a diverse set of tasks.

Hierarchical Macro Discourse Parsing Based on Topic Segmentation

Hierarchically constructing micro (i.e., intra-sentence or inter-sentence) discourse structure trees using explicit boundaries (e.g., sentence and paragraph boundaries) has been proved to be an effective strategy. However, it is difficult to apply this strategy to document-level macro (i.e., inter-paragraph) discourse parsing, the more challenging task, due to the lack of explicit boundaries at the higher level. To alleviate this issue, we introduce a topic segmentation mechanism to detect implicit topic boundaries and then help the document-level macro discourse parser to construct better discourse trees hierarchically. In particular, our parser first splits a document into several sections using the topic boundaries that the topic segmentation detects. Then it builds a smaller and more accurate discourse sub-tree in each section and sequentially forms a whole tree for a document. The experimental results on both Chinese MCDTB and English RST-DT show that our proposed method outperforms the state-of-the-art baselines significantly.

Focus and contrastive topic in questions and answers, with particular reference to Turkish

AbstractMuch recent research has recognized the importance of focus and contrastive topic in assertions for discourse coherence. However, with few exceptions, it has been neglected that focus and contrastive topic also occur in questions, and have a similar role in establishing coherence. We propose a framework of dynamic interpretation based on the notion of Commitment Spaces that show that a uniform interpretation of focus and contrastive topic is possible. The algebraic representation format is rich enough so that a separate introduction of discourse trees is not necessary. The paper discusses these phenomena for Turkish, a language with an explicit focus marker for polar and alternative questions, which distinguishes focus from contrastive topic.

El análisis de las relaciones discursivas como recurso en el seguimiento de la Enfermedad de Alzheimer

This paper explores the results of a measurement proposal for discourse coherence of elderly patients who have been diagnosed with dementia of the Alzheimer’s type (DAT), in contrast with cognitively-healthy elderly. By means of the Rhetorical Structure Theory, the rhetorical relations underlying conversations of seven elderly patients in different stages of dementia and six cognitively healthy elderly were analyzed. The procedure involved segmentation of semantic discourse units (SDUs), labeling of relations and making of discourse trees. Additionally, relations of caregivers were counted for analyzing their contributions and conversational strategies in semi-structured interviews. We found that elderly patients produce more relations and use more variety of them. Involuntary Cause, Solution, Concession, Elaboration and Involuntary Result are related to cognitively healthy subjects’ discourse. On the other hand, some rhetorical relations are used by caregivers for pragmatic purposes. Due to our results, we believe that it is plausible to add the discourse coherence into the catalog of linguistic affectations considered for the clinic profile of a patient with DTA.

Sentiment analysis based on rhetorical structure theory:Learning deep neural networks from discourse trees

Prominent applications of sentiment analysis are countless, covering areas such as marketing, customer service and communication. The conventional bag-of-words approach for measuring sentiment merely counts term frequencies; however, it neglects the position of the terms within the discourse. As a remedy, we develop a discourse-aware method that builds upon the discourse structure of documents. For this purpose, we utilize rhetorical structure theory to label (sub-)clauses according to their hierarchical relationships and then assign polarity scores to individual leaves. To learn from the resulting rhetorical structure, we propose a tensor-based, tree-structured deep neural network (named Discourse-LSTM) in order to process the complete discourse tree. The underlying tensors infer the salient passages of narrative materials. In addition, we suggest two algorithms for data augmentation (node reordering and artificial leaf insertion) that increase our training set and reduce overfitting. Our benchmarks demonstrate the superior performance of our approach. Moreover, our tensor structure reveals the salient text passages and thereby provides explanatory insights.

Detecting logical argumentation in text via communicative discourse tree

ABSTRACTWe solve the argument mining problem by investigating discourse and communicative text structure. A new formal graph-based structure called communicative discourse tree (CDT) is defined. It consists of a discourse tree with additional labels on edges, which stand for verbs. These verbs represent communicative actions. Discourse trees are based on rhetoric relations, extracted from a text according to Rhetoric Structure Theory. The problem is tackled as a binary classification task, where the positive class corresponds to texts with arguments and the negative class corresponds to texts with no arguments. The feature engineering for the classification task is conducted, deciding on which syntactic and discourse features are associated with logical argumentation. Text classification framework based on syntactic, discourse and communicative discourse text structures with a number of learning approaches is implemented. Evaluation on a combined data-set is provided.

Discovering Rhetoric Agreement between a Request and Response

To support a natural flow of a conversation between humans and automated agents, rhetoric structures of each message has to be analyzed. We classify a pair of paragraphs of text as appropriate for one to follow another, or inappropriate, based on both topic and communicative discourse considerations. To represent a multi-sentence message with respect to how it should follow a previous message in a conversation or dialogue, we build an extension of a discourse tree for it. Extended discourse tree is based on a discourse tree for RST relations with labels for communicative actions, and also additional arcs for anaphora and ontology-based relations for entities. We refer to such trees as Communicative Discourse Trees (CDTs). We explore syntactic and discourse features that are indicative of correct vs incorrect request-response or question-answer pairs. Two learning frameworks are used to recognize such correct pairs: deterministic, nearest-neighbor learning of CDTs as graphs, and a tree kernel learning of CDTs, where a feature space of all CDT sub-trees is subject to SVM learning. We form the positive training set from the correct pairs obtained from Yahoo Answers, social network, corporate conversations including Enron emails, customer complaints and interviews by journalists. The corresponding negative training set is artificially created by attaching responses for different, inappropriate requests that include relevant keywords. The evaluation showed that it is possible to recognize valid pairs in 70% of cases in the domains of weak request-response agreement and 80% of cases in the domains of strong agreement, which is essential to support automated conversations. These accuracies are comparable with the benchmark task of classification of discourse trees themselves as valid or invalid, and also with classification of multi-sentence answers in factoid question-answering systems. The applicability of proposed machinery to the problem of chatbots, social chats and programming via NL is demonstrated. We conclude that learning rhetoric structures in the form of CDTs is the key source of data to support answering complex questions, chatbots and dialogue management.

Rhetorical relations revisited across distinct levels of discourse unit granularity

In accordance with the compositionality criterion and hierarchy principle of Rhetorical Structure Theory (RST), this study reframes each tree in the RST Discourse Treebank into three new dependency trees with ultimate nodes being clauses, sentences, and paragraphs, respectively, which also draw on an analogy between syntactic and discourse trees. Detailed percentages of various RST relations at the three granularity levels are examined, illuminating the discourse processes of organizing units of one granularity level into those of the next upper level and suggesting certain homogeneity and interaction across levels in the Treebank, particularly at the two upper levels. The study demonstrates the applicability of RST analysis between same-level terminal units. With unique analytical advantages, the newly constructed discourse dependency trees provide new research prospects.

Visualizing structural “inverted pyramids” in English news discourse across levels

Abstract Drawing on an analogy between discourse and syntactic trees, this paper chooses 359 Wall Street Journal articles with multiple paragraphs from the Rhetorical Structure Theory (RST) Discourse Treebank, and converts each discourse tree into three additional dependency ones, at discourse, paragraph and sentence levels, with exclusively elementary discourse units of clauses, sentences and paragraphs, respectively. It empirically tests and visually presents the genre-specific “summary+details” or “inverted pyramid” structuring of news discourse. It further extends the idea of inverted pyramid structuring to the paragraph and sentence levels. It proves that the body of the report also has a similar schematic top-down installment organization with macro-propositions on top. It also visually and statistically presents the rhetorical structures at sentence level, which differ to some extent from grammatical structures. Operated in line with the compositionality criterion and hierarchy principle of RST, the converted trees provide unique analytical advantages and constitute new research prospects.

Annotations of Connectives and Arguments in Malayalam Language

Discourse relations in natural languages link clauses in text and compose overall text structure. Discourse connectives are an important part of modeling the Malayalam discourse structure. We followed the annotation procedure of Penn Discourse Tree Bank and worked on tagging of discourse connectives and arguments of Malayalam text and also report the senses of relation. We present our work on annotations of Malayalam discourse connectives and arguments which helps to know more about the discourse connectives and their appearance in case of semantic rules in Malayalam discourse. Discourse connectives may or may not be explicitly present in the relation. In our work, we focus on the annotation of both explicit and implicit connectives and arguments in Malayalam text and showed encouraging results.

Summarizing a Document by Trimming the Discourse Tree

Recent studies on extractive text summarization formulate it as a combinatorial optimization problem, extracting the optimal subset from a set of the textual units that maximizes an objective function without violating the length constraint. Although these methods successfully improve automatic evaluation scores, they do not consider the discourse structure in the source document. Thus, summaries generated by these methods may lack logical coherence. In previous work, we proposed a method that exploits a discourse tree structure to produce coherent summaries. By transforming a traditional discourse tree, namely a rhetorical structure theory-based discourse tree (RST-DT), into a dependency-based discourse tree (DEP-DT), we formulated the summarization procedure as a Tree Knapsack Problem whose tree corresponds to the DEP-DT. This paper extends the work with a detailed discussion of the approach together with a novel efficient dynamic programming algorithm for solving the Tree Knapsack Problem. Experiments show that our method not only achieved the highest score in both automatic and human evaluation, but also obtained good performance in terms of the linguistic qualities of the summaries.

CODRA: A Novel Discriminative Framework for Rhetorical Analysis

Clauses and sentences rarely stand on their own in an actual discourse; rather, the relationship between them carries important information that allows the discourse to express a meaning as a whole beyond the sum of its individual parts. Rhetorical analysis seeks to uncover this coherence structure. In this article, we present CODRA— a COmplete probabilistic Discriminative framework for performing Rhetorical Analysis in accordance with Rhetorical Structure Theory, which posits a tree representation of a discourse.CODRA comprises a discourse segmenter and a discourse parser. First, the discourse segmenter, which is based on a binary classifier, identifies the elementary discourse units in a given text. Then the discourse parser builds a discourse tree by applying an optimal parsing algorithm to probabilities inferred from two Conditional Random Fields: one for intra-sentential parsing and the other for multi-sentential parsing. We present two approaches to combine these two stages of parsing effectively. By conducting a series of empirical evaluations over two different data sets, we demonstrate that CODRA significantly outperforms the state-of-the-art, often by a wide margin. We also show that a reranking of the k-best parse hypotheses generated by CODRA can potentially improve the accuracy even further.

Annotating the meaning of discourse connectives in multilingual corpora

AbstractDiscourse connectives are lexical items indicating coherence relations between discourse segments. Even though many languages possess a whole range of connectives, important divergences exist cross-linguistically in the number of connectives that are used to express a given relation. For this reason, connectives are not easily paired with a univocal translation equivalent across languages. This paper is a first attempt to design a reliable method to annotate the meaning of discourse connectives cross-linguistically using corpus data. We present the methodological choices made to reach this aim and report three annotation experiments using the framework of the Penn Discourse Tree Bank.

HILDA: A Discourse Parser Using Support Vector Machine Classification

Discourse structures have a central role in several computational tasks, such as question-answering or dialogue generation. In particular, the framework of the Rhetorical Structure Theory (RST) offers a sound formalism for hierarchical text organization. In this article, we present HILDA, an implemented discourse parser based on RST and Support Vector Machine (SVM) classification. SVM classifiers are trained and applied to discourse segmentation and relation labeling. By combining labeling with a greedy bottom-up tree building approach, we are able to create accurate discourse trees in linear time complexity. Importantly, our parser can parse entire texts, whereas the publicly available parser SPADE (Soricut and Marcu 2003) is limited to sentence level analysis. HILDA outperforms other discourse parsers for tree structure construction and discourse relation labeling. For the discourse parsing task, our system reaches 78.3% of the performance level of human annotators. Compared to a state-of-the-art rule-based discourse parser, our system achieves a performance increase of 11.6%.

Document summarisation using combination and reduction of extracted sentences

An ideal summariser should produce a summary which contains all the crucial information present in the original text, while conforming to the size of the summary. We propose a novel method to perform single-document summarisation of the English text. We start by combining semantically related sentences using a rule-based approach to avoid loss of important information and maintain coherence in the resulting summary. The rules for combination rely on surface indicators present in the sentence i.e., cue-phrases. Then we extract important text from the combined sentences using lexical chains. This is followed by a sentence-reduction step involving removal of superfluous phrases from the extracted sentences using a rule-based approach. The rules for sentence pruning use discourse trees generated using the intra-sentential rhetorical relations. The summaries produced by our system were agreed by human subjects to be more concise and coherent than extraction-based summaries for the same documents.

Active Learning with Multiple Views

Active learners alleviate the burden of labeling large amounts of data by detecting and asking the user to label only the most informative examples in the domain. We focus here on active learning for multi-view domains, in which there are several disjoint subsets of features (views), each of which is sufficient to learn the target concept. In this paper we make several contributions. First, we introduce Co-Testing, which is the first approach to multi-view active learning. Second, we extend the multi-view learning framework by also exploiting weak views, which are adequate only for learning a concept that is more general/specific than the target concept. Finally, we empirically show that Co-Testing outperforms existing active learners on a variety of real world domains such as wrapper induction, Web page classification, advertisement removal, and discourse tree parsing.

Ellipsis and the Structure of Discourse

It is generally assumed that ellipsis requires certain parallelism between the clause containing the ellips is and some antecedent clause. We argue that the parallelism requirement generated by ellipsis must be applied in accordance with discourse structure: a matching antecedent clause must be found that locally c-commands the clause containing the ellipsis in the discourse tree. We show that the claim makes several correct predictions concerning the interpretation of ellipsis, both in terms of the selection of the antecedent (in Sluicing and Verb Phrase Ellipsis), and in terms of the possible readings given a particular antecedent (in the “many-clause” puzzle and in Antecedent-Contained Deletion).