Explanations are pervasive in our lives. Mostly, they occur in dialogical form where an {\em explainer} discusses a concept or phenomenon of interest with an {\em explainee}. Leaving the explainee with a clear understanding is not straightforward due to the knowledge gap between the two participants. Previous research looked at the interaction of explanation moves, dialogue acts, and topics in successful dialogues with expert explainers. However, daily-life explanations often fail, raising the question of what makes a dialogue successful. In this work, we study explanation dialogues in terms of the interactions between the explainer and explainee and how they correlate with the quality of explanations in terms of a successful understanding on the explainee's side. In particular, we first construct a corpus of 399 dialogues from the Reddit forum {\em Explain Like I am Five} and annotate it for interaction flows and explanation quality. We then analyze the interaction flows, comparing them to those appearing in expert dialogues. Finally, we encode the interaction flows using two language models that can handle long inputs, and we provide empirical evidence for the effectiveness boost gained through the encoding in predicting the success of explanation dialogues.
In recent years, large language models have achieved state-of-the-art performance across various NLP tasks. However, investigations have shown that these models tend to rely on shortcut features, leading to inaccurate predictions and causing the models to be unreliable at generalization to out-of-distribution (OOD) samples. For instance, in the context of relation extraction (RE), we would expect a model to identify the same relation independently of the entities involved in it. For example, consider the sentence "Leonardo da Vinci painted the Mona Lisa" expressing the created(Leonardo_da_Vinci, Mona_Lisa) relation. If we substiute "Leonardo da Vinci" with "Barack Obama", then the sentence still expresses the created relation. A robust model is supposed to detect the same relation in both cases. In this work, we describe several semantically-motivated strategies to generate adversarial examples by replacing entity mentions and investigate how state-of-the-art RE models perform under pressure. Our analyses show that the performance of these models significantly deteriorates on the modified datasets (avg. of -48.5% in F1), which indicates that these models rely to a great extent on shortcuts, such as surface forms (or patterns therein) of entities, without making full use of the information present in the sentences.
Many annotation tasks in natural language processing are highly subjective in that there can be different valid and justified perspectives on what is a proper label for a given example. This also applies to the judgment of argument quality, where the assignment of a single ground truth is often questionable. At the same time, there are generally accepted concepts behind argumentation that form a common ground. To best represent the interplay of individual and shared perspectives, we consider a continuum of approaches ranging from models that fully aggregate perspectives into a majority label to "share nothing"-architectures in which each annotator is considered in isolation from all other annotators. In between these extremes, inspired by models used in the field of recommender systems, we investigate the extent to which architectures that include layers to model the relations between different annotators are beneficial for predicting single-annotator labels. By means of two tasks of argument quality classification (argument concreteness and validity/novelty of conclusions), we show that recommender architectures increase the averaged annotator-individual F$_1$-scores up to $43\%$ over a majority label model. Our findings indicate that approaches to subjectivity can benefit from relating individual perspectives.
OTTR is a language for representing ontology modeling patterns, which enables to build ontologies or knowledge bases by instantiating templates. Thereby, particularities of the ontological representation language are hidden from the domain experts, and it enables ontology engineers to, to some extent, separate the processes of deciding about what information to model from deciding about how to model the information, e.g., which design patterns to use. Certain decisions can thus be postponed for the benefit of focusing on one of these processes. To date, only few works on ontology engineering where ontology templates are applied are described in the literature. In this paper, we outline our methodology and report findings from our ontology engineering activities in the domain of Material Science. In these activities, OTTR templates play a key role. Our ontology engineering process is bottom-up, as we begin modeling activities from existing data that is then, via templates, fed into a knowledge graph, and it is top-down, as we first focus on which data to model and postpone the decision of how to model the data. We find, among other things, that OTTR templates are especially useful as a means of communication with domain experts. Furthermore, we find that because OTTR templates encapsulate modeling decisions, the engineering process becomes flexible, meaning that design decisions can be changed at little cost.
Many NLP tasks exhibit human label variation, where different annotators give different labels to the same texts. This variation is known to depend, at least in part, on the sociodemographics of annotators. Recent research aims to model individual annotator behaviour rather than predicting aggregated labels, and we would expect that sociodemographic information is useful for these models. On the other hand, the ecological fallacy states that aggregate group behaviour, such as the behaviour of the average female annotator, does not necessarily explain individual behaviour. To account for sociodemographics in models of individual annotator behaviour, we introduce group-specific layers to multi-annotator models. In a series of experiments for toxic content detection, we find that explicitly accounting for sociodemographic attributes in this way does not significantly improve model performance. This result shows that individual annotation behaviour depends on much more than just sociodemographics.
Arguments often do not make explicit how a conclusion follows from its premises. To compensate for this lack, we enrich arguments with structured background knowledge to support knowledge-intense argumentation tasks. We present a new unsupervised method for constructing Contextualized Commonsense Knowledge Graphs (CCKGs) that selects contextually relevant knowledge from large knowledge graphs (KGs) efficiently and at high quality. Our work goes beyond context-insensitive knowledge extraction heuristics by computing semantic similarity between KG triplets and textual arguments. Using these triplet similarities as weights, we extract contextualized knowledge paths that connect a conclusion to its premise, while maximizing similarity to the argument. We combine multiple paths into a CCKG that we optionally prune to reduce noise and raise precision. Intrinsic evaluation of the quality of our graphs shows that our method is effective for (re)constructing human explanation graphs. Manual evaluations in a large-scale knowledge selection setup confirm high recall and precision of implicit CSK in the CCKGs. Finally, we demonstrate the effectiveness of CCKGs in a knowledge-insensitive argument quality rating task, outperforming strong baselines and rivaling a GPT-3 based system.
Key point analysis is the task of extracting a set of concise and high-level statements from a given collection of arguments, representing the gist of these arguments. This paper presents our proposed approach to the Key Point Analysis shared task, collocated with the 8th Workshop on Argument Mining. The approach integrates two complementary components. One component employs contrastive learning via a siamese neural network for matching arguments to key points; the other is a graph-based extractive summarization model for generating key points. In both automatic and manual evaluation, our approach was ranked best among all submissions to the shared task.
Aspect-based sentiment analysis involves the recognition of so called opinion target expressions (OTEs). To automatically extract OTEs, supervised learning algorithms are usually employed which are trained on manually annotated corpora. The creation of these corpora is labor-intensive and sufficiently large datasets are therefore usually only available for a very narrow selection of languages and domains. In this work, we address the lack of available annotated data for specific languages by proposing a zero-shot cross-lingual approach for the extraction of opinion target expressions. We leverage multilingual word embeddings that share a common vector space across various languages and incorporate these into a convolutional neural network architecture for OTE extraction. Our experiments with 5 languages give promising results: We can successfully train a model on annotated data of a source language and perform accurate prediction on a target language without ever using any annotated samples in that target language. Depending on the source and target language pairs, we reach performances in a zero-shot regime of up to 77% of a model trained on target language data. Furthermore, we can increase this performance up to 87% of a baseline model trained on target language data by performing cross-lingual learning from multiple source languages.
The task of answering natural language questions over knowledge bases has received wide attention in recent years. Various deep learning architectures have been proposed for this task. However, architectural design choices are typically not systematically compared nor evaluated under the same conditions. In this paper, we contribute to a better understanding of the impact of architectural design choices by evaluating four different architectures under the same conditions. We address the task of answering simple questions, consisting in predicting the subject and predicate of a triple given a question. In order to provide a fair comparison of different architectures, we evaluate them under the same strategy for inferring the subject, and compare different architectures for inferring the predicate. The architecture for inferring the subject is based on a standard LSTM model trained to recognize the span of the subject in the question and on a linking component that links the subject span to an entity in the knowledge base. The architectures for predicate inference are based on i) a standard softmax classifier ranging over all predicates as output, iii) a model that predicts a low-dimensional encoding of the property given entity representation and question, iii) a model that learns to score a pair of subject and predicate given the question as well as iv) a model based on the well-known FastText model. The comparison of architectures shows that FastText provides better results than other architectures.
The task of answering natural language questions over RDF data has received wide interest in recent years, in particular in the context of the series of QALD benchmarks. The task consists of mapping a natural language question to an executable form, e.g. SPARQL, so that answers from a given KB can be extracted. So far, most systems proposed are i) monolingual and ii) rely on a set of hard-coded rules to interpret questions and map them into a SPARQL query. We present the first multilingual QALD pipeline that induces a model from training data for mapping a natural language question into logical form as probabilistic inference. In particular, our approach learns to map universal syntactic dependency representations to a language-independent logical form based on DUDES (Dependency-based Underspecified Discourse Representation Structures) that are then mapped to a SPARQL query as a deterministic second step. Our model builds on factor graphs that rely on features extracted from the dependency graph and corresponding semantic representations. We rely on approximate inference techniques, Markov Chain Monte Carlo methods in particular, as well as Sample Rank to update parameters using a ranking objective. Our focus lies on developing methods that overcome the lexical gap and present a novel combination of machine translation and word embedding approaches for this purpose. As a proof of concept for our approach, we evaluate our approach on the QALD-6 datasets for English, German & Spanish.