Paraphrase identification involves identifying whether a pair of sentences express the same or similar meanings.While cross-encoders have achieved high performances across several benchmarks, bi-encoders such as SBERT have been widely applied to sentence pair tasks.They exhibit substantially lower computation complexity and are better suited to symmetric tasks.In this work, we adopt a biencoder approach to the paraphrase identification task, and investigate the impact of explicitly incorporating predicate-argument information into SBERT through weighted aggregation.Experiments on six paraphrase identification datasets demonstrate that, with a minimal increase in parameters, the proposed model is able to outperform SBERT/SRoBERTa significantly.Further, ablation studies reveal that the predicate-argument based component plays a significant role in the performance gain.

Previous works have demonstrated the effectiveness of utilising pre-trained sentence encoders based on their sentence representations for meaning comparison tasks. Though such representations are shown to capture hidden syntax structures, the direct similarity comparison between them exhibits weak sensitivity to word order and structural differences in given sentences. A single similarity score further makes the comparison process hard to interpret. Therefore, we here propose to combine sentence encoders with an alignment component by representing each sentence as a list of predicate-argument spans (where their span representations are derived from sentence encoders), and decomposing the sentence-level meaning comparison into the alignment between their spans for paraphrase identification tasks. Empirical results show that the alignment component brings in both improved performance and interpretability for various sentence encoders. After closer investigation, the proposed approach indicates increased sensitivity to structural difference and enhanced ability to distinguish non-paraphrases with high lexical overlap.

Previous work has demonstrated that pre-trained large language models (LLM) acquire knowledge during pre-training which enables reasoning over relationships between words (e.g, hyponymy) and more complex inferences over larger units of meaning such as sentences. Here, we investigate whether lexical entailment (LE, i.e. hyponymy or the is a relation between words) can be generalised in a compositional manner. Accordingly, we introduce PLANE (Phrase-Level Adjective-Noun Entailment), a new benchmark to test models on fine-grained compositional entailment using adjective-noun phrases. Our experiments show that knowledge extracted via In–Context and transfer learning is not enough to solve PLANE. However, a LLM trained on PLANE can generalise well to out–of–distribution sets, since the required knowledge can be stored in the representations of subwords (SW) tokens.

This paper addresses a deficiency in existing cross-lingual information retrieval (CLIR) datasets and provides a robust evaluation of CLIR systems’ disambiguation ability. CLIR is commonly tackled by combining translation and traditional IR. Due to translation ambiguity, the problem of ambiguity is worse in CLIR than in monolingual IR. But existing auto-generated CLIR datasets are dominated by searches for named entity mentions, which does not provide a good measure for disambiguation performance, as named entity mentions can often be transliterated across languages and tend not to have multiple translations. Therefore, we introduce a new evaluation dataset (MuSeCLIR) to address this inadequacy. The dataset focusses on polysemous common nouns with multiple possible translations. MuSeCLIR is constructed from multilingual Wikipedia and supports searches on documents written in European (French, German, Italian) and Asian (Chinese, Japanese) languages. We provide baseline statistical and neural model results on MuSeCLIR which show that MuSeCLIR has a higher requirement on the ability of systems to disambiguate query terms.

The automatic detection of hypernymy relationships represents a challenging problem in NLP.The successful application of stateof-the-art supervised approaches using distributed representations has generally been impeded by the limited availability of high quality training data.We have developed two novel data augmentation techniques which generate new training examples from existing ones.First, we combine the linguistic principles of hypernym transitivity and intersective modifier-noun composition to generate additional pairs of vectors, such as small dogdog or small dog -animal, for which a hypernymy relationship can be assumed.Second, we use generative adversarial networks (GANs) to generate pairs of vectors for which the hypernymy relation can also be assumed.We furthermore present two complementary strategies for extending an existing dataset by leveraging linguistic resources such as Word-Net.Using an evaluation across 3 different datasets for hypernymy detection and 2 different vector spaces, we demonstrate that both of the proposed automatic data augmentation and dataset extension strategies substantially improve classifier performance.

Distributional models are derived from cooccurrences in a corpus, where only a small proportion of all possible plausible cooccurrences will be observed.This results in a very sparse vector space, requiring a mechanism for inferring missing knowledge.Most methods face this challenge in ways that render the resulting word representations uninterpretable, with the consequence that semantic composition becomes hard to model.In this paper we explore an alternative which involves explicitly inferring unobserved co-occurrences using the distributional neighbourhood.We show that distributional inference improves sparse word representations on several word similarity benchmarks and demonstrate that our model is competitive with the state-of-the-art for adjectivenoun, noun-noun and verb-object compositions while being fully interpretable.

We present a new framework for compositional distributional semantics in which the distributional contexts of lexemes are expressed in terms of anchored packed dependency trees. We show that these structures have the potential to capture the full sentential contexts of a lexeme and provide a uniform basis for the composition of distributional knowledge in a way that captures both mutual disambiguation and generalization.

There has been a great deal of recent research into word sense disambiguation, particularly since the inception of the Senseval evaluation exercises. Because a word often has more than one meaning, resolving word sense ambiguity could benefit applications that need some level of semantic interpretation of language input. A major problem is that the accuracy of word sense disambiguation systems is strongly dependent on the quantity of manually sense-tagged data available, and even the best systems, when tagging every word token in a document, perform little better than a simple heuristic that guesses the first, or predominant, sense of a word in all contexts. The success of this heuristic is due to the skewed nature of word sense distributions. Data for the heuristic can come from either dictionaries or a sample of sense-tagged data. However, there is a limited supply of the latter, and the sense distributions and predominant sense of a word can depend on the domain or source of a document. (The first sense of “star” for example would be different in the popular press and scientific journals). In this article, we expand on a previously proposed method for determining the predominant sense of a word automatically from raw text. We look at a number of different data sources and parameterizations of the method, using evaluation results and error analyses to identify where the method performs well and also where it does not. In particular, we find that the method does not work as well for verbs and adverbs as nouns and adjectives, but produces more accurate predominant sense information than the widely used SemCor corpus for nouns with low coverage in that corpus. We further show that the method is able to adapt successfully to domains when using domain specific corpora as input and where the input can either be hand-labeled for domain or automatically classified.

Techniques that exploit knowledge of distributional similarity between words have been proposed in many areas of Natural Language Processing. For example, in language modeling, the sparse data problem can be alleviated by estimating the probabilities of unseen co-occurrences of events from the probabilities of seen co-occurrences of similar events. In other applications, distributional similarity is taken to be an approximation to semantic similarity. However, due to the wide range of potential applications and the lack of a strict definition of the concept of distributional similarity, many methods of calculating distributional similarity have been proposed or adopted. In this work, a flexible, parameterized framework for calculating distributional similarity is proposed. Within this framework, the problem of finding distributionally similar words is cast as one of co-occurrence retrieval (CR) for which precision and recall can be measured by analogy with the way they are measured in document retrieval. As will be shown, a number of popular existing measures of distributional similarity are simulated with parameter settings within the CR framework. In this article, the CR framework is then used to systematically investigate three fundamental questions concerning distributional similarity. First, is the relationship of lexical similarity necessarily symmetric, or are there advantages to be gained from considering it as an asymmetric relationship? Second, are some co-occurrences inherently more salient than others in the calculation of distributional similarity? Third, is it necessary to consider the difference in the extent to which each word occurs in each co-occurrence type? Two application-based tasks are used for evaluation: automatic thesaurus generation and pseudo-disambiguation. It is possible to achieve significantly better results on both these tasks by varying the parameters within the CR framework rather than using other existing distributional similarity measures; it will also be shown that any single unparameterized measure is unlikely to be able to do better on both tasks. This is due to an inherent asymmetry in lexical substitutability and therefore also in lexical distributional similarity.

In word sense disambiguation (WSD), the heuristic of choosing the most common sense is extremely powerful because the distribution of the senses of a word is often skewed. The problem with using the predominant, or first sense heuristic, aside from the fact that it does not take surrounding context into account, is that it assumes some quantity of hand-tagged data. Whilst there are a few hand-tagged corpora available for some languages, one would expect the frequency distribution of the senses of words, particularly topical words, to depend on the genre and domain of the text under consideration. We present work on the use of a thesaurus acquired from raw textual corpora and the WordNet similarity package to find predominant noun senses automatically. The acquired predominant senses give a precision of 64% on the nouns of the SENSEVAL-2 English all-words task. This is a very promising result given that our method does not require any hand-tagged text, such as SemCor. Furthermore, we demonstrate that our method discovers appropriate predominant senses for words from two domain-specific corpora.