Semantic Constraints Research Articles

Describing Linguistic Vagueness of Evaluative Expressions Using Fuzzy Natural Logic and Linguistic Constraints

In recent years, the study of evaluative linguistic expressions has crossed the field of theoretical linguistics and has aroused interest in very different research areas such as artificial intelligence, psychology or cognitive linguistics. The interest in this type of expressions may be due to its relevance in applications such as opinion mining or sentiment analysis. This paper brings together Fuzzy Natural Logic and Fuzzy Property Grammars to approach evaluative expressions. Our contribution includes the marriage of mathematical and linguistic methods for providing a formalism to deal with the linguistic vagueness of evaluative expressions by describing the syntax and semantics of these structures. We contribute to the study of evaluative linguistic expressions by proposing a formal characterization of them where the concepts of semantic prime, borderline evaluative expressions and markedness are defined and where the relation between the semantic constraints of evaluations and their sentiment can be established. A proof-of-concept of how to create a lexicon of evaluative expressions for future computational applications is presented. The results demonstrate that linguistic evaluative expressions are gradient, have sentiment, and that the evaluations work as a relation of hypernym and hyponym, the hypernym being a semantic prime. Our findings provide the basis for building an ontology of evaluative expressions for future computational applications.

Blueprint

Blueprint is a declarative domain-specific language for document extraction. Users describe document layout using spatial, textual, semantic, and numerical fuzzy constraints, and the language runtime extracts the field-value mappings that best satisfy the constraints in a given document. We used Blueprint to develop several document extraction solutions in a commercial setting. This approach to the extraction problem proved powerful. Concise Blueprint programs were able to generate good accuracy on a broad set of use cases. However, a major goal of our work was to build a system that non-experts, and in particular non-engineers, could use effectively, and we found that writing declarative fuzzy constraint-based extraction programs was not intuitive for many users: a large up-front learning investment was required to be effective, and debugging was often challenging. To address these issues, we developed a no-code IDE for Blueprint, called Studio, as well as program synthesis functionality for automatically generating Blueprint programs from training data, which could be created by labeling document samples in our IDE. Overall, the IDE significantly improved the Blueprint development experience and the results users were able to achieve. In this paper, we discuss the design, implementation, and deployment of Blueprint and Studio. We compare our system with a state-of-the-art deep-learning based extraction tool and show that our system can achieve comparable accuracy results, with comparable development time, for appropriately-chosen use cases, while providing better interpretability and debuggability.

BIM-based design decisions documentation using design episodes, explanation tags, and constraints

The process of designing a building involves producing design concepts while fulfilling various requirements and regulations. Furthermore, during the project's life-cycle, multiple experts from multiple domains collaborate in developing the different partial models, including architectural, structural, and HVAC among others. Accordingly, clearly communicating the rationale behind design decisions is crucial for developing regulatory compliant designs that also fit the owner’s needs. The developed designs are the main deliverables exchanged and handed over. However, these deliverables do not include any explanation of design intentions or documentation of design decisions. Communication among parties and reuse of knowledge are hindered by the absent explanation of existing design. To overcome this deficiency, this paper proposes a methodology for digitally documenting design decisions, incorporating their intention and rationale. Architectural concepts and evaluation criteria are represented in the form of explanation tags as well as spatial and semantic constraints, which are assigned to the individual model elements and properties. Additionally, to document how design decisions fulfill owner requirements and regulatory documents, natural language processing (NLP) is employed to facilitate querying those documents and then the individual requirements are linked to specific elements, properties, and constraints. To evaluate the proposed methodology, a prototype was implemented as a plugin inside a BIM- Authoring tool and multiple real-world use cases are discussed.

An experimental perspective on embedded gapping in Persian

Abstract This paper empirically tests the embedding constraints on gapping in Persian. It has been suggested that gapping differs from other kinds of ellipsis in banning embedding. However, the first counter-examples in the literature come from Persian. Following up on previous experiments on embedded gapping in several languages, we report the results of two acceptability judgment tasks. Our results show that, while embedded gapping is overall acceptable in Persian, speakers’ acceptability judgements also vary depending on the semantic type of the embedding predicate, as well as the presence/absence of the complementizer. Data from Persian highlight that, despite the cross-linguistic variation observed with respect to the acceptability of embedded gapping, a general semantic constraint is at work across languages: non-factive verbs embed more easily than factive ones; inside factive verbs, semi-factive (cognitive) predicates embed more easily than true factive (emotive) ones. Moreover, whereas previous theoretical literature indicates no systematic preference for the absence or the presence of the complementizer in Persian, these new experimental data suggest a preference for complementizer drop. To account for the gradience observed in our experimental data, we propose an approach of gapping based on acceptability rather than grammaticality.

A multimodal domain adaptive segmentation framework for IDH genotype prediction.

The gene mutation status of isocitrate dehydrogenase (IDH) in gliomas leads to a different prognosis. It is challenging to perform automated tumor segmentation and genotype prediction directly using label-deprived multimodal magnetic resonance (MR) images. We propose a novel framework that employs a domain adaptive mechanism to address this issue. Multimodal domain adaptive segmentation (MDAS) framework was proposed to solve the gap issue in cross dataset model transfer. Image translation was used to adaptively align the multimodal data from two domains at the image level, and segmentation consistency loss was proposed to retain more pathological information through semantic constraints. The data distribution between the labeled public dataset and label-free target dataset was learned to achieve better unsupervised segmentation results on the target dataset. Then, the segmented tumor foci were used as a mask to extract the radiomics and deep features. And the subsequent prediction of IDH gene mutation status was conducted by training a random forest classifier. The prediction model does not need any expert segmented labels. We implemented our method on the public BraTS 2019 dataset and 110 astrocytoma cases of grade II-IV brain tumors from our hospital. We obtained a Dice score of 77.41% for unsupervised tumor segmentation, a genotype prediction accuracy (ACC) of 0.7639 and an area under curve (AUC) of 0.8600. Experimental results demonstrate that our domain adaptive approach outperforms the methods utilizing direct transfer learning. The model using hybrid features gives better results than the model using radiomics or deep features alone. Domain adaptation enables the segmentation network to achieve better performance, and the extraction of mixed features at multiple levels on the segmented region of interest ensures effective prediction of the IDH gene mutation status.

DCT-net

This paper introduces DCT-Net, a novel image translation architecture for few-shot portrait stylization. Given limited style exemplars (~100), the new architecture can produce high-quality style transfer results with advanced ability to synthesize high-fidelity contents and strong generality to handle complicated scenes (e.g., occlusions and accessories). Moreover, it enables full-body image translation via one elegant evaluation network trained by partial observations (i.e., stylized heads). Few-shot learning based style transfer is challenging since the learned model can easily become overfitted in the target domain, due to the biased distribution formed by only a few training examples. This paper aims to handle the challenge by adopting the key idea of "calibration first, translation later" and exploring the augmented global structure with locally-focused translation. Specifically, the proposed DCT-Net consists of three modules: a content adapter borrowing the powerful prior from source photos to calibrate the content distribution of target samples; a geometry expansion module using affine transformations to release spatially semantic constraints; and a texture translation module leveraging samples produced by the calibrated distribution to learn a fine-grained conversion. Experimental results demonstrate the proposed method's superiority over the state of the art in head stylization and its effectiveness on full image translation with adaptive deformations. Our code is publicly available at https://github.com/menyifang/DCT-Net.

Medical visual question answering based on question-type reasoning and semantic space constraint

Medical visual question answering (Med-VQA) aims to accurately answer clinical questions about medical images. Despite its enormous potential for application in the medical domain, the current technology is still in its infancy. Compared with general visual question answering task, Med-VQA task involve more demanding challenges. First, clinical questions about medical images are usually diverse due to different clinicians and the complexity of diseases. Consequently, noise is inevitably introduced when extracting question features. Second, Med-VQA task have always been regarded as a classification problem for predefined answers, ignoring the relationships between candidate responses. Thus, the Med-VQA model pays equal attention to all candidate answers when predicting answers. In this paper, a novel Med-VQA framework is proposed to alleviate the above-mentioned problems. Specifically, we employed a question-type reasoning module severally to closed-ended and open-ended questions, thereby extracting the important information contained in the questions through an attention mechanism and filtering the noise to extract more valuable question features. To take advantage of the relational information between answers, we designed a semantic constraint space to calculate the similarity between the answers and assign higher attention to answers with high correlation. To evaluate the effectiveness of the proposed method, extensive experiments were conducted on a public dataset, namely VQA-RAD. Experimental results showed that the proposed method achieved better performance compared to other the state-of-the-art methods. The overall accuracy, closed-ended accuracy, and open-ended accuracy reached 74.1 %, 82.7 %, and 60.9 %, respectively. It is worth noting that the absolute accuracy of the proposed method improved by 5.5 % for closed-ended questions.

ОСМИ ФОРУМ „БЪЛГАРСКА ГРАМАТИКА“ / EIGHT FORUM ON BULGARIAN GRAMMAR

The Eighth Forum on Bulgarian Grammar titled Predication, Predicates, Predicatives was held on 21 and 22 October 2021. The Forum was organised by the Institute for Bulgarian Language Prof. Lyubomir Andreychin at the Bulgarian Academy of Sciences and the Department of Bulgarian Language of the Faculty of Slavic Studies at Sofia University St. Kliment Ohridski. The forum was dedicated to the 90th anniversary of one of the most renowned Bulgarian linguists of our time – Prof. DSc Yordan Penchev Penchev. The Forum’s topic encompassed a broad range of theoretical and applied issues related to the lexical peculiarities, the semantic description and the classification of (selected classes of) predicates and predicatives, the description of grammatical categories involved in the realisation of predication, the study of the syntactic patterns of argument expression and the identification of relevant semantic and grammatical features. The main focus of the discussion was centred around (though not limited to) the following issues: • Lexical and semantic characterisation of various classes of predicates and predicatives • Approaches to the description of valency and argument selection • The expression of predication at the morphological level • Semantic and syntactic structure of sentences with predicates and predicatives • Description of semantic and grammatical constraints on subject, complement and adjunct realisation. Following the established tradition, the papers presented at the Forum are published in Balgarski ezik’s Supplement.

Research into an Image Inpainting Algorithm via Multilevel Attention Progression Mechanism

Existing image inpainting schemes generally have the problems of structural disorder and blurred texture details. This is mainly because, in the reconstruction process of the damaged area of the image, it is difficult for the inpainting network to make full use of the information in the nondamaged area to accurately infer the content of the damaged area. Therefore, the paper has proposed an image inpainting network driven by multilevel attention progression mechanism. The proposed network has compressed the high-level features extracted from the full-resolution image into multiscale compact features and then drives the compact features to perform multilevel order according to the scale size. Attention feature progression is to achieve the goal of the full progression of high-level features including structure and details in the network. To further realize fine-grained image inpainting and reconstruction, the paper has also proposed a composite granular discriminator to achieve image inpainting process performing global semantic constraints and nonspecific local dense constraints. The related experimental results in the paper can show that the proposed method can achieve higher quality repair results than state-of-the-art ones.

Satisfiability and containment of recursive SHACL

The Shapes Constraint Language (SHACL) is the recent W3C recommendation language for validating RDF data, by verifying certain shapes on graphs. Previous work has largely focused on the validation problem, while the standard decision problems of satisfiability and containment, crucial for design and optimisation purposes, have only been investigated for simplified versions of SHACL. Moreover, the SHACL specification does not define the semantics of recursively-defined constraints, which led to several alternative recursive semantics being proposed in the literature. The interaction between these different semantics and important decision problems has not been investigated yet. In this article we provide a comprehensive study of the different features of SHACL, by providing a translation to a new first-order language, called SCL, that precisely captures the semantics of SHACL. We also present MSCL, a second-order extension of SCL, which allows us to define, in a single formal logic framework, the main recursive semantics of SHACL. Within this language we also provide an effective treatment of filter constraints which are often neglected in the related literature. Using this logic we provide a detailed map of (un)decidability and complexity results for the satisfiability and containment decision problems for different SHACL fragments. Notably, we prove that both problems are undecidable for the full language, but we present decidable combinations of interesting features, even in the face of recursion.

Design of Neural Network Model for Cross-Media Audio and Video Score Recognition Based on Convolutional Neural Network Model.

In this paper, the residual convolutional neural network is used to extract the note features in the music score image to solve the problem of model degradation; then, multiscale feature fusion is used to fuse the feature information of different levels in the same feature map to enhance the feature representation ability of the model. A network composed of a bidirectional simple loop unit and a chained time series classification function is used to identify notes, parallelizing a large number of calculations, thereby speeding up the convergence speed of training, which also makes the data in the dataset no longer need to be strict with labels. Alignment also reduces the requirements on the dataset. Aiming at the problem that the existing cross-modal retrieval methods based on common subspace are insufficient for mining local consistency within modalities, a cross-modal retrieval method fused with graph convolution is proposed. The K-nearest neighbor algorithm is used to construct modal graphs for samples of different modalities, and the original features of samples from different modalities are encoded through a symmetric graph convolutional coding network and a symmetric multilayer fully connected coding network, and the encoded features are fused and input. We jointly optimize the intramodal semantic constraints and intermodal modality-invariant constraints in the common subspace to learn highly locally consistent and semantically consistent common representations for samples from different modalities. The error value of the experimental results is used to illustrate the effect of parameters such as the number of iterations and the number of neurons on the network. In order to more accurately illustrate that the generated music sequence is very similar to the original music sequence, the generated music sequence is also framed, and finally the music sequence spectrogram and spectrogram are generated. The accuracy of the experiment is illustrated by comparing the spectrogram and the spectrogram, and genre classification predictions are also performed on the generated music to show that the network can generate music of different genres.

Phonological effects on word order: AB constructions in Hebrew

The literature on what we call AB constructions (freezes, irreversible binomials), such as odds and ends and copy paste, attributes the fixed word order to both phonological and non-phonological, mostly semantic constraints. However, some researchers attribute a prominent role to phonology, while others view semantics as the major contributor to word order of AB constructions. In this paper we evaluate the role of phonology in Hebrew AB constructions with reference to a harmonic grammar with weighted constraints, where constraint weight is calculated on the basis of its effect in our corpus. The grammar reveals that semantic constraints weigh more than phonological constraints in both the cumulative weight and the average weight. Nevertheless, phonology affects a great number of data items, in particular those where semantic constraints are mute. We thus conclude that although syntax and semantics are responsible for word order, phonology determines word order when the other modules do not have a say.

On the big silent appositive time in dà NP de construction in Chinese: A compositional approach

This article explores the compositionality of [dà NP de] construction in Chinese. The NP should be a direct time-denoting NPtime or an indirect time-denoting NPitime. Following Kayne (2016), we assume there is a silent TIME following de. Meanwhile, a less-discussed use of de—that is, the equational/appositional use of de (Chao 1968), has been revived to link the silent TIME and NPtime/NPitime. This analysis of de helps explain why the NP is preferably time-related, as it is easier for a time-denoting NP to establish an equational/appositional relation with the silent TIME. Dà 'big' is neither a size-denoting adjective, nor an evaluative morpheme. It is analyzed as an element like the restrictive adjective very (e.g., the very book) in English. Dà modifies the appositive conjunction phrase [NP de TIME], bringing emphasis to the construction, which in turn imposes certain semantic and pragmatic constraints on its subsequent clauses.

Understanding Putative Should: A Semantic Approach

Chinese learners of English have much difficulty in understanding putative should. This article attempts to find out the cause of this difficulty. It makes a semantic study of sentences with putative should used in that-clauses, discusses five distinctions between them, and presents a corpus-assisted study of the semantic constraint of factuality accompanying the emotive should. It finds that the learning difficulty results from the fuzzy nature of the term ‘putative should’, which fails to adequately describe and explain the five differences. It argues that the teaching and learning of putative should should focus on understanding its two distinct uses, the suasive should and the emotive should, in nominal that-clauses. The suasive should denotes obligation, applies to something yet to come, and goes with suasive key words in sentences in whose that-clauses the present subjunctive can be used instead. The emotive should denotes surprise, applies to a personal, psychological, subjective fact, and goes with an emotive element, linguistic or extralinguistic, in sentences in whose that-clauses the indicative can be used instead but the present subjunctive cannot. The emotive element can take the form of an emotive word, a negative expression, or even the tone of voice.

CLEC-based Error Analysis of Collocations of Chinese English Learners

China’s voice in the international arena is gradually increasing in the present era of economic globalization. In order to further strengthen the multifaceted communication with the world, Chinese students must improve their English learning and application skills. However, due to the distinct difference between Chinese and English, and the lack of an authentic communication environment, English collocation has always been a significant difficulty for Chinese students in the process of learning English. As a result, pragmatic errors frequently occur in their usage. Aimed to address Chinese students’ difficulties in applying collocation, this article takes the Chinese Learner English Corpus (CLEC) as the research object and AntConc as the retrieval tool to collect the noun-noun collocation errors (CC1) in CLEC, then further analyses and interprets the data obtained. It’s discovered that learners’ usage errors in noun-noun collocation are mainly divided into the following three categories: grammar error, Chinese English expression, and overgeneralization. By further interpretation and analysis, it is reckoned that negative transfer of native language, interlanguage fossilization, second-language teaching methods, negligence of collocation structure, and semantic constraints are the main reasons for noun-noun collocation errors. In terms of the above situation, this article proposes corresponding countermeasures for second-language teaching methods and English learning to enrich the current research on English collocations, seeking to provide a reference for the use and teaching of noun-noun collocations.

A Multi-Stage Visible and Infrared Image Fusion Network Based on Attention Mechanism.

Pixel-level image fusion is an effective way to fully exploit the rich texture information of visible images and the salient target characteristics of infrared images. With the development of deep learning technology in recent years, the image fusion algorithm based on this method has also achieved great success. However, owing to the lack of sufficient and reliable paired data and a nonexistent ideal fusion result as supervision, it is difficult to design a precise network training mode. Moreover, the manual fusion strategy has difficulty ensuring the full use of information, which easily causes redundancy and omittance. To solve the above problems, this paper proposes a multi-stage visible and infrared image fusion network based on an attention mechanism (MSFAM). Our method stabilizes the training process through multi-stage training and enhances features by the learning attention fusion block. To improve the network effect, we further design a Semantic Constraint module and Push–Pull loss function for the fusion task. Compared with several recently used methods, the qualitative comparison intuitively shows more beautiful and natural fusion results by our model with a stronger applicability. For quantitative experiments, MSFAM achieves the best results in three of the six frequently used metrics in fusion tasks, while other methods only obtain good scores on a single metric or a few metrics. Besides, a commonly used high-level semantic task, i.e., object detection, is used to prove its greater benefits for downstream tasks compared with single-light images and fusion results by existing methods. All these experiments prove the superiority and effectiveness of our algorithm.

A random walk sampling on knowledge graphs for semantic-oriented statistical tasks

A knowledge graph (KG) manages large-scale and real-world facts as a big graph in a schema-flexible manner, which has recently attracted considerable attention . It is very common that users deploy some statistical tasks on a KG to achieve the latent information of interest. There are two types of the statistical tasks, that are, topology-oriented and semantic-oriented statistical tasks. Many efforts have been made for the former one (e.g., finding the average degree of a KG). The basic idea is concluded as: estimating an approximate statistical result based on a random sample collected through a topology-aware KG sampling approach. Unfortunately, this method cannot be directly deployed to support semantic-oriented statistical tasks (e.g., achieving the average fuel economy of cars produced in Germany), because the topology-aware sampling does not consider the semantics of a KG (or we say the sample is collected only based on the topological information of a KG, while excluding the semantics of a KG), hence leading to a low-quality random sample and would significantly affect the accuracy. In this paper, we propose a semantic-aware random walk sampling on KGs to quickly and accurately collect samples that match the semantic constraint of the semantic-oriented statistical task, and obtain an approximate statistical result by well-designed unbiased estimators. Moreover, we propose an optimization on our semantic-aware sampling to improve the sampling efficiency. Finally, extensive experiments were conducted on our method, which confirmed the effectiveness and efficiency of our approach.

Linguistic and Aesthetic Constraints in Literary Translation: Phonic Considerations in Translating Śabda-śakti-mūla dhvani

By outlining two features of language termed ‘lexical drift’ and ‘linguistic clogging’ and employing certain explanatory concepts of classical Indian aesthetics, this study addresses the question of how phonic elements of language constrain or smooth the ways of translation. Linguistic clogging, the study argues, constraints translation of expressions with śabda-śakti-mūla dhvani, SSMD (phonic-based suggestion) , in two ways. As a semantic constraint, the presence of SSMD in source expressions impedes translation if unsatisfied with target expressions. Secondly, as an aesthetic constraint, it curtails the aesthetic pleasure for readers even if the semantic constraint is satisfied by paraphrasing the source expressions. This study, analysing verses in English, Sanskrit, and Malayalam languages with SSMD, illustrates that if the constraints are not satisfied in the target expressions, they bring down the quality of translation. Nonetheless, satisfying both these constraints, by recovering the literal and suggested meanings where possible, can improve the quality of translation. Śabda-śakti-mūla dhvani (SSMD) is a sub-variety of dhvani where a figure of speech is suggested because of the inherent power of words. The meaning that is different from literal and metaphorical is termed “dhvani” (suggested meaning). Around the ninth century CE, Ānandavardhana articulated the philosophy of aesthetic suggestion and systematically theorised the concept of dhvani (suggestion) in his magnum opus Dhvanyāloka (Light on the Doctrine of Suggestion). He argues that the significative power of words is of two types, vācya (literal) and pratīyamāna (suggestive). SSMD is generally translated as “word-based suggestion”. In this paper, the author has translated SSMD as “phonic-based suggestion”, as sound elements distinguish SSMD from other varieties of dhvani.

Towards reliable cardiac image segmentation: Assessing image-level and pixel-level segmentation quality via self-reflective references.

Cardiac image segmentation is a fundamental step in cardiovascular disease diagnosis, where many deep learning models have achieved promising performance. However, when deploying these well-trained models for real clinical usage, the network will inevitably produce inferior results due to domain shifts, motion artifacts, etc. How to avoid the potential poor-quality segmentations involved in clinical decision making is crucial for reliable computer-aided cardiac disease diagnosis. To this end, we develop a quality control method to identify failure segmentations by measuring their qualities, and report them to physicians for professional opinions. In specific, we propose a reference-based framework to assess the image-level quality (i.e.per-class Dice) for overall evaluation and pixel-level quality (i.e.pixel-wise correct map) to locate mis-segmented regions. Following previous works, we create informative references first, and investigate their relative relationships (e.g.differences) to the inputs to expose segmentation failures. However, we generate and leverage the references in different ways. We instantiate the references by recovering input images from segmentations by a self-reflective reference generator. If the segmentation is of good quality, the reference (i.e.the reconstructed image) will be close to the input image, and the inconsistency between them would be a good indicator to deduce the qualities. To effectively explore these inconsistency, we employ a difference investigator equipped with semantic class-aware compactness constraint to force the correctly-segmented features more separable to the wrongly-segmented ones. The experiments on ACDC and MSCMR datasets demonstrated our method could effectively capture segmentation failures, and the results on low-quality (Dice∈[0,0.6)), medium-quality (Dice∈[0.6,0.8)) and high-quality (Dice∈[0.8,1.0)) segmentations showed satisfying robustness of our method.

Age of Acquisition Modulates Alpha Power During Bilingual Speech Comprehension in Noise.

Research on bilingualism has grown exponentially in recent years. However, the comprehension of speech in noise, given the ubiquity of both bilingualism and noisy environments, has seen only limited focus. Electroencephalogram (EEG) studies in monolinguals show an increase in alpha power when listening to speech in noise, which, in the theoretical context where alpha power indexes attentional control, is thought to reflect an increase in attentional demands. In the current study, English/French bilinguals with similar second language (L2) proficiency and who varied in terms of age of L2 acquisition (AoA) from 0 (simultaneous bilinguals) to 15 years completed a speech perception in noise task. Participants were required to identify the final word of high and low semantically constrained auditory sentences such as “Stir your coffee with a spoon” vs. “Bob could have known about the spoon” in both of their languages and in both noise (multi-talker babble) and quiet during electrophysiological recording. We examined the effects of language, AoA, semantic constraint, and listening condition on participants’ induced alpha power during speech comprehension. Our results show an increase in alpha power when participants were listening in their L2, suggesting that listening in an L2 requires additional attentional control compared to the first language, particularly early in processing during word identification. Additionally, despite similar proficiency across participants, our results suggest that under difficult processing demands, AoA modulates the amount of attention required to process the second language.