Semantic Similarity Research Articles

From word clouds to Word Rain: Revisiting the classic word cloud to visualize climate change texts

Word Rain is a development of the classic word cloud. It addresses some of the limitations of word clouds, in particular the lack of a semantically motivated positioning of the words, and the use of font size as a sole indicator of word prominence. Word Rain uses the semantic information encoded in a distributional semantics-based language model – reduced into one dimension – to position the words along the x-axis. Thereby, the horizontal positioning of the words reflects semantic similarity. Font size is still used to signal word prominence, but this signal is supplemented with a bar chart, as well as with the position of the words on the y-axis. We exemplify the use of Word Rain by three concrete visualization tasks, applied on different real-world texts and document collections on climate change. In these case studies, word2vec models, reduced to one dimension with t-SNE, are used to encode semantic similarity, and TF-IDF is used for measuring word prominence. We evaluate the technique further by carrying out domain expert reviews.

Automatic Scoring of Metaphor Creativity with Large Language Models

ABSTRACT Metaphor is crucial in human cognition and creativity, facilitating abstract thinking, analogical reasoning, and idea generation. Typically, human raters manually score the originality of responses to creative thinking tasks – a laborious and error-prone process. Previous research sought to remedy these risks by scoring creativity tasks automatically using semantic distance and large language models (LLMs). Here, we extend research on automatic creativity scoring to metaphor generation – the ability to creatively describe episodes and concepts using nonliteral language. Metaphor is arguably more abstract and naturalistic than prior targets of automated creativity assessment. We collected 4,589 responses from 1,546 participants to various metaphor prompts and corresponding human creativity ratings. We fine-tuned two open-source LLMs (RoBERTa and GPT-2) – effectively “teaching” them to score metaphors like humans – before testing their ability to accurately assess the creativity of new metaphors. Results showed both models reliably predicted new human creativity ratings (RoBERTa r = .72, GPT-2 r = .70), significantly more strongly than semantic distance (r = .42). Importantly, the fine-tuned models generalized accurately to metaphor prompts they had not been trained on (RoBERTa r = .68, GPT-2 r = .63). We provide open access to the fine-tuned models, allowing researchers to assess metaphor creativity in a reproducible and timely manner.

Creativity on Tap? The Effect of Creativity Anxiety Under Evaluative Pressure

ABSTRACT Creative achievement is critical for education, work and wider society. Creativity anxiety – a distinct, generalizable form of anxiety that emerges at the prospect of having to be creative – is a newly established trait that varies between individuals and has been linked to lower creative achievement. This study tested the interaction between trait creativity anxiety and creative evaluative pressure on a well-established objective measure linked to creative ability – semantic distance. Participants performed a word association task under two instruction conditions designed to manipulate creative evaluative pressure. Participants were first instructed to respond with the first word that came to mind, before repeating the task with the added information that the task was measuring creativity. The results revealed that creative pressure increased semantic distance between cues and responses, similar to the established “be creative” effect. Critically, this “be creative” effect was reduced among participants with higher levels of creativity anxiety. The findings are hence consistent with the proposal that creativity anxiety hinders people’s ability to think creatively on demand. These findings will guide further research on creativity anxiety and inform intervention strategies to unlock people’s creative potential.

DenoSent: A Denoising Objective for Self-Supervised Sentence Representation Learning

Contrastive-learning-based methods have dominated sentence representation learning. These methods regularize the representation space by pulling similar sentence representations closer and pushing away the dissimilar ones and have been proven effective in various NLP tasks, e.g., semantic textual similarity (STS) tasks. However, it is challenging for these methods to learn fine-grained semantics as they only learn from the inter-sentence perspective, i.e., their supervision signal comes from the relationship between data samples. In this work, we propose a novel denoising objective that inherits from another perspective, i.e., the intra-sentence perspective. By introducing both discrete and continuous noise, we generate noisy sentences and then train our model to restore them to their original form. Our empirical evaluations demonstrate that this approach delivers competitive results on both semantic textual similarity (STS) and a wide range of transfer tasks, standing up well in comparison to contrastive-learning-based methods. Notably, the proposed intra-sentence denoising objective complements existing inter-sentence contrastive methodologies and can be integrated with them to further enhance performance. Our code is available at https://github.com/xinghaow99/DenoSent.

HiHPQ: Hierarchical Hyperbolic Product Quantization for Unsupervised Image Retrieval

Existing unsupervised deep product quantization methods primarily aim for the increased similarity between different views of the identical image, whereas the delicate multi-level semantic similarities preserved between images are overlooked. Moreover, these methods predominantly focus on the Euclidean space for computational convenience, compromising their ability to map the multi-level semantic relationships between images effectively. To mitigate these shortcomings, we propose a novel unsupervised product quantization method dubbed Hierarchical Hyperbolic Product Quantization (HiHPQ), which learns quantized representations by incorporating hierarchical semantic similarity within hyperbolic geometry. Specifically, we propose a hyperbolic product quantizer, where the hyperbolic codebook attention mechanism and the quantized contrastive learning on the hyperbolic product manifold are introduced to expedite quantization. Furthermore, we propose a hierarchical semantics learning module, designed to enhance the distinction between similar and non-matching images for a query by utilizing the extracted hierarchical semantics as an additional training supervision. Experiments on benchmark image datasets show that our proposed method outperforms state-of-the-art baselines.

VLM2Scene: Self-Supervised Image-Text-LiDAR Learning with Foundation Models for Autonomous Driving Scene Understanding

Vision and language foundation models (VLMs) have showcased impressive capabilities in 2D scene understanding. However, their latent potential in elevating the understanding of 3D autonomous driving scenes remains untapped. In this paper, we propose VLM2Scene, which exploits the potential of VLMs to enhance 3D self-supervised representation learning through our proposed image-text-LiDAR contrastive learning strategy. Specifically, in the realm of autonomous driving scenes, the inherent sparsity of LiDAR point clouds poses a notable challenge for point-level contrastive learning methods. This method often grapples with limitations tied to a restricted receptive field and the presence of noisy points. To tackle this challenge, our approach emphasizes region-level learning, leveraging regional masks without semantics derived from the vision foundation model. This approach capitalizes on valuable contextual information to enhance the learning of point cloud representations. First, we introduce Region Caption Prompts to generate fine-grained language descriptions for the corresponding regions, utilizing the language foundation model. These region prompts then facilitate the establishment of positive and negative text-point pairs within the contrastive loss framework. Second, we propose a Region Semantic Concordance Regularization, which involves a semantic-filtered region learning and a region semantic assignment strategy. The former aims to filter the false negative samples based on the semantic distance, and the latter mitigates potential inaccuracies in pixel semantics, thereby enhancing overall semantic consistency. Extensive experiments on representative autonomous driving datasets demonstrate that our self-supervised method significantly outperforms other counterparts. Codes are available at https://github.com/gbliao/VLM2Scene.

Unsupervised Cross-Domain Image Retrieval via Prototypical Optimal Transport

Unsupervised cross-domain image retrieval (UCIR) aims to retrieve images sharing the same category across diverse domains without relying on labeled data. Prior approaches have typically decomposed the UCIR problem into two distinct tasks: intra-domain representation learning and cross-domain feature alignment. However, these segregated strategies overlook the potential synergies between these tasks. This paper introduces ProtoOT, a novel Optimal Transport formulation explicitly tailored for UCIR, which integrates intra-domain feature representation learning and cross-domain alignment into a unified framework. ProtoOT leverages the strengths of the K-means clustering method to effectively manage distribution imbalances inherent in UCIR. By utilizing K-means for generating initial prototypes and approximating class marginal distributions, we modify the constraints in Optimal Transport accordingly, significantly enhancing its performance in UCIR scenarios. Furthermore, we incorporate contrastive learning into the ProtoOT framework to further improve representation learning. This encourages local semantic consistency among features with similar semantics, while also explicitly enforcing separation between features and unmatched prototypes, thereby enhancing global discriminativeness. ProtoOT surpasses existing state-of-the-art methods by a notable margin across benchmark datasets. Notably, on DomainNet, ProtoOT achieves an average P@200 enhancement of 24.44%, and on Office-Home, it demonstrates a P@15 improvement of 12.12%. Code is available at https://github.com/HCVLAB/ProtoOT.

Collaborative Weakly Supervised Video Correlation Learning for Procedure-Aware Instructional Video Analysis

Video Correlation Learning (VCL), which aims to analyze the relationships between videos, has been widely studied and applied in various general video tasks. However, applying VCL to instructional videos is still quite challenging due to their intrinsic procedural temporal structure. Specifically, procedural knowledge is critical for accurate correlation analyses on instructional videos. Nevertheless, current procedure-learning methods heavily rely on step-level annotations, which are costly and not scalable. To address this problem, we introduce a weakly supervised framework called Collaborative Procedure Alignment (CPA) for procedure-aware correlation learning on instructional videos. Our framework comprises two core modules: collaborative step mining and frame-to-step alignment. The collaborative step mining module enables simultaneous and consistent step segmentation for paired videos, leveraging the semantic and temporal similarity between frames. Based on the identified steps, the frame-to-step alignment module performs alignment between the frames and steps across videos. The alignment result serves as a measurement of the correlation distance between two videos. We instantiate our framework in two distinct instructional video tasks: sequence verification and action quality assessment. Extensive experiments validate the effectiveness of our approach in providing accurate and interpretable correlation analyses for instructional videos.

Bridging the Semantic Latent Space between Brain and Machine: Similarity Is All You Need

How our brain encodes complex concepts has been a longstanding mystery in neuroscience. The answer to this problem can lead to new understandings about how the brain retrieves information in large-scale data with high efficiency and robustness. Neuroscience studies suggest the brain represents concepts in a locality-sensitive hashing (LSH) strategy, i.e., similar concepts will be represented by similar responses. This finding has inspired the design of similarity-based algorithms, especially in contrastive learning. Here, we hypothesize that the brain and large neural network models, both using similarity-based learning rules, could contain a similar semantic embedding space. To verify that, this paper proposes a functional Magnetic Resonance Imaging (fMRI) semantic learning network named BrainSem, aimed at seeking a joint semantic latent space that bridges the brain and a Contrastive Language-Image Pre-training (CLIP) model. Given that our perception is inherently cross-modal, we introduce a fuzzy (one-to-many) matching loss function to encourage the models to extract high-level semantic components from neural signals. Our results claimed that using only a small set of fMRI recordings for semantic space alignment, we could obtain shared embedding valid for unseen categories out of the training set, which provided potential evidence for the semantic representation similarity between the brain and large neural networks. In a zero-shot classification task, our BrainSem achieves an 11.6% improvement over the state-of-the-art.

A Unified Knowledge Transfer Network for Generalized Category Discovery

Generalized Category Discovery (GCD) aims to recognize both known and novel categories in an unlabeled dataset by leveraging another labeled dataset with only known categories. Without considering knowledge transfer from known to novel categories, current methods usually perform poorly on novel categories due to the lack of corresponding supervision. To mitigate this issue, we propose a unified Knowledge Transfer Network (KTN), which solves two obstacles to knowledge transfer in GCD. First, the mixture of known and novel categories in unlabeled data makes it difficult to identify transfer candidates (i.e., samples with novel categories). For this, we propose an entropy-based method that leverages knowledge in the pre-trained classifier to differentiate known and novel categories without requiring extra data or parameters. Second, the lack of prior knowledge of novel categories presents challenges in quantifying semantic relationships between categories to decide the transfer weights. For this, we model different categories with prototypes and treat their similarities as transfer weights to measure the semantic similarities between categories. On the basis of two treatments, we transfer knowledge from known to novel categories by conducting pre-adjustment of logits and post-adjustment of labels for transfer candidates based on the transfer weights between different categories. With the weighted adjustment, KTN can generate more accurate pseudo-labels for unlabeled data, which helps to learn more discriminative features and boost model performance on novel categories. Extensive experiments show that our method outperforms state-of-the-art models on all evaluation metrics across multiple benchmark datasets. Furthermore, different from previous clustering-based methods that can only work offline with abundant data, KTN can be deployed online conveniently with faster inference speed. Code and data are available at https://github.com/yibai-shi/KTN.

Semantic-Guided Novel Category Discovery

The Novel Category Discovery problem aims to cluster an unlabeled set with the help of a labeled set consisting of disjoint but related classes. However, existing models treat class names as discrete one-hot labels and ignore the semantic understanding of these classes. In this paper, we propose a new setting named Semantic-guided Novel Category Discovery (SNCD), which requires the model to not only cluster the unlabeled images but also semantically recognize these images based on a set of their class names. The first challenge we confront pertains to effectively leveraging the class names of unlabeled images, given the inherent gap between the visual and linguistic domains. To address this issue, we incorporate a semantic-aware recognition mechanism. This is achieved by constructing dynamic class-wise visual prototypes as well as a semantic similarity matrix that enables the projection of visual features into the semantic space. The second challenge originates from the granularity disparity between the classification and clustering tasks. To deal with this, we develop a semantic-aware clustering process to facilitate the exchange of knowledge between the two tasks. Through extensive experiments, we demonstrate the mutual benefits of the recognition and clustering tasks, which can be jointly optimized. Experimental results on multiple datasets confirm the effectiveness of our proposed method. Our code is available at https://github.com/wang-weishuai/Semantic-guided-NCD.

Causal Adversarial Perturbations for Individual Fairness and Robustness in Heterogeneous Data Spaces

As responsible AI gains importance in machine learning algorithms, properties like fairness, adversarial robustness, and causality have received considerable attention in recent years. However, despite their individual significance, there remains a critical gap in simultaneously exploring and integrating these properties. In this paper, we propose a novel approach that examines the relationship between individual fairness, adversarial robustness, and structural causal models (SCMs) in heterogeneous data spaces, particularly when dealing with discrete sensitive attributes. We use SCMs and sensitive attributes to create a fair metric and apply it to measure semantic similarity among individuals. By introducing a novel causal adversarial perturbation (CAP) and applying adversarial training, we create a new regularizer that combines individual fairness, causality, and robustness in the classifier. Our method is evaluated on both real-world and synthetic datasets, demonstrating its effectiveness in achieving an accurate classifier that simultaneously exhibits fairness, adversarial robustness, and causal awareness.

SALSA: Semantically-Aware Latent Space Autoencoder

In deep learning for drug discovery, molecular representations are often based on sequences, known as SMILES, which allow for straightforward implementation of natural language processing methodologies, one being the sequence-to-sequence autoencoder. However, we observe that training an autoencoder solely on SMILES is insufficient to learn molecular representations that are semantically meaningful, where semantics are specified by the structural (graph-to-graph) similarities between molecules. We demonstrate by example that SMILES-based autoencoders may map structurally similar molecules to distant codes, resulting in an incoherent latent space that does not necessarily respect the semantic similarities between molecules. To address this shortcoming we propose Semantically-Aware Latent Space Autoencoder (SALSA) for molecular representations: a SMILES-based transformer autoencoder modified with a contrastive task aimed at learning graph-to-graph similarities between molecules. To accomplish this, we develop a novel dataset comprised of sets of structurally similar molecules and opt for a supervised contrastive loss that is able to incorporate full sets of positive samples. We evaluate semantic awareness of SALSA representations by comparing to its ablated counterparts, and show empirically that SALSA learns representations that maintain 1) structural awareness, 2) physicochemical awareness, 3) biological awareness, and 4) semantic continuity.

Mining Gaze for Contrastive Learning toward Computer-Assisted Diagnosis

Obtaining large-scale radiology reports can be difficult for medical images due to ethical concerns, limiting the effectiveness of contrastive pre-training in the medical image domain and underscoring the need for alternative methods. In this paper, we propose eye-tracking as an alternative to text reports, as it allows for the passive collection of gaze signals without ethical issues. By tracking the gaze of radiologists as they read and diagnose medical images, we can understand their visual attention and clinical reasoning. When a radiologist has similar gazes for two medical images, it may indicate semantic similarity for diagnosis, and these images should be treated as positive pairs when pre-training a computer-assisted diagnosis (CAD) network through contrastive learning. Accordingly, we introduce the Medical contrastive Gaze Image Pre-training (McGIP) as a plug-and-play module for contrastive learning frameworks. McGIP uses radiologist gaze to guide contrastive pre-training. We evaluate our method using two representative types of medical images and two common types of gaze data. The experimental results demonstrate the practicality of McGIP, indicating its high potential for various clinical scenarios and applications.

Beyond Prototypes: Semantic Anchor Regularization for Better Representation Learning

One of the ultimate goals of representation learning is to achieve compactness within a class and well-separability between classes. Many outstanding metric-based and prototype-based methods following the Expectation-Maximization paradigm, have been proposed for this objective. However, they inevitably introduce biases into the learning process, particularly with long-tail distributed training data. In this paper, we reveal that the class prototype is not necessarily to be derived from training features and propose a novel perspective to use pre-defined class anchors serving as feature centroid to unidirectionally guide feature learning. However, the pre-defined anchors may have a large semantic distance from the pixel features, which prevents them from being directly applied. To address this issue and generate feature centroid independent from feature learning, a simple yet effective Semantic Anchor Regularization (SAR) is proposed. SAR ensures the inter-class separability of semantic anchors in the semantic space by employing a classifier-aware auxiliary cross-entropy loss during training via disentanglement learning. By pulling the learned features to these semantic anchors, several advantages can be attained: 1) the intra-class compactness and naturally inter-class separability, 2) induced bias or errors from feature learning can be avoided, and 3) robustness to the long-tailed problem. The proposed SAR can be used in a plug-and-play manner in the existing models. Extensive experiments demonstrate that the SAR performs better than previous sophisticated prototype-based methods. The implementation is available at https://github.com/geyanqi/SAR.

Automated linguistic analysis in speech samples of Turkish-speaking patients with schizophrenia-spectrum disorders

Modern natural language processing (NLP) methods provide ways to objectively quantify language disturbances for potential use in diagnostic classification. We performed computerized language analysis in speech samples of 82 Turkish-speaking subjects, including 44 patients with schizophrenia spectrum disorders (SSD) and 38 healthy controls (HC). Exploratory analysis of speech samples involved 16 sentence-level semantic similarity features using SBERT (Sentence Bidirectional Encoder Representation from Text) as well as 8 generic and 8 part-of-speech (POS) features. The random forest classifier using SBERT-derived semantic similarity features achieved a mean accuracy of 85.6 % for the classification of SSD and HC. When semantic similarity features were combined with generic and POS features, the classifier's mean accuracy reached to 86.8 %. Our analysis reflected increased sentence-level semantic similarity scores in SSD. Generic and POS analyses revealed an increase in the use of verbs, proper nouns and pronouns in SSD while our results showed a decrease in the utilization of conjunctions, determiners, and both average and maximum sentence length in SSD compared to HC. Quantitative language features were correlated with the expressive deficit domain of BNSS (Brief Negative Symptom Scale) as well as with the duration of illness. These findings from Turkish-speaking interviews contribute to the growing evidence-based NLP-derived assessments in non-English-speaking patients.

FSD-CLCD: Functional semantic distillation graph learning for cross-language code clone detection

Code clone detection can find similar or the same code snippets, which is important in analyzing homologous components, discovering redundant code, and improving software system development and maintenance efficiency. A crucial challenge is to extract more functional semantic similarity from code in heterogeneous conditions, such as a cross-language scenario. Existing methods mainly exploit sequence models with only lexical and statistical features to compare code pairs, which are susceptible to linguistic feature noise and misclassify code pairs that have similar structure dependencies such as control flow. Meanwhile, there are issues with inconsistent node types and a great variation of node numbers while capturing structure-dependent features, resulting in a misaligned distribution of clone pairs, and weakening the detection precision. This work presents a novel cross-language code clone detection method. It represents code with a graph structure based on abstract syntax trees and introduces a global node to strengthen the connection between control flows. Prune the graph structure based on key node protection rules to reduce the impact of linguistic feature noise. Besides, optimize graph matching networks for cross-language abstract syntax trees by using contrastive loss to align the functional semantic distribution of clone pairs. The method distills the invariant functional semantic similarity with a huge discrepancy of the code graph in heterogeneous cross-language conditions. Experiment results show that the proposed method achieves scores of 0.95, 0.98, and 0.96 in terms of precision, recall and F1-score and substantially outperforms the state-of-the-art baselines.

Age differences in generalization, memory specificity, and their overnight fate in childhood.

Memory enables generalization to new situations, and memory specificity that preserves individual episodes. This study investigated generalization, memory specificity, and their overnight fate in 141 4- to 8-year-olds (computerized memory game; 71 females, tested 2020-2021 in Germany). The results replicated age effects in generalization and memory specificity, and a contingency of generalization on object conceptual properties and interobject semantic proximity. Age effects were stronger in generalization than in memory specificity, and generalization was more closely linked to the explicit regularity knowledge in older than in younger children. After an overnight delay, older children retained more generalized and specific memories and showed greater gains but only in generalization. These findings reveal distinct age differences in generalization and memory specificity across childhood.

MeSH-Based Semantic Weighting Scheme to Enhance Document Indexing: Application on Biomedical Document Classification

Document indexing phase plays a significant role in text mining applications such as text document classification. The common indexing paradigm is based on terms frequency in documents known as Bag Of Words (BOW)-based representation approach. However, such classical approach suffers from ambiguity and disparity of words. In addition, traditional term weighting schemes, such as TF-IDF, exploit only the statistical information of terms in documents. To overcome these problems, we have been interested in biomedical semantic document indexing using concepts extracted from the knowledge resource MeSH. Thus, we have focused first on a disambiguation method to identify the adequate senses of ambiguous MeSH concepts and we have considered four representation enrichment strategies to identify the best appropriate representatives of the adequate sense in the textual entities representation. Second, we propose to introduce a semantic weighting scheme that quantifies MeSH concept’s importance in documents through their occurrence frequency and semantic similarities with unambiguous MeSH concepts. Our contribution lies particularly in the in-depth experimental study of the performance of these methods and precisely the impact of the semantic weighting scheme on the performance. To do that, three benchmark datasets TREC 2004 genomics, BioCreative II and OHSUMED were used.

GRACE: Empowering LLM-based software vulnerability detection with graph structure and in-context learning

Software vulnerabilities inflict considerable economic and societal harm. Therefore, timely and accurate detection of these flaws has become vital. Large language models (LLMs) have emerged as a promising tool for vulnerability detection in recent studies. However, their effectiveness suffers when limited to plain text source code, which may ignore the syntactic and semantic information of the code. To address this limitation, we propose a novel vulnerability detection approach GRACE that empowers LLM-based software vulnerability detection by incorporating graph structural information in the code and in-context learning. We also design an effective demonstration retrieval approach that identifies highly relevant code examples by considering semantic, lexical, and syntactic similarities for the target code to provide better demonstrations for in-context learning. To evaluate the effectiveness of GRACE, we conducted an empirical study on three vulnerability detection datasets (i.e., Devign, Reveal, and Big-Vul). The results demonstrate that GRACE outperforms six state-of-the-art vulnerability detection baselines by at least 28.65% in terms of the F1 score across these three datasets. Therefore, our study highlights the effectiveness of integrating graph structural information and in-context learning in LLMs for vulnerability detection. These findings motivate further investigation into tailoring such approaches for specific vulnerability types or adapting them to other security tasks.