Search Results (30)

Fake news has eroded trust in credible news sources, driving the need for tools to verify the accuracy of circulating information. Fact verification addresses this issue by classifying claims as Supports (S), Refutes (R), or Not Enough Info (NEI) based on evidence. Neural Semantic Matching Networks (NSMN) is an algorithm designed for this purpose, but its reliance on BiLSTM has shown limitations, particularly overfitting. This study aims to enhance NSMN for fact verification through a structured framework comprising encoding, alignment, matching, and output layers. The proposed approach employed Siamese MaLSTM in the matching layer and introduced the Manhattan Fact Relatedness Score (MFRS) in the output layer, culminating in a novel algorithm called Deep One-Directional Neural Semantic Siamese Network (DOD–NSSN). Performance evaluation compared DOD–NSSN with NSMN and transformer-based algorithms (BERT, RoBERTa, XLM, XL-Net). Results demonstrated that DOD–NSSN achieved 91.86% accuracy and consistently outperformed other models, achieving over 95% accuracy across diverse topics, including sports, government, politics, health, and industry. The findings highlight the DOD–NSSN model’s capability to generalize effectively across various domains, providing a robust tool for automated fact verification. Full article

Music has been shown to increase arousal and attention and even facilitate processing during non-musical tasks, including those related to speech and language functions. Mind wandering has been studied in many sustained attention tasks. Here, we investigated the intersection of these two phenomena: the role of mind wandering while listening to familiar/unfamiliar musical excerpts, and its effects on concurrent linguistic processing. We hypothesized that familiar music would be less distracting than unfamiliar music, causing less mind wandering, and consequently benefit concurrent speech perception. Participants (N = 96 young adults) performed a lexical-semantic congruity task where they judged the relatedness of visually presented word pairs while listening to non-vocal classical music (familiar or unfamiliar orchestral pieces), or a non-music environmental sound clip (control) played in the background. Mind wandering episodes were probed intermittently during the task by explicitly asking listeners if their mind was wandering in that moment. The primary outcome was accuracy and reactions times measured during the lexical-semantic judgment task across the three background music conditions (familiar, unfamiliar, and control). We found that listening to familiar music, relative to unfamiliar music or environmental noise, was associated with faster lexical-semantic decisions and a lower incidence of mind wandering. Mind wandering frequency was similar when performing the task when listening to familiar music and control environmental sounds. We infer that familiar music increases task enjoyment, reduces mind wandering, and promotes more rapid lexical access during concurrent lexical processing, by modulating task-related attentional resources. The implications of using music as an aid during academic study and cognitive tasks are discussed. Full article

Transformers have made a significant breakthrough in natural language processing. These models are trained on large datasets and can handle multiple tasks. We compare monolingual and multilingual transformer models for semantic relatedness and verse retrieval. We leveraged data from the original QurSim dataset (Arabic) and used authentic multi-author translations in 22 languages to create a multilingual QurSim dataset, which we released for the research community. We evaluated the performance of monolingual and multilingual LLMs for Arabic and our results show that monolingual LLMs give better results for verse classification and matching verse retrieval. We incrementally built monolingual models with Arabic, English, and Urdu and multilingual models with all 22 languages supported by the multilingual paraphrase-MiniLM-L12-v2 model. Our results show improvement in classification accuracy with the incorporation of multilingual QurSim. Full article

Consumer-grade EEG devices, such as the InteraXon Muse 2 headband, present a promising opportunity to enhance the accessibility and inclusivity of neuroscience research. However, their effectiveness in capturing language-related ERP components, such as the N400, remains underexplored. This study thus aimed to investigate the feasibility of using the Muse 2 to measure the N400 effect in a semantic relatedness judgment task. Thirty-seven participants evaluated the semantic relatedness of word pairs while their EEG was recorded using the Muse 2. Single-trial ERPs were analyzed using robust Yuen t-tests and hierarchical linear modeling (HLM) to assess the N400 difference between semantically related and unrelated target words. ERP analyses indicated a significantly larger N400 effect in response to unrelated word pairs over the right frontal electrode. Additionally, dependability estimates suggested acceptable internal consistency for the N400 data. Overall, these findings illustrate the capability of the Muse 2 to reliably measure the N400 effect, reinforcing its potential as a valuable tool for language research. This study highlights the potential of affordable, wearable EEG technology to expand access to brain research by offering an affordable and portable way to study language and cognition in diverse populations and settings. Full article

Previous research has established that bilinguals automatically activate lexical items in both of their languages in a nonselectivemanner, even when processing linguistic information in the second language (L2) alone. However, whether this co-activation extends to the sub-lexical level remains debated. In this study, we investigate whether bilinguals access sub-lexical information while processing in their L2. Thirty-two Chinese–English bilinguals and thirty-one English monolinguals completed an EEG-based semantic relatedness task, during which they judged whether pairs of English words were related in meaning or not (±S). Unbeknownst to the participants, the form (±F) of the Chinese translations in half of the pairs shared a sub-lexical semantic radical. This leads to four conditions: +S+F, +S−F, −S+F, and −S−F. This design, along with the comparison to English monolinguals, allows us to examine if bilinguals’ native language is activated at the sub-lexical level when they are exposed only to L2. The results revealed that both groups showed sensitivity to semantic relatedness, as evidenced by a greater N400 for semantic unrelated pairs than related pairs, with monolinguals eliciting a more pronounced difference. Bilinguals, on the other hand, exhibited a greater P200 difference compared to monolinguals, indicating greater sensitivity to the hidden Chinese radical/form manipulation. These results suggest that highly proficient bilinguals automatically engage in lexical co-activation of their native language during L2 processing. Crucially, this co-activation extends to the sub-lexical semantic radical level. Full article

Lexical substitution aims to generate a list of equivalent substitutions (i.e., synonyms) to a sentence’s target word or phrase while preserving the sentence’s meaning to improve writing, enhance language understanding, improve natural language processing models, and handle ambiguity. This task has recently attracted much attention in many languages. Despite the richness of Arabic vocabulary, limited research has been performed on the lexical substitution task due to the lack of annotated data. To bridge this gap, we present the first Arabic lexical substitution benchmark dataset AraLexSubD for benchmarking lexical substitution pipelines. AraLexSubD is manually built by eight native Arabic speakers and linguists (six linguist annotators, a doctor, and an economist) who annotate the 630 sentences. AraLexSubD covers three domains: general, finance, and medical. It encompasses 2476 substitution candidates ranked according to their semantic relatedness. We also present the first Arabic lexical substitution pipeline, AraLexSub, which uses the AraBERT pre-trained language model. The pipeline consists of several modules: substitute generation, substitute filtering, and candidate ranking. The filtering step shows its effectiveness by achieving an increase of 1.6 in the F1 score on the entire AraLexSubD dataset. Additionally, an error analysis of the experiment is reported. To our knowledge, this is the first study on Arabic lexical substitution. Full article

The morpheme ‘give’ is among the most well-studied lexical items in the realm of grammaticalization. This study sets out to provide a typological and areal analysis of the distinct forms and multiple functions of ‘give’ in 27 varieties of Hui Chinese, a lesser-known group of Sinitic languages. Making use of both primary and secondary data, we have identified ten different functions of GIVE, namely (i) lexical verb ‘give’, (ii) recipient marker ‘to’, (iii) benefactive marker ‘for’, (iv) purpose marker, (v) permissive marker, (vi) passive marker, (vii) pretransitive disposal marker, (viii) allative marker, (ix) locative marker ‘at/in’, and (x) temporal marker ‘till’. The Hui varieties covered in this study generally showcase the syncretism of a minimum of five of the functions above simultaneously. Semantic extension, polygrammaticalization, and cooptation are shown to be the major mechanisms behind the polyfunctionality or polysemy sharing of the morpheme ‘give’. Our study contributes to the understanding of the role that grammaticalization, especially contact-induced grammaticalization, plays in forming linguistic areas. In addition, it casts doubt on the basicness of ‘give’ in assessing the genetic relatedness of languages in the world. Full article

Grammatical gender as a lexico-syntactic feature has been well explored, and the gender congruency effect has been observed in many languages (e.g., Dutch, German, Croatian, Czech, etc.). Yet, so far, this effect has not been found in Romance languages such as Italian, French, and Spanish. It has been argued that the absence of the effect in Romance languages is due the fact that the gender-marking definite article is not exclusively dependent on the grammatical gender of the head noun, but also on its onset phonology (e.g., lo zucchero is ‘the sugar’ in Italian, not il zucchero, il being the default masculine determiner in Italian). For Spanish, this argument has also been made because feminine words starting with a stressed /a/ take the masculine article (e.g., el água is ‘the water’, not la água). However, the number of words belonging to that set is rather small in Spanish, and it may be questionable whether or not this feature can be taken as an argument for the absence of a gender congruency effect in Spanish. In this study, we investigated the gender congruency effect in native Spanish noun phrase production. We measured 30 native Spanish speakers’ naming latencies in four conditions via the picture–word interference paradigm by manipulating gender congruency (i.e., gender-congruent vs. gender-incongruent) and semantic relatedness (i.e., semantically related vs. semantically unrelated). The results revealed significantly longer naming latencies in gender-incongruent and semantically related conditions compared to gender-congruent and semantically unrelated conditions. This result suggests that grammatical gender as a lexico-syntactic feature in Spanish is used to competitively select determiners in native Spanish speakers’ noun phrases. Our findings provide an important behavioral piece of evidence for the gender congruency effect in Romance languages. Full article

Measuring ontology matching is a critical issue in knowledge engineering and supports knowledge sharing and knowledge evolution. Recently, linguistic scientists have defined semantic relatedness as being more significant than semantic similarities in measuring ontology matching. Semantic relatedness is measured using synonyms and hypernym–hyponym relationships. In this paper, a systematic approach for measuring ontology semantic relatedness is proposed. The proposed approach is developed with a clear and fully described methodology, with illustrative examples used to demonstrate the proposed approach. The relatedness between ontologies has been measured based on class level by using lexical features, defining semantic similarity of concepts based on hypernym–hyponym relationships. For evaluating our proposed approach against similar works, benchmarks are generated using five properties: related meaning features, lexical features, providing technical descriptions, proving applicability, and accuracy. Technical implementation is carried out in order to demonstrate the applicability of our approach. The results demonstrate an achieved accuracy of 99%. The contributions are further highlighted by benchmarking against recent related works. Full article

This study investigates the extent to which highly proficient Spanish–Catalan bilinguals activate Spanish translation equivalents when they are presented with Catalan words. Participants performed a translation recognition task (Experiment 1) or a primed lexical decision task (Experiment 2) where the relationship between the first presented (Catalan) word and the second presented (Spanish) word was manipulated. Semantic and form relationships between the first and the second words were examined. Semantic relatedness produced a behavioral interference effect in the translation recognition task and a facilitation effect in the primed lexical decision task. The semantic manipulation also affected the N400 component. Form relatedness produced a behavioral interference effect only in the translation recognition task, which was accompanied by a modulation of the LPC component. In contrast, there were no effects of the formal manipulation in the primed lexical decision task. These results, which are discussed in relation to the revised hierarchical model (RHM), suggest that activation of translation equivalents is a by-product of the type of task. Full article

Process mining has piqued the interest of researchers and technology manufacturers. Process mining aims to extract information from event activities and their interdependencies from events recorded by some enterprise systems. An enterprise system’s transactions are labeled based on their information content, such as an activity that causes the occurrence of another, the timestamp between events, and the resource from which the transaction originated. This paper describes a novel process of parsing event-log activities based on information content (IC). The information content of attributes, especially activity names, which are used to describe the flow processes of enterprise systems, is grouped hierarchically as hypernyms and hyponyms in a subsume tree. The least common subsume (LCS) values of these activity names are calculated, and the corresponding relatedness values between them are obtained. These values are used to create a fuzzy causal matrix (FCM) for parsing the activities, from which a process mining algorithm is designed to mine the structural and semantic relationships among activities using an enhanced gray wolf optimizer and backpropagation algorithm. The proposed approach is resistant to noisy and incomplete event logs and can be used for process mining to reflect the structure and behavior of event logs. Full article

In the present investigation we adopted the Rapid Parallel Visual Presentation Paradigm with the aim of studying the timing of parafoveal semantic processing. The paradigm consisted in the simultaneous presentation of couple of words, one in fovea (W1) and one in parafovea (W2). In three experiments, we manipulated word frequency, semantic relatedness between the two words and the effect of stimulus duration (150, 100, 50 ms). Accuracy on W2 was higher when W1 and W2 were both of high-frequency and when they were semantically related. W1 reading times were faster when both words were highly-frequent but only when the two words were semantically related (150 ms); when W2 was highly frequent and semantically related to the foveal word (100 ms). When the stimuli were presented for 50 ms, the reading times were reduced when W1 was highly frequent and, crucially, in case of a semantic relation between the two words. Our results suggest that it is possible to extract semantic information from the parafovea very fast (within 100 ms) and in parallel to the processing of the foveal word, especially when the cognitive load required for the latter is reduced, as is the case for high-frequency words. We discuss the resulting data in terms of word recognition and eye movements’ models. Full article

The knowledge graph has received a lot of interest in the field of recommender systems as side information because it can address the sparsity and cold start issues associated with collaborative filtering-based recommender systems. However, when incorporating entities from a knowledge graph to represent semantic information, most current KG-based recommendation methods are unaware of the relationships between these users and items. As such, the learned semantic information representation of users and items cannot fully reflect the connectivity between users and items. In this paper, we present the PRKG-DI symmetry model, a Personalized Relationships-based Knowledge Graph for recommender systems with Dual-view Items that explores user-item relatedness by mining associated entities in the KG from user-oriented entity view and item-oriented entity view to augment item semantic information. Specifically, PRKG-DI utilizes a heterogeneous propagation strategy to gather information on higher-order user-item interactions and an attention mechanism to generate the weighted representation of entities. Moreover, PRKG-DI provides a score feature as a filter for individualized relationships to evaluate users’ potential interests. The empirical results demonstrate that our approach significantly outperforms several state-of-the-art baselines by 1.6%, 2.1%, and 0.8% on $AUC$, and 1.8%, 2.3%, and 0.8% on $F1$ when applied to three real-world scenarios for music, movie, and book recommendations, respectively. Full article

Due to the advancement in information technology and the boom of micro-blogging platforms, a growing number of online reviews are posted daily on product distributed platforms in the form of spontaneous and insightful user feedback, and these can be used as a significant data source to understand user experience (UX) and satisfaction. However, despite the vast amount of online reviews, the existing literature focuses on online ratings and ignores the real textual context in reviews. We proposed a three-step UX quantification model from online reviews to understand customer satisfaction using the effect-based Kano model. First, the relevant online reviews are selected using various filter mechanisms. Second, UX dimensions (UXDs) are extracted using a proposed method called UX word embedding Latent Dirichlet allocation (UXWE-LDA) and sentiment orientation using a transformer-based pipeline. Then, the casual relationships are identified for the extracted UXDs. Third, the UXDs are mapped on the customer satisfaction model (effect-based Kano) to understand the user perspective about the system, product, or services. Finally, the different parts of the proposed quantification model are evaluated to examine the performance of this method. We present different results of the proposed method in terms of accuracy, topic coherence (TC), Topic-wise performance, and expert-based evaluation for the proposed framework validation. For review quality filters, we achieved 98.49% accuracy for the spam detection classifier and 95% accuracy for the relatedness detection classifier. The results show that the proposed method for the topic extractor module always gives a higher TC value than other models such as WE-LDA and LDA. Regarding topic-wise performance measures, UXWE-LDA achieves a 3% improvement on average compared to LDA due to the incorporation of semantic domain knowledge. We also compute the Jaccard coefficient similarity between the extracted dimensions using UXWE-LDA and UX experts-based analysis for checking the mutual agreement, which is 0.3, 0.5, and 0.4, respectively. Based on the Kano model, the presented study has potential implications concerning issues and knowing the product’s strengths and weaknesses in product design. Full article

In multilingual textual archives, the availability of textual annotation, that is keywords either manually or automatically associated with texts, is something worth exploiting to improve user experience and successful navigation, search and visualization. It is therefore necessary to study and develop tools for this exploitation. The paper aims to define models and tools for handling textual annotations, in our case keywords of a scientific library. With the background of NLP, machine learning and deep learning approaches are presented. They allow us, in supervised and unsupervised ways, to increase the quality of keywords. The different steps of the pipeline are addressed, and different solutions are analyzed, implemented, evaluated and compared, using statistical methods, machine learning and artificial neural networks as appropriate. If possible, off-the-shelf solutions will also be compared. The models are trained on different datasets already available or created ad hoc with common characteristics with the starting dataset. The results obtained are presented, commented and compared with each other. Full article