Hasil untuk "Computational linguistics. Natural language processing"

Qian Luo

Chunmei Xie, Cong Xie

Abstract To cope with the inventory control problem of an intelligent warehousing system, this paper proposes a neuro-fuzzy dynamic inventory regulation model (NFDIRM), which integrates radial basis function neural network (RBFNN) and fuzzy logic algorithm, aiming to solve the problems of poor prediction accuracy and poor decision flexibility of traditional models. The experiment is based on historical inventory data from a large e-commerce platform, encompassing over 500 commodities across three years. NFDIRM is compared with the economic order quantity (EOQ) model and the ARIMA model, and an ablation analysis is conducted. The results show that the comprehensive average inventory turnover rate of NFDIRM is 22.08, the average out-of-stock rate is 2.77%, and the average inventory cost is 324,600 yuan, which is significantly better than the control model. Ablation analysis reveals that after removing the RBFNN module, the comprehensive average turnover rate decreases to 15.65, while the comprehensive average out-of-stock rate increases to 6.93%. After removing the fuzzy logic decision module, the comprehensive average turnover rate drops to 18.0, and the comprehensive average out-of-stock rate rises to 5.1%. The NFDIRM model proposed in this study enhances the accuracy and efficiency of inventory control, offering a novel solution for intelligent warehouse inventory management. However, the applicability of the model in different industry scenarios still needs to be verified by further research.

Xi Lu

Abstract To solve the low efficiency, high cost, and poor user experience in traditional clothing design and display, this study integrates 3D reconstruction stereo and virtual reality technology to develop an intelligent clothing wearable display system, improving design efficiency and user experience. By introducing Transformer and overlapping attention mechanism, a 3D reconstruction stereo (Transformer with Overlapping Attention for Multi-View Stereo 3D Reconstruction, TransOver-MVS) method is constructed to achieve high-precision 3D reconstruction. Based on this model, a virtual reality intelligent clothing wearing display platform is designed, which includes clothing modeling, human body modeling, and human clothing interaction modules. Intelligent stitching and conflict detection algorithms are used to enhance the authenticity of virtual try on. The results showed that TransOver-MVS had an average deviation of 0.215 mm, 0.186 mm, and 0.329 mm in accuracy, completeness, and integrity on the Danish Technical University Dataset, which was superior to the comparison techniques. The average F-score on the Tanks and Temples dataset was 32.09, which also performed the best. The intelligent clothing display platform showed that the score for role model saving and virtual matching clothing addition and deletion functions reached 9.7 points. The research has verified the superiority of TransOver-MVS in 3D reconstruction, initially achieving digital closed-loop management of clothing and human body, providing innovative solutions for clothing design and display, promoting the digital transformation of the clothing industry, and improving design efficiency and market competitiveness.

Phong Nguyen Duc, Thang Tang Duc, Giap Pham Van et al.

Abstract Salinity intrusion stands out as a severe yet escalating challenge facing the water resource management and agricultural production of the Mekong Delta in Vietnam as a result of climate change and upstream hydrological changes. This paper assesses the efficacy of six different machine learning (ML) and deep learning models (DL) for hourly prediction of salinity in the Mekong Delta at four stations (Cau Quan, Tra Vinh, Ben Trai, and Tran De). The six models are XGB, GB, SVR, LSTM, RNN and ANN. Using hourly hydrological data of 2015–2020 with upstream discharge and tidal water levels as major inputs we designed training and testing of models (training: Jan 2015-mid 2018; testing: mid 2018-Feb 2020). Our results prove that LSTM and XGB models have the best prediction. In particular, they showed good accuracy in predicting upstream salinity (RMSE: 0.25 to 0.30, R2 > 0.97) and downstream salinity (RMSE: 1.5 to 1.6, R2 > 0.88). This success is due to capacity of high temporal resolution as well as spatio-temporal dynamics of salinity variation. The LSTM structure has proven to be effective at capturing long-term temporal dependencies, such as seasonal discharge patterns, while XGB successfully models non-linear interactions between stations with the greatest success, particularly discharge-tidal level interactions. The ML/DL models are capable of successfully forecasting salinity which can open doors to data-driven water management in the Mekong Delta. Future studies should further add hydro-meteorological parameters, other hybrid architectures, and real-time prediction systems, which could be useful operationally and have wider applicability.

Hamed Ouloto YEO

Résumé : À la faveur de la nouvelle génération de dramaturges africains qui s’ouvrent au monde à partir des années 1990, le théâtre africain va prendre une autre tournure avec des auteurs comme José Pliya que nous avons décidé d’étudier à travers le sujet suivant : « Séisme scriptural et créativité scénologique dans le théâtre de José PLIYA. » Dès lors, les concepts d’écriture et de créativité font appelle à celui du chamboulement constaté dans ses œuvres théâtrales où se mêlent et s’entremêlent différents genres littéraires. Pour le dramaturge, l’accent doit être mis sur l’universalité de l’écriture et de la connaissance. Le théâtre de ce Béninois d’origine apparaît, de ce fait, comme un théâtre nouveau, révolutionnaire et intemporel. Ainsi, tourner en dérision la structure dramatique traditionnelle suggère le refus de l’immobilité et de la fixité chez Pliya. Mots-clés : Art, créativité, discours, mutation, Séisme scriptural, hybridité.

Bruno Gonçalves Carneiro, Roselba Gomes de Miranda, Karylleila dos Santos Andrade et al.

This article presents the results of a research on toponyms of cities in the state of Tocantins, Brazil, in Brazilian Sign Language (Libras). Signs were collected from observation and interviews with deaf people. The state of Tocantins consists of 139 municipalities and, so far, we have surveyed 61 toponyms. Signs were categorized into (1) native, (2) initialized and (3) fingerspelled (exclusive categories). Regarding motivation, toponyms categorized as iconic motivation, covering (1) material and (2) cultural and Portuguese language motivation, covering (3) calque and (4) writing. The spelling motivation is the most prevalent in the toponyms of the cities of Tocantins, also combining with other types of motivation, such as (5) material and spelling, (6) cultural and writing, and (7) calque and spelling. We bring further considerations on these categories and suggest forearm as a productive point of articulation for toponymic signs in the context of Tocantins.

BADO GNOUMOU Boromoussan

Résumé : Le corps féminin dans L’homme à la bagnole rouge, sous une lecture sémiotique est un corps violé, violenté, transgressé aussi bien dans son enveloppe corporelle que dans sa chair mouvante. C’est un corps hors d’usage, meurtrie et meurtrier. En effet Pamela est une jeune fille qui a été rejetée par son géniteur Alphonse Sawadogo depuis le ventre de sa mère. Profondément marquée et sur son enveloppe corporelle et dans sa chair mouvante, sa mère Salimata va se suicider. Remise à ses grands-parents qui prennent le relai pour son éducation, Pamela devient la bonne à tout faire et un moyen de transformation de leur statut social qui est de passer de l’ extrême pauvreté à l’ opulence. Le corps féminin se constitue donc en un réseau de marquages successifs. Ce qui fait de ce corps, un corps féminin signifiant et donc porteur de la sémiotisation. L’œuvre L’homme à la bagnole rouge se situe du point de vue de la victime et elle met en fiction l’exhumation de la mémoire de ces tortures souvent tues, voire niées. La problématique des enfants rejetés, abandonnés porterait –elle en elle-même le germe d’une société en crise ? Notre objectif dans cet article, est de partir de la théorie de la sémiotique du corps de Jacques Fontanille pour montrer comment le corps féminin peut être perçu comme n’importe quel objet faisant parti de ce qu’il est convenu d’appeler des « sémiotiques-objets » c’est-à dire des ensembles observables supposés exprimer des contenus signifiants. Il s’agit d’identifier « la manière dont les figures de l’expression prennent forme à partir du substrat matériel des inscriptions, et du geste qui les y a inscrites » (Fontanille, 2004 : 264), afin d’accéder au plan du contenu. Mots-clés : corps féminin, empreintes, rejet, meurtrier, société.

Pamela Parenti

Ormai in tarda età, sembra che Verga abbia confidato la sua insofferenza rispetto al mancato riconoscimento della sua veste autoriale: «Per chi dovrei scrivere?» avrebbe detto «Di ciò che ho scritto sopravvive soltanto Cavalleria rusticana e non per virtù mia, ma di Pietro Mascagni. Le porto, quelle paginette, come un cappio al collo!». È abbastanza noto che lo scrittore siciliano uscì provato dalle disavventure giudiziarie che lo avevano visto protagonista, insieme a Mascagni, di una lunga disputa, fatta di numerose querele, denunce e processi sui ricavi conseguiti ai successi di Cavalleria e sulle proprie competenze autoriali. Si tratta di vicende che potrebbero tranquillamente rappresentare il soggetto di un film giudiziario dai toni talvolta drammatici e talvolta grotteschi. Nel corso di questo intervento, saranno sinteticamente ricostruiti i vari passaggi della lunga storia, che parte dalla scrittura dei testi (la novella, il dramma, il libretto per il melodramma) per giungere fino alle numerose riprese cinematografiche, con alcuni cenni su altri melodrammi derivati dalla novella in oggetto, ma meno conosciuti: Malapasqua di Stanislao Gastaldon e Cavalleria rusticana di Giovanni e Domenico Monleone.

Jiaxu Zhang, Gaoxiang Ye, Zhigang Tu et al.

Abstract Current studies have shown that the spatial‐temporal graph convolutional network (ST‐GCN) is effective for skeleton‐based action recognition. However, for the existing ST‐GCN‐based methods, their temporal kernel size is usually fixed over all layers, which makes them cannot fully exploit the temporal dependency between discontinuous frames and different sequence lengths. Besides, most of these methods use average pooling to obtain global graph feature from vertex features, resulting in losing much fine‐grained information for action classification. To address these issues, in this work, the authors propose a novel spatial attentive and temporal dilated graph convolutional network (SATD‐GCN). It contains two important components, that is, a spatial attention pooling module (SAP) and a temporal dilated graph convolution module (TDGC). Specifically, the SAP module can select the human body joints which are beneficial for action recognition by a self‐attention mechanism and alleviates the influence of data redundancy and noise. The TDGC module can effectively extract the temporal features at different time scales, which is useful to improve the temporal perception field and enhance the robustness of the model to different motion speed and sequence length. Importantly, both the SAP module and the TDGC module can be easily integrated into the ST‐GCN‐based models, and significantly improve their performance. Extensive experiments on two large‐scale benchmark datasets, that is, NTU‐RGB + D and Kinetics‐Skeleton, demonstrate that the authors’ method achieves the state‐of‐the‐art performance for skeleton‐based action recognition.

Anthony Sicilia, Tristan Maidment, Pat Healy et al.

AbstractInvestigating cooperativity of interlocutors is central in studying pragmatics of dialogue. Models of conversation that only assume cooperative agents fail to explain the dynamics of strategic conversations. Thus, we investigate the ability of agents to identify non-cooperative interlocutors while completing a concurrent visual-dialogue task. Within this novel setting, we study the optimality of communication strategies for achieving this multi-task objective. We use the tools of learning theory to develop a theoretical model for identifying non-cooperative interlocutors and apply this theory to analyze different communication strategies. We also introduce a corpus of non-cooperative conversations about images in the GuessWhat?! dataset proposed by De Vries et al. (2017). We use reinforcement learning to implement multiple communication strategies in this context and find that empirical results validate our theory.

Stuart C. Shapiro

Yuval Marom, Ingrid Zukerman

Hinrich Schütze, Michael Walsh

Retno Yuliati

This study offers an in-depth reading of Jasmine Warga’s Other Word for Home. Deploying descriptive analytical method and P.H. Collins’ concepts of Matrix of Domination and Self-Definition, this study aims to analyze the themes of women’s oppression and women’s empowerment in this book. Findings suggest that Jude and her mother, Syrian refugee women, are oppressed both in Syria and in the US due to their overlapping identity markers (gender, class, ethnicity, religion) in four domains of power (structural, disciplinary, hegemonic, interpersonal). In Syria, they are oppressed in structural domain by Syrian government. Meanwhile in the US, they are oppressed in disciplinary domain to preserve her subordination. Both Jude and Mama are also oppressed by the negative hegemonic ideologies that evolving around Syrian refugees, Muslim, and Arab women. At last, they are oppressed in interpersonal level in everyday life. However, by nurturing certain behaviors in Self-Definition (self-valuation, self-reliance, personal empowerment), they help fostering their empowerment to resist the intersecting oppressions.   Penelitian ini menyajikan pembacaan komprehensif terhadap buku Other Words for Home karya Jasmine Warga. Menggunakan metode deskriptif-analitis dan konsep Matriks Dominasi dan Definisi-Diri P.H. Collins, penelitian ini bermaksud untuk menganalisis tema penindasan terhadap perempuan dan pemberdayaan perempuan dalam buku ini. Ditemukan bahwa Jude dan ibunya tertindas baik di Suriah dan Amerika Serikat karna beragam penanda identitas menyilang mereka (jenis kelamin, kelas, etnisitas, agama) dalam empat wilayah kekuasaan (struktural, disipliner, hegemonik, antar individu). Di Suriah, Jude dan ibunya tertindas dalam wilayah struktural oleh pemerintah Suriah. Sementara di Amerika, mereka tertindas dalam wilayah disipliner agar mereka tetap menjadi seorang bawahan. Jude dan ibunya juga tertindas oleh ideologi hegemonik negatif mengenai pengungsi Suriah, Muslim, dan perempuan Arab. Terakhir, mereka juga tertindas dalam wilayah antar individu dalam kehidupan sehari-hari. Meski begitu, sikap-sikap dalam Definisi-Diri (penghargaan diri, kemandirian, pemberdayaan pribadi) mendukung pemberdayaan mereka dalam melawan penindasan menyilang.

Sana Salman, Deborah Richards, Patrina Caldwell

Embodied Conversational Agents (ECAs) are virtual agents that exhibit humanlike verbal and non-verbal behaviours. When it comes to eHealth, ECAs can provide vital support to patients by being more reachable. In order to make ECAs more effective, humanlike empathy expressed during conversation through relational cues is essential. Empathy revolves around a wide range of verbal and non-verbal behaviours that include, for example, the choice of words in social dialogues. Owing to the COVID-19 situation, there was an opportunity to record online consultations in the Incontinence Clinic and Sleep Clinic at the Children’s Hospital at Westmead in Sydney, Australia. The present study analysed these human dialogues using qualitative research methods to understand the role of empathic dialogue used by the medical team. The qualitative analysis of the live calls used psychology-based relational cues derived from conversational characteristics of humans to build a coding framework around the most relevant themes. Statistical analysis was used to compare relational cue usage between healthcare roles. Finally, using the framework dialogues of the medical team and two ECAs known as Dr Evie (eVirtual Agent for Incontinence and Enuresis) and SAM (Sleep Adherence Mentor) were compared to provide recommendations for health practitioners and future ECA dialogue development.

Shuly Wintner

Mohammed Modhaffer, Sivaramakrishna Challavenkata

In this paper, we investigate the form, salient patterns and core functions of word-level total reduplication in Classical Arabic (CA) and Modern Standard Arabic (MSA). Using a multi-genre corpus of 469 million words, we extract total reduplication (TR) candidates into an SQL database, manually filter them, and perform concordance search to identify the patterns and functions. Data analysis reveals nine patterns and eleven functions of TR and compares their relative frequency in each variety. The functions of TR are mapped into two broad categories: morphological and semantic/pragmatic. Results show an interesting variation in terms of top functions being favored by the two varieties. While TR is favored by CA to express serial ordering, MSA is noticed to favor it to express intensification. The empirical findings of this study provide a reliable quantification of the status of TR in CA and MSA which is rather difficult to obtain by theoretical means: on the one hand, TR in Arabic is not as productive as in other languages such as Indonesian. On the other hand, it is more common in CA than in MSA because the latter usually resorts to using loose phrases to express the same concepts expressed by TR in CA.

Rita KHATER

تروم هذه الدراسة رصد تاريخ "صندوق العجب" وما يزخر به من حكايات شّعبيّة تاريخيّة وخرافيّة وما عرفه من انتشار عربي في نهاية القرن السابع عشر. فيَطفق الحكواتيّ يسرد حكاياته الشعبيّة المبنيّة على الخيال تارةً والمستمدّة من الواقع والمعتقدات والثقافة طورًا.  فهل تقتصر مهنة الحكواتيّ، المتوارثة عادةً، على مجرّد الترفيه والتسلية؟ وإلى أيّ مدى يمكن النّظر إلى حكايات صندوق الفُرجة بوصفها قصصًا تعليميّة قادرة على تحريك خيال الأطفال وترسيخ القِيَم الأخلاقيّة والمفاهيم الثقافيّة في أذهانهم؟ وبتعبيرٍ آخر، إلى أيّ مدى كان صندوق الفُرجَة يُلبّي حاجات عصره الترفيهيّة والمشهديّة والتثقيفيّة؟  وترسيخًا لتراثنا العربيّ وأصالته وإبداعه، سننفض الغبار عن صندوق الفُرجة، وسننظر من طاقاته السحرية لنستكشف الاستراتيجيّات الأسلوبيّة والسيميائيّة والأدائيّة الصوتيّة والتمثيليّة التي كان يتوسَّلها الحكواتيّ. هذا الراوي المتمكِّن من فنون الفُرجة الشّعبيّة لسرد حكاياته وإيصال رسائله التربويّة بأسلوب يزخر بالحيويّة والتشويق.

Suhail Mumtaz Khan

Hanna Prihodko

The paper targets at lingual means of actualizing frame-scripts’ evaluative potential. The article pursues a number of objectives: the study of realization of the structure of evaluative utterances in the frame; identification of functional-semantic features of this realization. The term “frame” is used primarily for the characterization of such structures of consciousness, which are formed for displaying situations in object-human cognitive activity. Frame is a declarative way of knowledge representation, which is formulated in terms of descriptions and is a bundle of knowledge about a particular area of human activity, on the ontology of the world, the human representation of the system, thematically related to one picture, one act and one script. Such understanding of the frame gives grounds to speak of it as a definitely organized system (set) of propositions which schematize corresponding denotative situations, that is, as a minimum informative block. Therefore, the frame can be called the cognitive-communicative field, or communicative frame. Formation of the frame as the specific structure of consciousness, corresponding to the representation of the event, has an ontological basis. The logical analysis of concepts, which provides for the establishment of the laws of its internal organization in order to identify its components and modeling their interactions, confirms the notion of the frame as a stereotypical situation. As far as the person’s life-world is made up of many situations, their language and speech fixation needs the combination of situations into the utterances. Thus, the evaluative utterance is the product of a certain reflection pattern, scene, or a script in the communicative act. In accordance with socio-role status relationships between participants of communication are spread mainly in the social sphere of communication, where social role reflects interactional conditions between the subject of communicative action and its object. According to some linguists, conditions of social interaction between communicants are based on three types of relations — equality, subordination and dominance, which are implemented in the familiar, unconstrained, neutral and elevated communication (speech) registers. Communicative role as a kind of invariant unit of behavior, is a part of the general scheme of activities and is related to relevant normative expectations, which may be shown by the communicants in a given communicative and particular social situations. Every act of communication is characterized by the definite form of interaction, which is based on its correlation with the situation-type, which is the frame with the features and functional conditions inherent to it. Frame structure can be regarded as an independent configuration consisting of a core, a set of standard forms of speech acts, participants of speech event. In addition to these components, an important role belongs to the objective, plan and consequent. Thus, the evaluative situation can be attributed to the frame, as it includes evaluation of the phenomena of the outer world and illustrates the continuity of images of the object and the subject, objectified in the system parts of speech, as well as all the constituents of utterances that make up the situation. Considering the above stated, the cognitive-communicative field of evaluative situation can be represented as hyperframe of verbal interaction, which displays all components of the frame in their interconnection and interdependence, which allows to determine the sequence of the constituents of data in the process of updating and predetermine the appearance of certain actions that characterize the core of the frame structure in functional and semantic aspects. An important component of the mechanism of formation of the evaluative utterance is a cognitive factor. On the one hand, it helps to conceptualize the relations between situations of extralinguistic reality (sender’s perspective), and on the other hand, it serves as a specific signal for the process of the mental perception of text (recipient’s perspective). Thus, participants of interaction use are general schemes of encoding and decoding of information that is contained in the utterance. Consequently, the actualized structure of the evaluative utterance is connected with the realization of the frame structure of a typical act of communication in the form of a holistic multi-level formation — functional-semantic representation. Actualization of a frame structure takes place on the background of social interaction, where the functional-semantic representation of the evaluative frame is promoted by implementing in its structure such items as the illocutionary act-event, topical proposition and typical grammatical construction. As a result, there is a picture of interconnected and interdependent in their development concepts — from parts of the utterance — to all evaluative utterance, which is the core of the cognitive-communicative field of interaction.