Search results for: impersonal sentence model

Authors: Shrabani Ghosh

Abstract:

Automatic offensive language detection in social media has become a stirring task in today's NLP. Manual Offensive language detection is tedious and laborious work where automatic methods based on machine learning are only alternatives. Previous works have done sentiment analysis over social media in different ways such as supervised, semi-supervised, and unsupervised manner. Domain adaptation in a semi-supervised way has also been explored in NLP, where the source domain and the target domain are different. In domain adaptation, the source domain usually has a large amount of labeled data, while only a limited amount of labeled data is available in the target domain. Pretrained transformers like BERT, RoBERTa models are fine-tuned to perform text classification in an unsupervised manner to perform further pre-train masked language modeling (MLM) tasks. In previous work, hate speech detection has been explored in Gab.ai, which is a free speech platform described as a platform of extremist in varying degrees in online social media. In domain adaptation process, Twitter data is used as the source domain, and Gab data is used as the target domain. The performance of domain adaptation also depends on the cross-domain similarity. Different distance measure methods such as L2 distance, cosine distance, Maximum Mean Discrepancy (MMD), Fisher Linear Discriminant (FLD), and CORAL have been used to estimate domain similarity. Certainly, in-domain distances are small, and between-domain distances are expected to be large. The previous work finding shows that pretrain masked language model (MLM) fine-tuned with a mixture of posts of source and target domain gives higher accuracy. However, in-domain performance of the hate classifier on Twitter data accuracy is 71.78%, and out-of-domain performance of the hate classifier on Gab data goes down to 56.53%. Recently self-supervised learning got a lot of attention as it is more applicable when labeled data are scarce. Few works have already been explored to apply self-supervised learning on NLP tasks such as sentiment classification. Self-supervised language representation model ALBERTA focuses on modeling inter-sentence coherence and helps downstream tasks with multi-sentence inputs. Self-supervised attention learning approach shows better performance as it exploits extracted context word in the training process. In this work, a self-supervised attention mechanism has been proposed to detect hate speech on Gab.ai. This framework initially classifies the Gab dataset in an attention-based self-supervised manner. On the next step, a semi-supervised classifier trained on the combination of labeled data from the first step and unlabeled data. The performance of the proposed framework will be compared with the results described earlier and also with optimized outcomes obtained from different optimization techniques.

Keywords: attention learning, language model, offensive language detection, self-supervised learning

Procedia PDF Downloads 82

Authors: Tiago do Carmo Nogueira, Cássio Dener Noronha Vinhal, Gélson da Cruz Júnior, Matheus Rudolfo Diedrich Ullmann

Abstract:

Generating automatic image descriptions through natural language is a challenging task. Image captioning is a task that consistently describes an image by combining computer vision and natural language processing techniques. To accomplish this task, cutting-edge models use encoder-decoder structures. Thus, Convolutional Neural Networks (CNN) are used to extract the characteristics of the images, and Recurrent Neural Networks (RNN) generate the descriptive sentences of the images. However, cutting-edge approaches still suffer from problems of generating incorrect captions and accumulating errors in the decoders. To solve this problem, we propose a model based on the encoder-decoder structure, introducing a module that generates the weights according to the importance of the word to form the sentence, using the part-of-speech (PoS). Thus, the results demonstrate that our model surpasses state-of-the-art models.

Keywords: gated recurrent units, caption generation, convolutional neural network, part-of-speech

Procedia PDF Downloads 70

Authors: Yao-Ying Lai, Maria Pinango, Ashwini Deo

Abstract:

While grammatical devices are essential to language processing, how comprehension utilizes cognitive mechanisms is less emphasized. This study addresses this issue by probing the complement coercion phenomenon: an entity-denoting complement following verbs like begin and finish receives an eventive interpretation. For example, (1) “The queen began the book” receives an agentive reading like (2) “The queen began [reading/writing/etc.…] the book.” Such sentences engender additional processing cost in real-time comprehension. The traditional account attributes this cost to an operation that coerces the entity-denoting complement to an event, assuming that these verbs require eventive complements. However, in closer examination, examples like “Chapter 1 began the book” undermine this assumption. An alternative, Structured Individual (SI) hypothesis, proposes that the complement following aspectual verbs (AspV; e.g. begin, finish) is conceptualized as a structured individual, construed as an axis along various dimensions (e.g. spatial, eventive, temporal, informational). The composition of an animate subject and an AspV such as (1) engenders an ambiguity between an agentive reading along the eventive dimension like (2), and a constitutive reading along the informational/spatial dimension like (3) “[The story of the queen] began the book,” in which the subject is interpreted as a subpart of the complement denotation. Comprehenders need to resolve the ambiguity by searching contextual information, resulting in additional cost. To evaluate the SI hypothesis, a questionnaire was employed. Method: Target AspV sentences such as “Shakespeare began the volume.” were preceded by one of the following types of context sentence: (A) Agentive-biasing, in which an event was mentioned (…writers often read…), (C) Constitutive-biasing, in which a constitutive meaning was hinted (Larry owns collections of Renaissance literature.), (N) Neutral context, which allowed both interpretations. Thirty-nine native speakers of English were asked to (i) rate each context-target sentence pair from a 1~5 scale (5=fully understandable), and (ii) choose possible interpretations for the target sentence given the context. The SI hypothesis predicts that comprehension is harder for the Neutral condition, as compared to the biasing conditions because no contextual information is provided to resolve an ambiguity. Also, comprehenders should obtain the specific interpretation corresponding to the context type. Results: (A) Agentive-biasing and (C) Constitutive-biasing were rated higher than (N) Neutral conditions (p< .001), while all conditions were within the acceptable range (> 3.5 on the 1~5 scale). This suggests that when lacking relevant contextual information, semantic ambiguity decreases comprehensibility. The interpretation task shows that the participants selected the biased agentive/constitutive reading for condition (A) and (C) respectively. For the Neutral condition, the agentive and constitutive readings were chosen equally often. Conclusion: These findings support the SI hypothesis: the meaning of AspV sentences is conceptualized as a parthood relation involving structured individuals. We argue that semantic representation makes reference to spatial structured-ness (abstracted axis). To obtain an appropriate interpretation, comprehenders utilize contextual information to enrich the conceptual representation of the sentence in question. This study connects semantic structure to human’s conceptual structure, and provides a processing model that incorporates contextual retrieval.

Keywords: ambiguity resolution, contextual retrieval, spatial structured-ness, structured individual

Procedia PDF Downloads 301

Authors: Riona Dighe

Abstract:

This research attempts to identify an alternative approach to the measurement of Gross National Happiness (GNH). It uses artificial intelligence (AI), incorporating natural language processing (NLP) and sentiment analysis to measure GNH. We use ‘off the shelf’ NLP models responsible for the sentiment analysis of a sentence as a building block for this research. We constructed an algorithm using NLP models to derive a sentiment analysis score against sentences. This was then tested against a sample of 20 respondents to derive a sentiment analysis score. The scores generated resembled human responses. By utilising the MLP classifier, decision tree, linear model, and K-nearest neighbors, we were able to obtain a test accuracy of 89.97%, 54.63%, 52.13%, and 47.9%, respectively. This gave us the confidence to use the NLP models against sentences in websites to measure the GNH of a country.

Keywords: artificial intelligence, NLP, sentiment analysis, gross national happiness

Procedia PDF Downloads 74

Authors: Abdullah Eqal Al Mazrooei

Abstract:

In this work, a new nonlinear model will be introduced. The model is in the state-space form. The nonlinearity of this model is in the state equation where the state vector is multiplied by its self. This technique makes our model generalizes many famous models as Lotka-Volterra model and Lorenz model which have many applications in the real life. We will apply our new model to estimate the wind speed by using a new nonlinear estimator which suitable to work with our model.

Keywords: nonlinear systems, state-space model, Kronecker product, nonlinear estimator

Procedia PDF Downloads 661

Authors: Tsvetomira Braynova, Miglena Simonska

Abstract:

The article presents the results of research on phonological processes in three and four-year-old children. For the purpose of the study, an author's test was developed and conducted among 120 children. The study included three areas of research - at the level of words (96 words), at the level of sentence repetition (10 sentences) and at the level of generating own speech from a picture (15 pictures). The test also gives us additional information about the articulation errors of the assessed children. The main purpose of the icing is to analyze all phonological processes that occur at this age in Bulgarian children and to identify which are typical and atypical for this age. The results show that the most common phonology errors that children make are: sound substitution, an elision of sound, metathesis of sound, elision of a syllable, and elision of consonants clustered in a syllable. All examined children were identified with the articulatory disorder from type bilabial lambdacism. Measuring the correlation between the average length of repeated speech and the average length of generated speech, the analysis proves that the more words a child can repeat in part “repeated speech,” the more words they can be expected to generate in part “generating sentence.” The results of this study show that the task of naming a word provides sufficient and representative information to assess the child's phonology.

Keywords: assessment, phonology, articulation, speech-language development

Procedia PDF Downloads 150

Authors: Chaiwat Tantarangsee

Abstract:

The purposes of this study are: 1) to study the frequent English writing errors of students registering the course: Reading and Writing English for Academic Purposes II, and 2) to find out the results of writing error correction by using coded indirect corrective feedback and writing error treatments. Samples include 28 2nd year English Major students, Faculty of Education, Suan Sunandha Rajabhat University. Tool for experimental study includes the lesson plan of the course; Reading and Writing English for Academic Purposes II, and tool for data collection includes 4 writing tests of short texts. The research findings disclose that frequent English writing errors found in this course comprise 7 types of grammatical errors, namely Fragment sentence, Subject-verb agreement, Wrong form of verb tense, Singular or plural noun endings, Run-ons sentence, Wrong form of verb pattern and Lack of parallel structure. Moreover, it is found that the results of writing error correction by using coded indirect corrective feedback and error treatment reveal the overall reduction of the frequent English writing errors and the increase of students’ achievement in the writing of short texts with the significance at .05.

Keywords: coded indirect corrective feedback, error correction, error treatment, frequent English writing errors

Procedia PDF Downloads 210

Authors: Haiyan Wu, Ying Liu, Shaoyun Shi

Abstract:

Authorship attribution is to extract features to identify authors of anonymous documents. Many previous works on authorship attribution focus on statistical style features (e.g., sentence/word length), content features (e.g., frequent words, n-grams). Modeling these features by regression or some transparent machine learning methods gives a portrait of the authors' writing style. But these methods do not capture the syntactic (e.g., dependency relationship) or semantic (e.g., topics) information. In recent years, some researchers model syntactic trees or latent semantic information by neural networks. However, few works take them together. Besides, predictions by neural networks are difficult to explain, which is vital in authorship attribution tasks. In this paper, we not only utilize the statistical style and content features but also take advantage of both syntactic and semantic features. Different from an end-to-end neural model, feature selection and prediction are two steps in our method. An attentive n-gram network is utilized to select useful features, and logistic regression is applied to give prediction and understandable representation of writing style. Experiments show that our extracted features can improve the state-of-the-art methods on three benchmark datasets.

Keywords: authorship attribution, attention mechanism, syntactic feature, feature extraction

Procedia PDF Downloads 105

Authors: H. Al-Awaisi, M. H. Al-Azri, S. Al-Rasbi, M. Al-Moundhri

Abstract:

Breast cancer is the most common cancer among females worldwide. It is also the most common cancer among females in Oman with 100 new breast cancer cases diagnosed every year. It has been found that breast cancer have a devastating effect on women’s life. Women diagnosed with breast cancer might develop negative attitudes towards the illness and their bodies. They might also suffer from psychological ailments such as depression. Despite the evidence on the impact of breast cancer diagnosis on women, there was no study found to explore the impact of breast cancer diagnosis among women in Oman. A phenomenological qualitative study was conducted to explore the impact of breast cancer diagnosis on Omani women. Data was collected through semi-structured individual interviews with 11 Omani women diagnosed with breast cancer. Interviews were transcribed verbatim and data were analyzed thematically. From the data, there are four main themes identified in relation to the impact of cancer diagnosis on Omani women. These are 'shock and disbelieve', 'a death sentence', “uncertain future” and “social stigma”. At the time of interviews, all participants had advanced breast cancer with some participants having metastatic disease. The impact of the word “cancer” had a profound and catastrophic effect on the women and their close relatives. In conclusion, breast cancer diagnosis was shocking and mainly perceived as a death sentence by Omani women with uncertain future and social stigma. Regardless of age, maternal status and education level, it is evident that Omani women participated in this study lacked awareness about breast cancer diagnosis, treatment and prognosis.

Keywords: breast cancer, coping, diagnosis, Oman, women

Procedia PDF Downloads 463

Authors: Qiong Li

Abstract:

This study investigated in which condition computer-mediated communication (CMC) could promote pragmatic development. The focal feature included four Chinese sentence final particles (SFPs), a, ya, ba, and ne. They occur frequently in Chinese, and function as mitigators to soften the tone of speech. However, L2 acquisition of SFPs is difficult, suggesting the necessity of additional exposure to or explicit instruction on Chinese SFPs. This study follows this line and aims to explore two research questions: (1) Is CMC combined with data-driven instruction more effective than CMC alone in promoting L2 Chinese learners’ SFP use? (2) How does L2 Chinese learners’ SFP use change over time, as compared to the production of native Chinese speakers? The study involved 19 intermediate-level learners of Chinese enrolled at a private American university. They were randomly assigned to two groups: (1) the control group (N = 10), which was exposed to SFPs through CMC alone, (2) the treatment group (N = 9), which was exposed to SFPs via CMC and data-driven instruction. Learners interacted with native speakers on given topics through text-based CMC over Skype. Both groups went through six 30-minute CMC sessions on a weekly basis, with a one-week interval after the first two CMC sessions and a two-week interval after the second two CMC sessions (nine weeks in total). The treatment group additionally received a data-driven instruction after the first two sessions. Data analysis focused on three indices: token frequency, type frequency, and acceptability of SFP use. Token frequency was operationalized as the raw occurrence of SFPs per clause. Type frequency was the range of SFPs. Acceptability was rated by two native speakers using a rating rubric. The results showed that the treatment group made noticeable progress over time on the three indices. The production of SFPs approximated the native-like level. In contrast, the control group only slightly improved on token frequency. Only certain SFPs (a and ya) reached the native-like use. Potential explanations for the group differences were discussed in two aspects: the property of Chinese SFPs and the role of CMC and data-driven instruction. Though CMC provided the learners with opportunities to notice and observe SFP use, as a feature with low saliency, SFPs were not easily noticed in input. Data-driven instruction in the treatment group directed the learners’ attention to these particles, which facilitated the development.

Keywords: computer-mediated communication, data-driven instruction, pragmatic development, second language Chinese, sentence final particles

Procedia PDF Downloads 391

Authors: Hsin-Yang Wang, K. M. A. Salam, Ying-Jia Lin, Daniel Tan, Tzu-Hsuan Chou, Hung-Yu Kao

Abstract:

Detecting user intent from natural language expression has a wide variety of use cases in different natural language processing applications. Recently few-shot training has a spike of usage on commercial domains. Due to the lack of significant sample features, the downstream task performance has been limited or leads to an unstable result across different domains. As a state-of-the-art method, the pre-trained BERT model gathering the sentence-level information from a large text corpus shows improvement on several NLP benchmarks. In this research, we are proposing a method to change multi-class classification tasks into binary classification tasks, then use the confidence score to rank the results. As a language model, BERT performs well on sequence data. In our experiment, we change the objective from predicting labels into finding the relations between words in sequence data. Our proposed method achieved 71.0% accuracy in the internal intent detection dataset and 63.9% accuracy in the HuffPost dataset. Acknowledgment: This work was supported by NCKU-B109-K003, which is the collaboration between National Cheng Kung University, Taiwan, and SoftBank Corp., Tokyo.

Keywords: OSML, BERT, text classification, one shot

Procedia PDF Downloads 80

Authors: Chaiwat Tantarangsee

Abstract:

The purposes of this study are 1) to study the frequent English writing errors of students registering the course: Reading and Writing English for Academic Purposes II, and 2) to find out the results of writing error correction by using coded indirect corrective feedback and writing error treatments. Samples include 28 2nd year English Major students, Faculty of Education, Suan Sunandha Rajabhat University. Tool for experimental study includes the lesson plan of the course; Reading and Writing English for Academic Purposes II, and tool for data collection includes 4 writing tests of short texts. The research findings disclose that frequent English writing errors found in this course comprise 7 types of grammatical errors, namely Fragment sentence, Subject-verb agreement, Wrong form of verb tense, Singular or plural noun endings, Run-ons sentence, Wrong form of verb pattern and Lack of parallel structure. Moreover, it is found that the results of writing error correction by using coded indirect corrective feedback and error treatment reveal the overall reduction of the frequent English writing errors and the increase of students’ achievement in the writing of short texts with the significance at .05.

Keywords: coded indirect corrective feedback, error correction, error treatment, English writing

Procedia PDF Downloads 277

Authors: Yufen Qin

Abstract:

Text sentiment transfer is an important branch of text style transfer. The goal is to generate text with another sentiment attribute based on a text with a specific sentiment attribute while maintaining the content and semantic information unrelated to sentiment unchanged in the process. There are currently two main challenges in this field: no parallel corpus and text attribute entanglement. In response to the above problems, this paper proposed a novel solution: Mask-Prompt-Rerank. Use the method of masking the sentiment words and then using prompt regeneration to transfer the sentence sentiment. Experiments on two sentiment benchmark datasets and one formality transfer benchmark dataset show that this approach makes the performance of small pre-trained language models comparable to that of the most advanced large models, while consuming two orders of magnitude less computing and memory.

Keywords: language model, natural language processing, prompt, text sentiment transfer

Procedia PDF Downloads 48

Authors: Naouel Zoghlami

Abstract:

Spoken-word recognition involves the simultaneous activation of potential word candidates which compete with each other for final correct recognition. In continuous speech, the activation-competition process gets more complicated due to speech reductions existing at word boundaries. Lexical processing is more difficult in L2 than in L1 because L2 listeners often lack phonetic, lexico-semantic, syntactic, and prosodic knowledge in the target language. In this study, we investigate the on-line lexical segmentation hypotheses that French listeners of L2 English form and then revise as subsequent perceptual evidence is revealed. Our purpose is to shed further light on the processes of L2 spoken-word recognition in context and better understand L2 listening difficulties through a comparison of skilled and unskilled reactions at the point where their working hypothesis is rejected. We use a variant of the gating experiment in which subjects transcribe an English sentence presented in increments of progressively greater duration. The spoken sentence was “And this amazing athlete has just broken another world record”, chosen mainly because it included common reductions and phonetic features in English, such as elision and assimilation. Our preliminary results show that there is an important difference in the manner in which proficient and less-proficient L2 listeners handle connected speech. Less-proficient listeners delay recognition of words as they wait for lexical and syntactic evidence to appear in the gates. Further statistical results are currently being undertaken.

Keywords: gating paradigm, spoken word recognition, online lexical segmentation, L2 listening

Procedia PDF Downloads 442

Authors: Tayebeh Norouzi

Abstract:

This study aims to investigate the acquisition process of intonation. It examines the intonation structure of Tokyo Japanese and its realization by Iranian learners of Japanese. Seven Iranian learners of Japanese, differing in fluency, and two Japanese speakers participated in the experiment. Two sentences were used to test the phonological and phonetic characteristics of lexical pitch-accent as well as the intonation patterns produced by the speakers. Both sentences consisted of similar words with the same number of syllables and lexical pitch-accents but different syntactic structure. Speakers were asked to read each sentence three times at normal speed, and the data were analyzed by Praat. The results show that lexical pitch-accent, Accentual Phrase (AP) and AP boundary tone realization vary depending on sentence type. For sentences of type XdeYwo, the lexical pitch-accent is realized properly. However, there is a rise in AP boundary tone regardless of speakers’ level of fluency. In contrast, in sentences of type XnoYwo, the lexical pitch-accent and AP boundary tone vary depending on the speakers’ fluency level. Advanced speakers are better at grouping words into phrases and produce more native-like intonation patterns, though they are not able to realize downstep properly. The non-native speakers tried to realize proper intonation patterns by making changes in lexical accent and boundary tone.

Keywords: intonation, Iranian learners, Japanese prosody, lexical accent, second language acquisition.

Procedia PDF Downloads 130

Authors: Randy S. Magdaluyo, Lea M. Cabar, Jefferson Q. Correa

Abstract:

One recurring issue in ESL writing is the confusing differences in the writing conventions of the first language and the target language. Culture may play an intriguing role in specifying writing features and structures that ESL writers have to follow. Although writing is typically organized in a three-part structure with introduction, body, and conclusion, it is important to analyze the complex nature of ESL writing. This study investigated the organizational features and structures of argumentative essays written in English by thirty college ESL students from three linguistic backgrounds (Cebuano, Chavacao, and Tausug) in a Philippine university. The nature of word order and sentence construction in the students’ essays and the specific components of the introduction, body, and conclusion were quantitatively and qualitatively analyzed based on ESL writing models. Focus group discussions were also conducted to help clarify the possible influence of students’ first language on the ways their essays were conceptualized and organized. Results indicate that while there was no significant difference in the overall introduction, body, and conclusion in all essays, the sentence length was interestingly different for each linguistic group of ESL students, and the word order was notably inconsistent with the S-V-O pattern of the target language. The first language was also revealed to have a facilitative role in the cognitive translation process of these ESL students. As such, implications for a multicultural writing pedagogy was discussed and recommended considering both the students’ native resources in their first language and the ESL writing models in their target language.

Keywords: community funds of knowledge, contrastive rhetoric, ESL writing, multicultural writing pedagogy

Procedia PDF Downloads 107

Authors: Nancy Retzlaff, Sarah J. Berkemer, Trudie Strauss

Abstract:

In the last couple of decades, the advent of digitalization of any kind of data was probably one of the major advances in all fields of study. This paves the way for also analysing these data even though they might come from disciplines where there was no initial computational necessity to do so. Especially in linguistics, one can find a rather manual tradition. Still when considering studies that involve the history of language families it is hard to overlook the striking similarities to bioinformatics (phylogenetic) approaches. Alignments of words are such a fairly well studied example of an application of bioinformatics methods to historical linguistics. In this paper we will not only consider alignments of strings, i.e., words in this case, but also alignments of syntax trees of selected Indo-European languages. Based on initial, crude alignments, a sophisticated scoring model is trained on both letters and syntactic features. The aim is to gain a better understanding on which features in two languages are related, i.e., most likely to have the same root. Initially, all words in two languages are pre-aligned with a basic scoring model that primarily selects consonants and adjusts them before fitting in the vowels. Mixture models are subsequently used to filter ‘good’ alignments depending on the alignment length and the number of inserted gaps. Using these selected word alignments it is possible to perform tree alignments of the given syntax trees and consequently find sentences that correspond rather well to each other across languages. The syntax alignments are then filtered for meaningful scores—’good’ scores contain evolutionary information and are therefore used to train the sophisticated scoring model. Further iterations of alignments and training steps are performed until the scoring model saturates, i.e., barely changes anymore. A better evaluation of the trained scoring model and its function in containing evolutionary meaningful information will be given. An assessment of sentence alignment compared to possible phrase structure will also be provided. The method described here may have its flaws because of limited prior information. This, however, may offer a good starting point to study languages where only little prior knowledge is available and a detailed, unbiased study is needed.

Keywords: alignments, bioinformatics, comparative linguistics, historical linguistics, statistical methods

Procedia PDF Downloads 129

Authors: Ebipatei Victoria Tunyan, T. A. Cao, Cheol Young Ock

Abstract:

Detecting subjectively biased statements is a vital task. This is because this kind of bias, when present in the text or other forms of information dissemination media such as news, social media, scientific texts, and encyclopedias, can weaken trust in the information and stir conflicts amongst consumers. Subjective bias detection is also critical for many Natural Language Processing (NLP) tasks like sentiment analysis, opinion identification, and bias neutralization. Having a system that can adequately detect subjectivity in text will boost research in the above-mentioned areas significantly. It can also come in handy for platforms like Wikipedia, where the use of neutral language is of importance. The goal of this work is to identify the subjectively biased language in text on a sentence level. With machine learning, we can solve complex AI problems, making it a good fit for the problem of subjective bias detection. A key step in this approach is to train a classifier based on BERT (Bidirectional Encoder Representations from Transformers) as upstream model. BERT by itself can be used as a classifier; however, in this study, we use BERT as data preprocessor as well as an embedding generator for a Bi-LSTM (Bidirectional Long Short-Term Memory) network incorporated with attention mechanism. This approach produces a deeper and better classifier. We evaluate the effectiveness of our model using the Wiki Neutrality Corpus (WNC), which was compiled from Wikipedia edits that removed various biased instances from sentences as a benchmark dataset, with which we also compare our model to existing approaches. Experimental analysis indicates an improved performance, as our model achieved state-of-the-art accuracy in detecting subjective bias. This study focuses on the English language, but the model can be fine-tuned to accommodate other languages.

Keywords: subjective bias detection, machine learning, BERT–BiLSTM–Attention, text classification, natural language processing

Procedia PDF Downloads 99

Authors: Entisar A. Elgmati

Abstract:

Recurrent infant diarrhea is studied using daily data collected in Salvador, Brazil over one year and three months. A logistic regression model is fitted instead of Aalen's additive model using the same covariates that were used in the analysis with the additive model. The model gives reasonably similar results to that using additive regression model. In addition, the problem with the estimated conditional probabilities not being constrained between zero and one in additive model is solved here. Also martingale residuals that have been used to judge the goodness of fit for the additive model are shown to be useful for judging the goodness of fit of the logistic model.

Keywords: additive model, cumulative probabilities, infant diarrhoea, recurrent event

Procedia PDF Downloads 608

Authors: Feiying Wu

Abstract:

Chatbots are computer programs that attempt to engage a human in a dialogue, which originated in the 1960s with MIT's Eliza. However, they have become widespread more recently as advances in language technology have produced chatbots with increasing linguistic quality and sophistication, leading to their potential to serve as a tool for Computer-Assisted Language Learning(CALL). The aim of this article is to assess the feasibility of using two chatbots, Mitsuku and CleverBot, as pedagogical tools for learning English as a second language by stimulating L2 learners with distinct English proficiencies. Speaking of the input of stimulated learners, they are measured by AntWordProfiler to match the user's expected vocabulary proficiency. Totally, there are four chat sessions as each chatbot will converse with both beginners and advanced learners. For evaluation, it focuses on chatbots' responses from a linguistic standpoint, encompassing vocabulary and sentence levels. The vocabulary level is determined by the vocabulary range and the reaction to misspelled words. Grammatical accuracy and responsiveness to poorly formed sentences are assessed for the sentence level. In addition, the assessment of this essay sets 25% lexical and grammatical incorrect input to determine chatbots' corrective ability towards different linguistic forms. Based on statistical evidence and illustration of examples, despite the small sample size, neither Mitsuku nor CleverBot is ideal as educational tools based on their performance through word range, grammatical accuracy, topic range, and corrective feedback for incorrect words and sentences, but rather as a conversational tool for beginners of L2 English.

Keywords: chatbots, CALL, L2, corrective feedback

Procedia PDF Downloads 54

Authors: Vishal Raj, Noorhan Abbas

Abstract:

Ambiguity in NLP (Natural language processing) refers to the ability of a word, phrase, sentence, or text to have multiple meanings. This results in various kinds of ambiguities such as lexical, syntactic, semantic, anaphoric and referential am-biguities. This study is focused mainly on solving the issue of Lexical ambiguity. Word Sense Disambiguation (WSD) is an NLP technique that aims to resolve lexical ambiguity by determining the correct meaning of a word within a given context. Most WSD solutions rely on words for training and testing, but we have used lemma and Part of Speech (POS) tokens of words for training and testing. Lemma adds generality and POS adds properties of word into token. We have designed a novel method to create an affinity matrix to calculate the affinity be-tween any pair of lemma_POS (a token where lemma and POS of word are joined by underscore) of given training set. Additionally, we have devised an al-gorithm to create the sense clusters of tokens using affinity matrix under hierar-chy of POS of lemma. Furthermore, three different mechanisms to predict the sense of target word using the affinity/similarity value are devised. Each contex-tual token contributes to the sense of target word with some value and whichever sense gets higher value becomes the sense of target word. So, contextual tokens play a key role in creating sense clusters and predicting the sense of target word, hence, the model is named Contextual SenSe Model (CSM). CSM exhibits a noteworthy simplicity and explication lucidity in contrast to contemporary deep learning models characterized by intricacy, time-intensive processes, and chal-lenging explication. CSM is trained on SemCor training data and evaluated on SemEval test dataset. The results indicate that despite the naivety of the method, it achieves promising results when compared to the Most Frequent Sense (MFS) model.

Keywords: word sense disambiguation (wsd), contextual sense model (csm), most frequent sense (mfs), part of speech (pos), natural language processing (nlp), oov (out of vocabulary), lemma_pos (a token where lemma and pos of word are joined by underscore), information retrieval (ir), machine translation (mt)

Procedia PDF Downloads 69

Authors: Kenji Gwee

Abstract:

In Singapore, the death penalty remains in active use for murder and drug trafficking of controlled drugs such as heroin. As such, the psychological assessment of defendants can often be of high stakes. The Test of Memory Malingering (TOMM) is employed by government psychologists to determine the degree of effort invested by defendants, which in turn inform on the veracity of overall psychological findings that can invariably determine the life and death of defendants. The purpose of this study was to find out if defendants facing the death penalty were more likely to invest less effort during psychological assessment (to fake bad in hopes of escaping the death sentence) compared to defendants facing lesser penalties. An archival search of all forensic cases assessed in 2012-2013 by Singapore’s designated forensic psychiatric facility yielded 186 defendants’ TOMM scores. Offence severity, coded into 6 rank-ordered categories, was analyzed in a one-way ANOVA with TOMM score as the dependent variable. There was a statistically significant difference (F(5,87) = 2.473, p = 0.038). A Tukey post-hoc test with Bonferroni correction revealed that defendants facing lower charges (Theft, shoplifting, criminal breach of trust) invested less test-taking effort (TOMM = 37.4±12.3, p = 0.033) compared to those facing the death penalty (TOMM = 46.2±8.1). The surprising finding that those facing death penalties actually invested more test taking effort than those facing relatively minor charges could be due to higher levels of cooperation when faced with death. Alternatively, other legal avenues to escape the death sentence may have been preferred over the mitigatory chance of a psychiatric defence.

Keywords: capital sentencing, offence severity, Singapore, Test of Memory Malingering

Procedia PDF Downloads 406

Authors: Sherly Ferro Lensun, Donal Matheos Ratu, Elni Jeini Usoh, Helena M. L. Pandi, Mayske Rinny Liando

Abstract:

This study aims to gain a deep understanding of the process and the increase resulting in mastery of Kanji with a Language Learning Strategies through multimedia. This research aims to gain scientific data on process and the result of improving kanji mastery by using Chokusetsu strategy in Kanji learning. The method used in this research is Action Research developed by Kemmis and Mc. Taggart is known as Spiral Model. This model consists of following stages: planning, implementation, observation, and reflection. The research results in following findings: (1) Kanji mastery comprises 4 major aspects, those are reading, writing, the use in sentence, and memorizing, and those aspects show gradual improvement from time to time. (2) Students have more participation in learning activities which can be identified from some positive behaviours such giving respond in finishing exercise in class. (3) Students’ better attention to the lesson shown by active behaviour in giving more questions or asking for more explanation to the lecturers, memorizing Kanji card, finishing the task of making Kanji card/house, doing the exercises more seriously, and finishing homework assignment punctually. (4) More attractive learning activities and tasks in the forms of more engaging colour and pictures enables students to conduct self-evaluation on their learning process.

Keywords: Kanji, action research, language learning strategies, multimedia

Procedia PDF Downloads 154

Authors: Hui Shi

Abstract:

This article examines the online popularity of Chinese neologism simida, which is a loanword derived from Korean declarative sentence-final suffix seumnida. Facilitated by corpus data obtained from Weibo, the Chinese counterpart of Twitter, this study analyzes the morphological and syntactical processes behind simida’s coinage, as well as the causes of its prevalence on Chinese social media. The findings show that simida is used by Weibo bloggers in two manners: (1) as an alternative word of 'Korea' and 'Korean'; (2) as a redundant sentence-final particle which adds a Korean-like speech style to a statement. Additionally, Weibo user profile analysis further reveals demographical distribution patterns concerning this neologism and highlights young Weibo users in the third-tier cities as the leading adopters of simida. These results are accounted for under the theoretical framework of social indexicality, especially how variations generate style in the indexical field. This article argues that the creation of such an ethnically-targeted neologism is a linguistic demonstration of Chinese netizen’s two-sided attitudes toward the previously heated Korean-wave. The exotic suffix seumnida is borrowed to Chinese as simida due to its high-frequency in Korean cultural exports. Therefore, it gradually becomes a replacement of Korea-related lexical items due to markedness, regardless of semantic prosody. Its innovative implantation to Chinese syntax, on the other hand, reflects Chinese netizens’ active manipulation of language for their online identity building. This study has implications for research on the linguistic construction of identity and style and lays the groundwork for linguistic creativity in the Chinese new media.

Keywords: Chinese neologism, loanword, humor, new media

Procedia PDF Downloads 153

Authors: Linnan Zhang

Abstract:

Linguists, as well as psychologists, have shown great interests in implicit causality in reference processing. However, most frequently-used approaches to this issue are psychological experiments (such as eye tracking or self-paced reading, etc.). This research is a corpus-based one and is assisted with statistical tool – software R. The main focus of the present study is about the competition between verb-based implicit causality and theme structure’s influence on anaphora bias in Mandarin Chinese sentences. In Accessibility Theory, it is believed that salience, which is also known as accessibility, and relevance are two important factors in reference processing. Theme structure, which is a special syntactic structure in Chinese, determines the salience of an antecedent on the syntactic level while verb-based implicit causality is a key factor to the relevance between antecedent and anaphora. Therefore, it is a study about anaphora, combining psychology with linguistics. With analysis of the sentences from corpus as well as the statistical analysis of Multinomial Logistic Regression, major findings of the present study are as follows: 1. When the sentence is stated in a ‘cause-effect’ structure, the theme structure will always be the antecedent no matter forward biased verbs or backward biased verbs co-occur; in non-theme structure, the anaphora bias will tend to be the opposite of the verb bias; 2. When the sentence is stated in a ‘effect-cause’ structure, theme structure will not always be the antecedent and the influence of verb-based implicit causality will outweigh that of theme structure; moreover, the anaphora bias will be the same with the bias of verbs. All the results indicate that implicit causality functions conditionally and the noun in theme structure will not be the high-salience antecedent under any circumstances.

Keywords: accessibility theory, anaphora, theme strcture, verb-based implicit causality

Procedia PDF Downloads 168

Authors: Manal Mohammed Hisham Said Najjar

Abstract:

The Arab Grammarians shed light on the redundant kana (was) and gave it a considerable attention. However, their considerations and interpretations pertaining to using this verb varied: is it used to determine tense? Or used for further emphasis or for another function? Does it have a syntactic function? Morphologically, could it be used in other forms than the past? In addition, Arab Grammarians discussed the possibility of using kana to locate itself in between the syntactic constructs of a sentence, a phrase, or a collocation. Others questioned its position whether it is in initial or final. This study found out that the redundant kana (was) is cited in Quran and was used by the Arabs in their speech and poetry. This redundant kana, whether used in initial position or in a final position, or in between the constructs of a sentence, a phrase, or a collocation, implies pragmatic meanings intended by the speaker or the poet to serve different functions, such as to indicate the past tense, to provide emphasis, and to refer to the continuity of the effect and meaning of a verb or adjective. The study concludes that this verb kana can be utilized in different contexts to achieve a specific effect as did the old Arabs who used it to add specific shades of meanings. Kana as a redundant word could be added to further highlight the meaning aimed at in a specific utterance. In addition, this verb can be used in both the past and the present morphological form; and its availability in an utterance could be functional and could not be. In other words, the study found out that the redundant kana can be used in various positions in an utterance, initial, final, or in between a syntactic structure, provided that this use is pragmatically functional. In conclusion, this paper seeks to invite the scholars of the Arabic language to coin a new term which is the “pragmatic kana” to replace the term “kana alzae’da (redundant kana)” which might mean that its use is redundant and void of significance – a fact that is illogical due to its recurrent use in the Holy Quran. NOTE: Please take this study not the other one (sent by mistake) and titled kana alnaqisa

Keywords: redundan, kana, grammarians, quran

Procedia PDF Downloads 97

Authors: Tatiana Stanchuliak

Abstract:

It is common knowledge that intonation is not laid over a ready text. On the contrary, intonation forms and accompanies the text on the level of its birth in the speaker’s mind. As a result, intonation plays one of the fundamental roles in the process of transferring a thought into external speech. Intonation structure can highlight the semantic significance of textual elements and become a ranging mark in understanding the information structure of the text. Intonation functions by means of prosodic characteristics, one of which is intonation salience, whose function in texts results in making some textual elements more prominent than others. This function of intonation, therefore, performs as organizing. It helps to form the frame of key elements of the text. The study under consideration made an attempt to look into the inner nature of salience and create a sort of a text intonation model. This general goal brought to some more specific intermediate results. First, there were established degrees of salience on the level of the smallest semantic element - intonation group, as well as prosodic means of creating salience, were examined. Second, the most frequent combinations of prosodic means made it possible to distinguish patterns of salience, which then became constituent elements of a text intonation model. Third, the analysis of the predicate structure allowed to divide the whole text into smaller parts, or units, which performed a specific function in the developing of the general communicative intention. It appeared that such units can be found in any text and they have common characteristics of their intonation arrangement. These findings are certainly very important both for the theory of intonation and their practical application.

Keywords: accentuation , inner speech, intention, intonation, intonation functions, models, patterns, predicate, salience, semantics, sentence stress, text

Procedia PDF Downloads 235

Authors: Mechlouch Ridha Fethi

Abstract:

This paper presents a recent mathematical model in political sciences concerning democracy. The model is represented by a logarithmic equation linking the Relative Index of Democracy (RID) to Participation Ratio (PR). Firstly the meanings of the different parameters of the model were presented; and the variation curve of the RID according to PR with different critical areas was discussed. Secondly, the model was applied to a virtual group where we show that the model can be applied depending on the gender. Thirdly, it was observed that the model can be extended to different language models of democracy and that little use to assess the state of democracy for some International organizations like UNO.

Keywords: democracy, mathematic, modelization, quantification

Procedia PDF Downloads 332

Authors: Ganga Brahma

Abstract:

Boro employs verbs hooked up with morphemes which lead verbs to adjust with their arguments and hence, affecting the whole of sentence structures. This paper is based on few such syntactic parameters which are usually considered as argument decreasing parameters in linguistic works. Passivizing of few transitive clauses which are usually construed from the verbs occurring with certain morphemes and representation in middle constructions are few of such strategies which lead to conceptualizing of decreasing of syntactic arguments from a sentence. This paper focuses on the mentioned linguistic strategies and attempts to describe the linguistic processes as for how these parameters work in languages especially by concentrating on a particular Tibeto-Burman language i.e. Boro. Boro is a Tibeto-Burman language widely spoken in parts of the north-eastern regions of India. It has an agglutinative nature in forming words as well as clauses. There is a morpheme ‘za’ which means ‘to happen, become’ in Boro whose appearances with verb roots denotes an idea of the subject being passivized. Passivization, usually has notions that it is a reversed representation of its active sentence forms in the terms of argument placements. (However, it is not accountably true as passives and actives have some distinct features of their own and independent of one and the other.) This particular work will concentrate on the semantics of passivization at the same time along with its syntactic reality. The verb khɑo meaning ‘to steal’ offers a sense of passivization with the appearance of the morpheme zɑ which means ‘to happen, become’ (e.g Zunu-ɑ lama-ɑo phɯisɑ khɑo-zɑ-bɑi; Junu-NOM road-LOC money steal-PASS-PRES: Junu got her money stolen on the road). The focus, here, is more on the argument placed at the subject position (i.e. Zunu) and the event taken place. The semantics of such construction asks for the agent because without an agent the event could not have taken place. However, the syntactic elements fill the slots of relegated or temporarily deleted agent which, infact, is the actual subject cum agent in its active representation. Due to the event marker ‘zɑ’ in this presentation it affords to reduce one participant from such a situation which in actual is made up of three participants. Hence, the structure of di-transitive construction here reduces to mono-transitive structure. Unlike passivization, middle construction does not allow relegation of the agents. It permanently deletes agents. However, it also focuses on the fore-grounded subject and highlighting on the changed states on the subjects which happens to be the underlying objects of their respective transitive structures (with agents). This work intends to describe how these two parameters which are different at their semantic realization can meet together at a syntactic level in order to create a linguistic parameter that decreases participants from their actual structures which are with more than one participant.

Keywords: argument-decrease, middle-construction, passivization, transitivity-intransitivity

Procedia PDF Downloads 210

Authors: Le Kang

Abstract:

On the research question of 'how to achieve USR', this contribution reflects the concept of university social responsibility, identify three achievement models of USR as the society - diversified model, the university-cooperation model, the government - compound model, also conduct a case study to explore characteristics of Chinese achievement model of USR. The contribution concludes with discussion of how the university, government and society balance demands and roles, make necessarily strategic adjustment and innovative approach to repair the shortcomings of each achievement model.

Keywords: modern university, USR, achievement model, compound model

Procedia PDF Downloads 724