Abstract: Analysis of the human microbiome using metagenomic
sequencing data has demonstrated high ability in discriminating
various human diseases. Raw metagenomic sequencing data require
multiple complex and computationally heavy bioinformatics steps
prior to data analysis. Such data contain millions of short sequences
read from the fragmented DNA sequences and stored as fastq files.
Conventional processing pipelines consist in multiple steps including
quality control, filtering, alignment of sequences against genomic
catalogs (genes, species, taxonomic levels, functional pathways,
etc.). These pipelines are complex to use, time consuming and
rely on a large number of parameters that often provide variability
and impact the estimation of the microbiome elements. Training
Deep Neural Networks directly from raw sequencing data is a
promising approach to bypass some of the challenges associated with
mainstream bioinformatics pipelines. Most of these methods use the
concept of word and sentence embeddings that create a meaningful
and numerical representation of DNA sequences, while extracting
features and reducing the dimensionality of the data. In this paper
we present an end-to-end approach that classifies patients into disease
groups directly from raw metagenomic reads: metagenome2vec. This
approach is composed of four steps (i) generating a vocabulary of
k-mers and learning their numerical embeddings; (ii) learning DNA
sequence (read) embeddings; (iii) identifying the genome from which
the sequence is most likely to come and (iv) training a multiple
instance learning classifier which predicts the phenotype based on
the vector representation of the raw data. An attention mechanism
is applied in the network so that the model can be interpreted,
assigning a weight to the influence of the prediction for each genome.
Using two public real-life data-sets as well a simulated one, we
demonstrated that this original approach reaches high performance,
comparable with the state-of-the-art methods applied directly on
processed data though mainstream bioinformatics workflows. These
results are encouraging for this proof of concept work. We believe
that with further dedication, the DNN models have the potential to
surpass mainstream bioinformatics workflows in disease classification
tasks.
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.
Abstract: The low availability of well-trained, unlimited, dynamic-access models for specific languages makes it hard for corporate users to adopt quick translation techniques and incorporate them into product solutions. As translation tasks increasingly require a dynamic sequence learning curve; stable, cost-free opensource models are scarce. We survey and compare current translation techniques and propose a modified sequence to sequence model repurposed with attention techniques. Sequence learning using an encoder-decoder model is now paving the path for higher precision levels in translation. Using a Convolutional Neural Network (CNN) encoder and a Recurrent Neural Network (RNN) decoder background, we use Fairseq tools to produce an end-to-end bilingually trained Spanish-English machine translation model including source language detection. We acquire competitive results using a duo-lingo-corpus trained model to provide for prospective, ready-made plug-in use for compound sentences and document translations. Our model serves a decent system for large, organizational data translation needs. While acknowledging its shortcomings and future scope, it also identifies itself as a well-optimized deep neural network model and solution.
Abstract: This paper presents and benchmarks a number of
end-to-end Deep Learning based models for metaphor detection in
Greek. We combine Convolutional Neural Networks and Recurrent
Neural Networks with representation learning to bear on the metaphor
detection problem for the Greek language. The models presented
achieve exceptional accuracy scores, significantly improving the
previous state-of-the-art results, which had already achieved accuracy
0.82. Furthermore, no special preprocessing, feature engineering or
linguistic knowledge is used in this work. The methods presented
achieve accuracy of 0.92 and F-score 0.92 with Convolutional
Neural Networks (CNNs) and bidirectional Long Short Term Memory
networks (LSTMs). Comparable results of 0.91 accuracy and 0.91
F-score are also achieved with bidirectional Gated Recurrent Units
(GRUs) and Convolutional Recurrent Neural Nets (CRNNs). The
models are trained and evaluated only on the basis of training tuples,
the related sentences and their labels. The outcome is a state-of-the-art
collection of metaphor detection models, trained on limited labelled
resources, which can be extended to other languages and similar
tasks.
Abstract: The Arabic language is one of the most important languages. Learning it is so important for many people around the world because of its religious and economic importance and the real challenge lies in practicing it without grammatical or syntactical mistakes. This research focused on detecting and correcting the syntactic mistakes of Arabic syntax according to their position in the sentence and focused on two of the main syntactical rules in Arabic: Dual and Plural. It analyzes each sentence in the text, using Stanford CoreNLP morphological analyzer and machine-learning approach in order to detect the syntactical mistakes and then correct it. A prototype of the proposed system was implemented and evaluated. It uses support vector machine (SVM) algorithm to detect Arabic grammatical errors and correct them using the rule-based approach. The prototype system has a far accuracy 81%. In general, it shows a set of useful grammatical suggestions that the user may forget about while writing due to lack of familiarity with grammar or as a result of the speed of writing such as alerting the user when using a plural term to indicate one person.
Abstract: The present paper is preoccupied with the different styles of three translators in their translating a Chinese classical novel Shuihu Zhuan. Based on a parallel corpus, it adopts a target-oriented approach to look into whether and what stylistic differences and shifts the three translations have revealed. The findings show that the three translators demonstrate different styles concerning their word choices and sentence preferences, which implies that identification of recurrent textual patterns may be a basic step for investigating the style of a translator.
Abstract: The more an educational system knows about a learner, the more personalised interaction it can provide, which leads to better learning. However, asking a learner directly is potentially disruptive, and often ignored by learners. Especially in the booming realm of MOOC Massive Online Learning platforms, only a very low percentage of users disclose demographic information about themselves. Thus, in this paper, we aim to predict learners’ demographic characteristics, by proposing an approach using linguistically motivated Deep Learning Architectures for Learner Profiling, particularly targeting gender prediction on a FutureLearn MOOC platform. Additionally, we tackle here the difficult problem of predicting the gender of learners based on their comments only – which are often available across MOOCs. The most common current approaches to text classification use the Long Short-Term Memory (LSTM) model, considering sentences as sequences. However, human language also has structures. In this research, rather than considering sentences as plain sequences, we hypothesise that higher semantic - and syntactic level sentence processing based on linguistics will render a richer representation. We thus evaluate, the traditional LSTM versus other bleeding edge models, which take into account syntactic structure, such as tree-structured LSTM, Stack-augmented Parser-Interpreter Neural Network (SPINN) and the Structure-Aware Tag Augmented model (SATA). Additionally, we explore using different word-level encoding functions. We have implemented these methods on Our MOOC dataset, which is the most performant one comparing with a public dataset on sentiment analysis that is further used as a cross-examining for the models' results.
Abstract: The recognition of the objects contained in images has always presented a challenge in the field of research because of several difficulties that the researcher can envisage because of the variability of shape, position, contrast of objects, etc. In this paper, we will be interested in the recognition of objects. The classical Hough Transform (HT) presented a tool for detecting straight line segments in images. The technique of HT has been generalized (GHT) for the detection of arbitrary forms. With GHT, the forms sought are not necessarily defined analytically but rather by a particular silhouette. For more precision, we proposed to combine the results from the GHT with the results from a calculation of similarity between the histograms and the spatiograms of the images. The main purpose of our work is to use the concepts from recognition to generate sentences in Arabic that summarize the content of the image.
Abstract: This study is intended to describe and analyze the effects of polysyllabic shortening and word or phrase boundary on the duration patterns of spoken utterances by Mandarin learners of English in comparison with native speakers of English. To investigate the relative contribution of these effects, two production experiments were conducted. The study included 11 native British English speakers and 20 Mandarin learners of English who were asked to produce four sets of tokens consisting of a mono-syllabic base form, disyllabic, and trisyllabic words derived from the base by the addition of suffixes, and a set of short sentences with a particular combination of phrase size, stress pattern, and boundary location. The duration of words and segments was measured, and results from the data analysis suggest that the amount of polysyllabic shortening and the effect of word or phrase position are likely to affect a Chinese accent for Mandarin ESL speakers. This study sheds light on research on the duration patterns of language by demonstrating the effect of duration-related factors on the foreign accent of Mandarin ESL speakers. It can also benefit both L2 learners and language teachers by increasing their sensitivity to the duration differences and difficulties experienced by L2 learners of English. An understanding of the amount of polysyllabic shortening and the effect of position in words and phrase on syllable duration can also facilitate L2 teachers to establish priorities for teaching pronunciation to ESL learners.
Abstract: Sentiment analysis is a recent field of study that computationally assesses the emotional nature of a body of text. To assess its test-validity, sentiment analysis was carried out on the emotional corpus of text from a personal 15-day mood diary. Self-reported mood scores varied more or less accurately with daily mood evaluation score given by the software. On further assessment, it was found that while sentiment analysis was good at assessing ‘global’ mood, it was not able to ‘locally’ identify and differentially score synonyms of various emotional words. It is further critiqued for treating the intensity of an emotion as universal across cultures. Finally, the software is shown not to account for emotional complexity in sentences by treating emotions as strictly positive or negative. Hence, it is posited that a better output could be two (positive and negative) affect scores for the same body of text.
Abstract: Article 3 of the European Convention for the Protection of Human Rights and Fundamental Freedoms (E.C.H.R.) proclaims that no one may be subjected to torture, punishment or degrading treatment. The legislative correlate in Spain is embodied in Article 15 of the Spanish Constitution, and there must be an overlapping interpretation of both precepts on the ideal plane. While it is true that there are not many cases in which the European Court of Human Rights (E.C.t.H.R. (The Strasbourg Court)) has sanctioned Spain for its failure to investigate complaints of torture, it must be emphasized that the tendency to violate Article 3 of the Convention appears to be on the rise, being necessary to know possible factors that may be affecting it. This paper addresses the analysis of sentences that directly or indirectly reveal the violation of Article 3 of the European Convention. To carry out the analysis, sentences of the Strasbourg Court have been consulted from 2012 to 2016, being able to address any previous sentences to this period if it provided justified information necessary for the study. After the review it becomes clear that there are two key groups of subjects that request a response to the Strasbourg Court on the understanding that they have been tortured or degradingly treated. These are: immigrants and terrorists. Both phenomena, immigration and terrorism, respond to patterns that have mutated in recent years, and it is important for this study to know if national regulations begin to be dysfunctional.
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.
Abstract: Sign Language (SL) is used by deaf and other people who cannot speak but can hear or have a problem with spoken languages due to some disability. It is a visual gesture language that makes use of either one hand or both hands, arms, face, body to convey meanings and thoughts. SL automation system is an effective way which provides an interface to communicate with normal people using a computer. In this paper, an avatar based dictionary has been proposed for text to Indian Sign Language (ISL) generation system. This research work will also depict a literature review on SL corpus available for various SL s over the years. For ISL generation system, a written form of SL is required and there are certain techniques available for writing the SL. The system uses Hamburg sign language Notation System (HamNoSys) and Signing Gesture Mark-up Language (SiGML) for ISL generation. It is developed in PHP using Web Graphics Library (WebGL) technology for 3D avatar animation. A multilingual ISL dictionary is developed using HamNoSys for both English and Hindi Language. This dictionary will be used as a database to associate signs with words or phrases of a spoken language. It provides an interface for admin panel to manage the dictionary, i.e., modification, addition, or deletion of a word. Through this interface, HamNoSys can be developed and stored in a database and these notations can be converted into its corresponding SiGML file manually. The system takes natural language input sentence in English and Hindi language and generate 3D sign animation using an avatar. SL generation systems have potential applications in many domains such as healthcare sector, media, educational institutes, commercial sectors, transportation services etc. This research work will help the researchers to understand various techniques used for writing SL and generation of Sign Language systems.
Abstract: This study presents a methodology of specialized translation with the objective of helping teachers to improve the strategies in teaching translation. In order to allow students to acquire skills to translate specialized texts, they need to become familiar with the semantic and syntactic features of source texts and target texts. The aim of our study is to use a corpus-based approach in the teaching of specialized translation between Chinese and Italian. This study proposes to construct a specialized Chinese - Italian comparable corpus that consists of 50 economic contracts from the domain of food. With the help of AntConc, we propose to compile a comparable corpus in for translation teaching purposes. This paper attempts to provide insight into how teachers could benefit from comparable corpus in the teaching of specialized translation from Italian into Chinese and through some examples of passive sentences how students could learn to apply different strategies for translating appropriately the voice.
Abstract: Phonological disorder is a serious and disturbing issue to many parents and teachers. Efforts towards resolving the problem have been undermined by other specific disabilities which were hidden to many regular and special education teachers. It is against this background that this study was motivated to provide data on the prevalence of phonological disorders in children with specific language impairment (CWSLI) as the first step towards critical intervention. The study was a survey of 15 CWSLI from St. Louise Inclusive schools, Ikot Ekpene in Akwa Ibom State of Nigeria. Phonological Processes Diagnostic Scale (PPDS) with 17 short sentences, which cut across the five phonological processes that were examined, were validated by experts in test measurement, phonology and special education. The respondents were made to read the sentences with emphasis on the targeted sounds. Their utterances were recorded and analyzed in the language laboratory using Praat Software. Data were also collected through friendly interactions at different times from the clients. The theory of generative phonology was adopted for the descriptive analysis of the phonological processes. Data collected were analyzed using simple percentage and composite bar chart for better understanding of the result. The study found out that CWSLI exhibited the five phonological processes under investigation. It was revealed that 66.7%, 80%, 73.3%, 80%, and 86.7% of the respondents have severe deficit in fricative stopping, velar fronting, liquid gliding, final consonant deletion and cluster reduction, respectively. It was therefore recommended that a nationwide survey should be carried out to have national statistics of CWSLI with phonological deficits and develop intervention strategies for effective therapy to remediate the disorder.
Abstract: Speaking skills merit meticulous attention both on the side of the learners and the teachers. In particular, accuracy is a critical component to guarantee the messages to be conveyed through conversation because a wrongful change may adversely alter the content and purpose of the talk. Different types of tasks have served teachers to meet numerous educational objectives. Besides, negotiation of meaning and the use of different strategies have been areas of concern in socio-cultural theories of SLA. Negotiation of meaning is among the conversational processes which have a crucial role in facilitating the understanding and expression of meaning in a given second language. Conversational strategies are used during interaction when there is a breakdown in communication that leads to the interlocutor attempting to remedy the gap through talk. Therefore, this study was an attempt to investigate if there was any significant difference between the effect of reasoning gap tasks and information gap tasks on the frequency of conversational strategies used in negotiation of meaning in classrooms on one hand, and on the accuracy in speaking of Iranian intermediate EFL learners on the other. After a pilot study to check the practicality of the treatments, at the outset of the main study, the Preliminary English Test was administered to ensure the homogeneity of 87 out of 107 participants who attended the intact classes of a 15 session term in one control and two experimental groups. Also, speaking sections of PET were used as pretest and posttest to examine their speaking accuracy. The tests were recorded and transcribed to estimate the percentage of the number of the clauses with no grammatical errors in the total produced clauses to measure the speaking accuracy. In all groups, the grammatical points of accuracy were instructed and the use of conversational strategies was practiced. Then, different kinds of reasoning gap tasks (matchmaking, deciding on the course of action, and working out a time table) and information gap tasks (restoring an incomplete chart, spot the differences, arranging sentences into stories, and guessing game) were manipulated in experimental groups during treatment sessions, and the students were required to practice conversational strategies when doing speaking tasks. The conversations throughout the terms were recorded and transcribed to count the frequency of the conversational strategies used in all groups. The results of statistical analysis demonstrated that applying both the reasoning gap tasks and information gap tasks significantly affected the frequency of conversational strategies through negotiation. In the face of the improvements, the reasoning gap tasks had a more significant impact on encouraging the negotiation of meaning and increasing the number of conversational frequencies every session. The findings also indicated both task types could help learners significantly improve their speaking accuracy. Here, applying the reasoning gap tasks was more effective than the information gap tasks in improving the level of learners’ speaking accuracy.
Abstract: To improve students’ skill, writing is the most challenging skill to be developed. The reason is that besides helping the students to develop their skill, this activity also helps them to express themselves. This paper depicts how paraphrasing is very helpful to interpret students’ writing. Syntactic units, used tenses and meanings will indeed change once the writings were paraphrased. The objectives of this research are to reveal the inappropriate structure of syntactic units, to show what types of sentences the students often make, and to show how paraphrasing can help to infer the message. The methodology of this research is descriptive qualitative research. In addition, theories of linguistics are also included. This includes theory of Syntax to describe syntactic units and tenses and theory of Semantics to describe theories of meaning and how paraphrasing works. The theories of general linguistics, grammar and writing are also provided to support the theories of Syntax and Semantics. The results of this research are concerned with how the message is received in the end. The message written in the students’ essay is not clear because of the improper structure of syntactic units and use of incorrect of tenses. The students tend to use simple sentences, compound sentences and complex sentences with a few mistakes in their writing. In addition, they tend to create unnecessary phrases. The last point is that this research shows how paraphrase works to attain complete meaning of a sentence.
Abstract: The enthusiasm of foreigners studying the Indonesian language by Foreign Speakers (BIPA) was documented in a sitcom "International Class". Tone and stress when they speak the Indonesian language is unique and different from Indonesian pronunciation. By using the Praat program, this research aims to describe prosodic Indonesian language which is spoken by ‘International Class” actors consisting of Abbas from Nigeria, Lee from Korea, and Kotaro from Japan. Data for the research are taken from the video sitcom "International Class" that aired on Indonesian television. The results of this study revealed that pitch movement that arises when pronouncing Indonesian sentences was up and down gradually, there is also a rise and fall sharply. In terms of stress, respondents tend to contain a lot of stress when pronouncing Indonesian sentences. Meanwhile, in terms of temporal structure, the duration pronouncing Indonesian sentences tends to be longer than that of Indonesian speakers.
Abstract: The present approach deals with the identification of Emotions and classification of Emotional patterns at Phrase-level with respect to Positive and Negative Orientation. The proposed approach considers emotion triggered terms, its co-occurrence terms and also associated sentences for recognizing emotions. The proposed approach uses Part of Speech Tagging and Emotion Actifiers for classification. Here sentence patterns are broken into phrases and Neuro-Fuzzy model is used to classify which results in 16 patterns of emotional phrases. Suitable intensities are assigned for capturing the degree of emotion contents that exist in semantics of patterns. These emotional phrases are assigned weights which supports in deciding the Positive and Negative Orientation of emotions. The approach uses web documents for experimental purpose and the proposed classification approach performs well and achieves good F-Scores.
Abstract: Over the past decade, there have been promising developments in Natural Language Processing (NLP) with several investigations of approaches focusing on Recognizing Textual Entailment (RTE). These models include models based on lexical similarities, models based on formal reasoning, and most recently deep neural models. In this paper, we present a sentence encoding model that exploits the sentence-to-sentence relation information for RTE. In terms of sentence modeling, Convolutional neural network (CNN) and recurrent neural networks (RNNs) adopt different approaches. RNNs are known to be well suited for sequence modeling, whilst CNN is suited for the extraction of n-gram features through the filters and can learn ranges of relations via the pooling mechanism. We combine the strength of RNN and CNN as stated above to present a unified model for the RTE task. Our model basically combines relation vectors computed from the phrasal representation of each sentence and final encoded sentence representations. Firstly, we pass each sentence through a convolutional layer to extract a sequence of higher-level phrase representation for each sentence from which the first relation vector is computed. Secondly, the phrasal representation of each sentence from the convolutional layer is fed into a Bidirectional Long Short Term Memory (Bi-LSTM) to obtain the final sentence representations from which a second relation vector is computed. The relations vectors are combined and then used in then used in the same fashion as attention mechanism over the Bi-LSTM outputs to yield the final sentence representations for the classification. Experiment on the Stanford Natural Language Inference (SNLI) corpus suggests that this is a promising technique for RTE.