Search

Identifying Financial Statement Fraud (FSF) events is very crucial in text mining. The researcher’s community is mostly utilized the data mining method for detecting FSF. In this direction, mostly the quantitative data has utilized by research i.e. the financial ratio is presented for detecting fraud in financial statements. On the text investigation there is no researches like auditor's remarks present in published reports. For this reason, this paper develops the optimized deep neural network-based FSF detection in the qualitative data present in financial reports. The pre-processing of text is performed initially using filtering, lemmatization, and tokenization. Then, the feature selection is done by the Harris Hawks Optimization (HHO) algorithm. Finally, a Deep Neural Network-Based Deer Hunting Optimization (DNN-DHO) is utilized to identify the fraud or no-fraud report in the financial statements. The developed FSF detection methodology executed in Python environment using financial statement datasets. The output of the developed approach gives high classification accuracy (96%) in comparison to the standard classifiers like DNN, CART, LR, SVM, Bayes, BP-NN, and KNN. Also, it provides better outcomes in all performance metrics.

PurposeBlockchain and distributed ledger technologies are set to disrupt the real estate sector in all areas: ownership, sale, management and investment. Tokenization moves physical real estate to the digital space and could result in substantial cost savings in the pre- and post-tokenization process. This article discusses whether real estate as an asset class is ready for digitalization in the Asia-Pacific (APAC) region.Design/methodology/approachGlobally, the APAC region has the highest digital adaptation/adoption rates. Regulators in the region are also moving fast to clarify their stance on digital assets. This article adopts a holistic view, from trends, regulations, and technology, to discuss the benefits and challenges of digitalizing real estate in APAC.FindingsReal estate tokenization is a nascent market but platforms like BrickX, KASA, ADDX, and Minterest have successfully launched real estate tokens in Australia, South Korea, and Singapore, respectively. Tokenization may prove to be a viable funding source for those relatively poorly capitalized financial markets in the APAC region.Practical implicationsThis paper discusses the current regulatory and business contexts in relation to the pace of tokenization of real estate in APAC. Opportunities and difficulties are outlined in a concise manner to facilitate more discussion in this area.Originality/valueExisting reports and research articles tend to focus on the western markets. This article provides a new perspective on tokenization, specifically in the APAC context.

Non-autoregressive machine translation aims to speed up the decoding procedure by discarding the autoregressive model and generating the target words independently. Because non-autoregressive machine translation fails to exploit target-side information, the ability to accurately model source representations is critical. In this paper, we propose an approach to enhance the encoder’s modeling ability by using a pre-trained BERT model as an extra encoder. With a different tokenization method, the BERT encoder and the Raw encoder can model the source input from different aspects. Furthermore, having a gate mechanism, the decoder can dynamically determine which representations contribute to the decoding process. Experimental results on three translation tasks show that our method can significantly improve the performance of non-autoregressive MT, and surpass the baseline non-autoregressive models. On the WMT14 EN $$\rightarrow$$ DE translation task, our method achieves 27.87 BLEU with a single decoding step. This is a comparable result with the baseline autoregressive Transformer model which obtains a score of 27.8 BLEU.

A vast amount of data is generated every second for microblogs, content sharing via social media sites, and social networking. Twitter is an essential popular microblog where people voice their opinions about daily issues. Recently, analyzing these opinions is the primary concern of Sentiment analysis or opinion mining. Efficiently capturing, gathering, and analyzing sentiments have been challenging for researchers. To deal with these challenges, in this research work, we propose a highly accurate approach for SA of fake news on COVID-19. The fake news dataset contains fake news on COVID-19; we started by data preprocessing (replace the missing value, noise removal, tokenization, and stemming). We applied a semantic model with term frequency and inverse document frequency weighting for data representation. In the measuring and evaluation step, we applied eight machine-learning algorithms such as Naive Bayesian, Adaboost, K -nearest neighbors, random forest, logistic regression, decision tree, neural networks, and support vector machine and four deep learning CNN, LSTM, RNN, and GRU. Afterward, based on the results, we boiled a highly efficient prediction model with python, and we trained and evaluated the classification model according to the performance measures (confusion matrix, classification rate, true positives rate...), then tested the model on a set of unclassified fake news on COVID-19, to predict the sentiment class of each fake news on COVID-19. Obtained results demonstrate a high accuracy compared to the other models. Finally, a set of recommendations is provided with future directions for this research to help researchers select an efficient sentiment analysis model on Twitter data.

Email is quite popular as a digital communication media. This is because the message sending process via email is easy. Unfortunately, most messages in emails are spam emails. Spam is a message that the recipient of the message does not want because spam usually contains advertising messages or fraudulent messages. Ham is the message that the recipient wants. One way to sort these messages is to classify email messages into spam or Ham. Naïve Bayes and decision tree J48 are the algorithms that can be used to classify email messages. Therefore, this study aims to compare the effectiveness of the Naïve Bayes algorithm and decision tree J48 in sorting spam emails. The method used is text mining. Data containing the text of the email message in English will be processed before being classified with Naïve Bayes and decision tree J48. The pre-process stage includes tokenization, disposal of stop word lists, stemming, and attribute selection. Furthermore, Data text for email message will be processed using the Naïve Bayes algorithm and decision tree J48. The Naïve Bayes algorithm is a classification algorithm based on Bayesian Decision Theory, while the J48 decision tree algorithm is the development of the ID3 decision tree algorithm. The result of this research is that the decision tree J48 algorithm gets higher accuracy than the Naïve Bayes algorithm. The decision tree J48 algorithm has an accuracy of 93,117% while Naïve Bayes has an accuracy of 88,5284%. The conclusion of this study is that the decision tree J48 algorithm is superior to Naive Bayes for sorting spam emails when viewed from the level of accuracy of each algorithm., Email cukup populer sebagai salah satu media komunikasi digital. Hal tersebut dikarenakan proses pengiriman pesan dengan email yang mudah. Sayangnya, kebanyakan pesan dalam email adalah email spam. Spam adalah pesan yang tidak diinginkan penerima pesan karena spam biasanya berisi pesan iklan maupun pesan penipuan. Ham adalah pesan yang diinginkan penerima pesan. Salah satu cara untuk menyortir pesan-pesan tersebut adalah dengan melakukan pengklasifikasian pesan email menjadi spam maupun ham. Naïve Bayes dan decision tree J48 ialah algoritma yang dapat digunakan untuk mengklasifikasikan pesan email. Oleh karena itu, penelitian ini bertujuan membandingkan efektifitas algoritma Naïve Bayes dan decision tree J48 dalam penyortiran email spam. Metode yang digunakan adalah text mining. Data yang berisi teks pesan email berbahasa Inggris akan diproses terlebih dahulu sebelum diklasifikasikan dengan Naïve Bayes dan decision tree J48. Tahap pra proses tersebut meliputi tokenisasi, pembuangan stop word list, stemming, dan seleksi atribut. Selanjutnya, data teks pesan email akan diproses dengan algoritma Naïve Bayes dan decision tree J48. Algoritma Naïve Bayes adalah algoritma pengklasifikasi yang berdasarkan pada teori keputusan Bayesian sedangkan algoritma decision tree J48 ialah pengembangan dari algoritma decision tree ID3. Hasil penelitian ini adalah algoritma decision tree J48 mendapat akurasi yang lebih tingggi dari algoritma Naïve Bayes. Algoritma decision tree J48 mendapat 93,117% sedangkan Naïve Beyes memiliki akurasi 88,5284%. Kesimpulan dari penelitian ini adalah algoritma decision tree J48 lebih unggul dibanding Naive Bayes untuk menyortir email spam jika dilihat dari tingkat akurasi masing-masing algoritma.

Memodelkan Entity Relationship Diagram (ERD) dapat dilakukan secara manual, namun umumnya memperoleh pemodelan ERD secara manual membutuhkan waktu yang lama. Maka, dibutuhkan pembangkit ERD dari spesifikasi kebutuhan untuk mempermudah dalam melakukan pemodelan ERD. Penelitian ini bertujuan untuk mengembangkan sebuah sistem pembangkit ERD dari spesifikasi kebutuhan dalam Bahasa Indonesia dengan menerapkan beberapa tahapan-tahapan dari Natural Language Processing (NLP) sesuai kebutuhan penelitian. Spesifikasi kebutuhan yang digunakan tim peneliti menggunakan teknik document analysis. Untuk tahapan-tahapan dari NLP yang digunakan oleh peneliti yaitu: case folding, sentence segmentation, tokenization, POS tagging, chunking dan parsing. Kemudian peneliti melakukan identifikasi terhadap kata-kata dari teks yang sudah diproses pada tahapan-tahapan dari NLP dengan metode rule-based untuk menemukan daftar kata-kata yang memenuhi dalam komponen ERD seperti: entitas, atribut, primary key dan relasi. ERD kemudian digambarkan menggunakan Graphviz berdasarkan komponen ERD yang telah diperoleh Evaluasi hasil ERD yang berhasil dibangkitkan kemudian di evaluasi menggunakan metode evaluasi expert judgement. Dari hasil evaluasi berdasarkan beberapa studi kasus diperoleh hasil rata-rata precision, recall, F1 score berturut-turut dari tiap ahli yaitu: pada ahli 1 diperoleh 91%, 90%, 90%; pada ahli 2 diperoleh 90%, 90%, 90%; pada ahli 3 diperoleh 98%, 94%, 96%; pada ahli 4 diperoleh 93%, 93%, 93%; dan pada ahli 5 diperoleh 98%, 83%, 90%.

PurposeTwitter is gaining popularity as a microblogging and social networking service to discuss various social issues. Coronavirus disease 2019 (COVID-19) has become a global pandemic and is discussed worldwide. Social media is an instant platform to deliberate various dimensions of COVID-19. The purpose of the study is to explore and analyze the public sentiments related to COVID-19 vaccines across the Twitter messages (positive, neutral, and negative) and the impact tweets make across digital social circles.Design/methodology/approachTo fetch the vaccine-related posts, a manual examination of randomly selected 500 tweets was carried out to identify the popular hashtags relevant to the vaccine conversation. It was found that the hashtags “covid19vaccine” and “coronavirusvaccine” were the two popular hashtags used to discuss the communications related to COVID-19 vaccines. 23,575 global tweets available in public domain were retrieved through “Twitter Application Programming Interface” (API), using “Orange Software”, an open-source machine learning, data visualization and data mining toolkit. The study was confined to the tweets posted in English language only. The default data cleaning and preprocessing techniques available in the “Orange Software” were applied to the dataset, which include “transformation”, “tokenization” and “filtering”. The “Valence Aware Dictionary for sEntiment Reasoning” (VADER) tool was used for classification of tweets to determine the tweet sentiments (positive, neutral and negative) as well as the degree of sentiments (compound score also known as sentiment score). To assess the influence/impact of tweets account wise (verified and unverified) and sentiment wise (positive, neutral, and negative), the retweets and likes, which offer a sort of reward or acknowledgment of tweets, were used.FindingsA gradual decline in the number of tweets over the time is observed. Majority (11,205; 47.52%) of tweets express positive sentiments, followed by neutral (7,948; 33.71%) and negative sentiments (4,422; 18.75%), respectively. The study also signifies a substantial difference between the impact of tweets tweeted by verified and unverified users. The tweets related to verified users have a higher impact both in terms of retweets (65.91%) and likes (84.62%) compared to the tweets tweeted by unverified users. Tweets expressing positive sentiments have the highest impact both in terms of likes (mean = 10.48) and retweets (mean = 3.07) compared to those that express neutral or negative sentiments.Research limitations/implicationsThe main limitation of the study is that the sentiments of the people expressed over one single social platform, that is, Twitter have been studied which cannot generalize the global public perceptions. There can be a variation in the results when the datasets from other social media platforms will be studied.Practical implicationsThe study will help to know the people's sentiments and beliefs toward the COVID-19 vaccines. Sentiments that people hold about the COVID-19 vaccines are studied, which will help health policymakers understand the polarity (positive, negative, and neutral) of the tweets and thus see the public reaction and reflect the types of information people are exposed to about vaccines. The study can aid the health sectors to intensify positive messages and eliminate negative messages for an enhanced vaccination uptake. The research can also help design more operative vaccine-advocating communication by customizing messages using the obtained knowledge from the sentiments and opinions about the vaccines.Originality/valueThe paper focuses on an essential aspect of COVID-19 vaccines and how people express themselves (positively, neutrally and negatively) on Twitter.

Quora is a common online platform for users to obtain answers to their questions, which is often subjected to the posting of irrelevant questions and answers by the users. This needs to be filtered and only relevant contents are to be allowed. The present work emphasized on filtering of community-oriented irrelevant questions by machine learning and classifying the questions as genuine/irrelevant. Preprocessing techniques like tokenization, lemmatization and stop words removal were used to convert the input data into a structured data, which was then fed to Word2Vec embedding model which mapped every unique word to a corresponding vector in the space. This vector model was given as an input to long short-term memory (LSTM) network which used ReLU and Adam for optimization. The accuracy of the model was verified at the end of every epoch, and the training was halted with the reduction of accuracy. An accuracy of 97% was obtained at the end of the final epoch. Such an approach was able to successfully filter the community-oriented irrelevant questions. This could be further extended for other anti-social aspects, in the near future.

This paper presents a novel blockchain-based spectrum tokenization method used to crowdsource wireless network deployment projects. Crowdsourcing is a method of financing certain projects and ideas through the funds collected by individuals or businesses in an open marketplace. The method presented in this paper finances the wireless network deployment projects belonging to service providers or governments. The method tokenizes proposed novel wireless resource units, and sells these units to investors. A new Value Unit Per User (VUPU) resource unit is introduced with a new pricing scheme depending on a load of a base station. A novel Proof of Data Load (PoDLO) consensus algorithm is proposed which is used to verify data and traffic load of a base station. Device Diversity Factor (DDF) and Subscriber Unique Permanent Identifier (SUPI) Factor (SUF) are proposed new ways to determine the value of a base station and a network cluster.

In software development, source code documents play essential role during program comprehension and software maintenance. Natural language descriptions and identifier names are the main parts of the source code document. Source code document generation spares the working hours of developers. Automatic source code documentation is a rapidly growing research area at the present time. Researchers have proposed various template based, IR based (information retrieval), and learning-based techniques for automatic source code documentation. There is not much work related to preprocessing and its effect on the automatic source code documentation. Tokenization is one of the essential steps in preprocessing. We found some important flaws in the basic tokenization steps that could affect automatic source code documentation performance. Therefore, we propose an updated tokenization approach to remove the flaws of basic tokenization steps. We performed method name prediction and comment generation studies to analyze the effect of updated tokenization approach. We found that the updated tokenization helped in improving the performance of the automatic source code documentation. Name prediction and comment generation performance improved by more than 2.5% and 3.5%, respectively, in terms of F1 score.

In this paper, it has been proposed an approach to assess the performance of part-of-speech tagging of the Bengali text. The tagging can be viewed as a process of interpreting a syntactic category for tokens in text documents. The difficulty occurs when choosing an appropriate category of the speech part for tokens. To overcome this limitation, we proposed an effective method to carry out part of the speech tagging on 5 corpora independent of the domain. Subsequently, we performed a tagging performance assessment to verify the efficiency of our system. Our system is developed in 10 phases: initialization of the dataset, sentence boundary determination, tokenization, identification of unique tokens, part-of-speech tagging, retrieving the token and the tagged portion of the speech class, record the retrieved outcomes, query processing, performance evaluation and rank generation. Five corpora have been used for the experiment of the system. The system is successfully tagged 98.97%, 98.35%, 89.93%, 88.46% and 90.01% tokens of experimental corpora. The system has obtained excellent tagging performance for the POS category (Common Noun) compared to other POS categories. The efficiency of this system is visualized through detailed performance appraisal in 18 part-of-speech categories. The system is successfully tagged 16,504,118 tokens over 18,047,593 numbers of distinct tokens in the total corpora. It has been achieved 91.44% overall tagging effectiveness, which represents an improvement of about 3.24% compared to the baseline method.

In evaluating internet circles, the dataset in textual form turns out to be an important tool for communication. Recognizing user’s emotion-based textual data is a difficult task. The original big textual data contains lots of noisy information, which may lead to degrading the system performance. So, in this research, novel hybrid based pre-processing approaches are employed to overcome these issues. Initially, a big text dataset is fed into the pre-processing stage, which carries a lot of hybrid approaches for eliminating noisy contents. Some steps are followed to carry out pre-processing steps, which are lower case conversion, HTML/XML tag removal, stop word removal, Tokenization, Text normalization, and numerical conversion. Afterwards, the string information is transformed into numerical values for extracting the feature information. The pre-processing steps make the final classifier step easier and achieve a good prediction measure with the minimum error value. The proposed methodology is validated with three conventional classifier algorithms, namely PNN, ANN and k‑NN. At last, the performance metrics are evaluated to show the effectiveness of the proposed methodology.

Depression and social anxiety are the two main negative impacts of cyberbullying. Unfortunately, a survey conducted by UNICEF on 3rd September 2019 showed that 1 in 3 young people in 30 countries had been victims of cyberbullying. Sentiment analysis research will be conducted to detect a comment that contains cyberbullying. Dataset of cyberbullying is obtained from the Kaggle website, named, Toxic Comment Classification Challenge. The pre-processing process consists of 4 stages, namely comment generalization (convert text into lowercase and remove punctuation), tokenization, stop words removal, and lemmatization. Word Embedding will be used to conduct sentiment analysis by implementing Word2Vec. After that, One-Against-All (OAA) method with the Support Vector Machine (SVM) model will be used to make predictions in the form of multi labelling. The SVM model will go through a hyperparameter tuning process using Randomized Search CV. Then, evaluation will be carried out using Micro Averaged F1 Score to assess the prediction accuracy and Hamming Loss to assess the numbers of pairs of sample and label that are incorrectly classified. Implementation result of Word2Vec and OAA SVM model provide the best result for the data undergoing the process of pre-processing using comment generalization, tokenization, stop words removal, and lemmatization which is stored into 100 features in Word2Vec model. Micro Averaged F1 and Hamming Loss percentage that is produced by the tuned model is 83.40% and 15.13% respectively. Index Terms— Sentiment Analysis; Word Embedding; Word2Vec; One-Against-All; Support Vector Machine; Toxic Comment Classification Challenge; Multi Labelling

Since the emergence of deep learning-based chatbots for knowledge services, numerous research and development projects have been conducted in various industries. A high demand for chatbots has drastically increased the global market size; however, the limited functional scalability of open-domain chatbots is a challenge to their application to industries. Moreover, as most chatbot frameworks employ English, it is necessary to create chatbots customized for other languages. To address this problem, this paper proposes KoRASA as a pipeline-optimization method, which uses a deep learning-based open-source chatbot framework to understand the Korean language. KoRASA is a closed-domain chatbot that is applicable across a wide range of industries in Korea. KoRASA’s operation consists of four stages: tokenization, featurization, intent classification, and entity extraction. The accuracy and F1-score of KoRASA were measured based on datasets taken from common tasks carried out in most industrial fields. The algorithm for intent classification and entity extraction was optimized. The accuracy and F1-score were 98.2% and 98.4% for intent classification and 97.4% and 94.7% for entity extraction, respectively. Furthermore, these results are better than those achieved by existing models. Accordingly, KoRASA can be applied to various industries, including mobile services based on closed-domain chatbots using Korean, robotic process automation (RPA), edge computing, and Internet of Energy (IoE) services.

This article examines the mechanisms now available to tokenize real asset ownership and create active secondary markets in tokenized or fractionalized units. Real estate is lumpy and illiquid, and tokenization would allow participation in the market by a broader group of investors and the ability to build diversified portfolios with modest sums of money. However, two developments are necessary for the digital tokenization of single real estate assets: (1) an expressed demand for the fractionalization of single real estate assets, and (2) market participants need to be comfortable with blockchain, the digital underpinning of tokenization. Further, in many land markets fractionalization requires an intermediate structure to be established because the direct ownership of land cannot be shared amongst many, increasing the cost of tokenization. Larger assets already held in fund structures may eventually be tokenized successfully; there may also be an alternative market for tokenized residential, social impact, or community assets where investment regulation and risk/return are not the main drivers of behavior. The mass market for the tokenization of single commercial real estate assets, however, may be some way down the road. TOPICS:Real estate, exchanges/markets/clearinghouses, currency, performance measurement Key Findings ▪ The author examines the mechanisms now available to tokenize real asset ownership. ▪ The author presents new survey-based evidence of real estate tokenization deals. ▪ The author concludes that an intermediate structure is likely to be both necessary and convenient when fractionalizing a single asset. For that reason, funds are more natural targets for tokenization.

One of the solutions to the problem of spatio-temporal data anisotropy is the use of a multilevel system of digital twins based on the corresponding industry models and the updated archive data base. The application of this approach has successfully proved itself in information systems for monitoring the parameters of the geomagnetic field and its variations, providing spatio-temporal interpolation of geomagnetic data with an accuracy of 0.81 nT in magnetically quiet periods. At the same time, the problem of information interaction between the levels of the system of digital twins remained unresolved, which is greatly aggravated by the constantly growing volume of data and their heterogeneous nature. The paper proposes a solution to the indicated problem by means of a formalized mechanism for packaging space-time information, in which the identification of data sources is performed on the basis of a hierarchical binary tokenization system. In addition, the proposed software implementation of such an approach is considered, a distinctive feature of which is the combination of traditional clientserver and innovative serverless architectures to implement a highly loaded reactive web application for working with analyzed data. The main stages of the implementation of information interaction are highlighted and programmatically formalized - from obtaining initial information from its sources to verifying data, analyzing them, processing and forming the output information flow of the system. The results of the computational experiments carried out on the example of the problem of monitoring the parameters of the Earth's magnetic field and its variations confirmed the effectiveness of the proposed solutions, expressed both in increasing the reactivity of web-based applications and in increasing the computational speed of formation and filling of information storages that aggregate information from distributed heterogeneous sources.

Machine Translation is an effort to bridge language barriers and misinterpretations, making communication more convenient through the automatic translation of languages. The quality of translations produced by corpus-based approaches predominantly depends on the availability of a large parallel corpus. Although machine translation of many Indian languages has progressively gained attention, there is very limited research on machine translation and the challenges of using various machine translation techniques for a low-resource language such as Mizo. In this article, we have implemented and compared statistical-based approaches with modern neural-based approaches for the English–Mizo language pair. We have experimented with different tokenization methods, architectures, and configurations. The performance of translations predicted by the trained models has been evaluated using automatic and human evaluation measures. Furthermore, we have analyzed the prediction errors of the models and the quality of predictions based on variations in sentence length and compared the model performance with the existing baselines.

Healthcare service has the role to help and serve people to access medical services, i.e. providing medicines, medical consultation, or health control. Healthcare service has been transforming to a digital platform. Halodoc is one of the digital platforms that people can use for free or paid, user can also give reviews of Halodoc’s performance and services on Google Play Store to give feedback that Halodoc can use to evaluate and improve the app. The Google Play Store review is increasing every day. Therefore an analysis for the review with sentiment analysis for Halodoc’s review is needed, first phase of sentiment analysis for the review is preprocessing which has tokenization, transform to lower cases, filter stopword, dan filter token (by length) processes. The data is divided into two positive and negative classes with cross-validation and a k-fold validation value of 10, using Naïve Bayes Classifier algorithm with 81,68% accuracy and AUC 0.756, categorized as fair classification.

International audience; phase in the Software Development Life Cycle (SDLC). The design phase follows it. Traceability is one of the core concepts in software engineering; it is used to follow updates to make consistent items. This paper aimed to cover consistency through bi-directional traceability between requirements and design phase in a semi-automatic way. The Natural Language Processing (NLP) was used to analyze the requirements text and generate a class diagram; then, the generated items can be traced back to requirements. We developed a novel process to support consistency and bi-directional traceability. To ensure our proposed process's practical applicability, we implemented a tool named as Requirements and Design Bi-directional Traceability (RDBT). RDTB receives textual format requirements, performs NLP tasks (Tokenization, Part-of-Speech Tagging, etc.), generates UML class diagram, and finally performs traceability management to ensure consistency of requirements and UML class diagram. The work evaluation reveals good results, which indicates it can be used efficiently as a guide to generate the UML class diagram semi-automatically and manage traceability.

Mobile advertising plays a vital role in the mobile app ecosystem. A major threat to the sustainability of this ecosystem is click fraud , i.e., ad clicks performed by malicious code or automatic bot problems. Existing click fraud detection approaches focus on analyzing the ad requests at the server side. However, such approaches may suffer from high false negatives since the detection can be easily circumvented, e.g., when the clicks are behind proxies or globally distributed. In this paper, we present AdSherlock, an efficient and deployable click fraud detection approach at the client side (inside the application) for mobile apps. AdSherlock splits the computation-intensive operations of click request identification into an offline procedure and an online procedure. In the offline procedure, AdSherlock generates both exact patterns and probabilistic patterns based on URL (Uniform Resource Locator) tokenization. These patterns are used in the online procedure for click request identification and further used for click fraud detection together with an ad request tree model. We implement a prototype of AdSherlock and evaluate its performance using real apps. The online detector is injected into the app executable archive through binary instrumentation. Results show that AdSherlock achieves higher click fraud detection accuracy compared with state of the art, with negligible runtime overhead.

The Sentiment Analysis is used for the text analysing, and classification of the text attitude.We are using the computing advancement in the form of Machine Learning (ML) and Support Vector Machine (SVM) algorithm to train a dataset which is collected automatically through ArabiTools and Twitter API. The dataset contents are labelled by both means, automatic and manual, in order to maintain the efficiency of the detection of CyberBullying tweets. The dataset is automatically labelled with respect to the nature of the tweet. If a tweet contains one or more CyberBullying words, it is labelled as CyberBullying, while if there isn’t any word with aggressive meaning found, it is marked as the NonCyberBullying. After the data collection, there are several pre-processing techniques utilized, including the Normalization, Tokenization, Light Stemmer, ArabicStemmerKhoja, and Term Frequency-Inverse Document Frequency (TF-IDF)” term weighting schema.” After the preliminary steps, (SVM), a “supervised algorithm,” is used with WEKA and Python. There are three experiments that take place one with the WEKA tool using the Light Stemmer, the other is again with WEKA using ArabicStemmerKhoja, and the final experiment was performed with Python. The results are showing the WEKA is more efficient in classifying the text correctly, while Python is more effective with time to build the model. WEKA using the Light Stemmer have the efficiency of 85.49% and taken 352.51 seconds, and the WEKA using ArabicStemmerKhoja have the efficiency of 85.38% and taken 212.12 seconds, while the Python have the efficiency of 84.03% and taken 142.68 seconds

Emotions detection in natural languages is very effective in analyzing the user's mood about a concerned product, news, topic, and so on. However, it is really a challenging task to extract important features from a burst of raw social text, as emotions are subjective with limited fuzzy boundaries. These subjective features can be conveyed in various perceptions and terminologies. In this article, we proposed an IoT-based framework for emotions classification of tweets using a hybrid approach of Term Frequency Inverse Document Frequency (TFIDF) and deep learning model. First, the raw tweets are filtered using the tokenization method for capturing useful features without noisy information. Second, the TFIDF statistical technique is applied to estimate the importance of features locally as well as globally. Third, the Adaptive Synthetic (ADASYN) class balancing technique is applied to solve the imbalance class issue among different classes of emotions. Finally, a deep learning model is designed to predict the emotions with dynamic epoch curves. The proposed methodology is analyzed on two different Twitter emotions datasets. The dynamic epoch curves are shown to show the behavior of test and train data points. It is proved that this methodology outperformed the popular state-of-the-art methods.

Initial coin offerings (ICOs) are an emerging form of fundraising for blockchain-based startups. We examine how ICOs can be leveraged in the context of asset tokenization, whereby firms issue tokens backed by future assets (i.e., inventory) to finance growth. We (i) make suggestions on how to design such “asset-backed” ICOs—including optimal token floating and pricing for both utility and equity tokens (a.k.a. security token offerings)—taking into account moral hazard (cash diversion), product characteristics, and customer demand uncertainty; (ii) make predictions on ICO success/failure; and (iii) discuss implications on firm operating strategy. We show that in unregulated environments, ICOs can lead to significant agency costs, underproduction, and loss of firm value. These inefficiencies, however, fade as product margins and demand characteristics (mean/variance) improve, and they are less severe under equity (rather than utility) token issuance. Importantly, the advantage of equity tokens stems from their inherent ability to better align incentives and thus continues to hold even absent regulation. This paper was accepted by Vishal Gaur, operations management.

PurposeThe purpose of the paper is to study multiple viewpoints which are required to access the more informative similarity features among the tweets documents, which is useful for achieving the robust tweets data clustering results.Design/methodology/approachLet “N” be the number of tweets documents for the topics extraction. Unwanted texts, punctuations and other symbols are removed, tokenization and stemming operations are performed in the initial tweets pre-processing step. Bag-of-features are determined for the tweets; later tweets are modelled with the obtained bag-of-features during the process of topics extraction. Approximation of topics features are extracted for every tweet document. These set of topics features of N documents are treated as multi-viewpoints. The key idea of the proposed work is to use multi-viewpoints in the similarity features computation. The following figure illustrates multi-viewpoints based cosine similarity computation of the five tweets documents (here N = 5) and corresponding documents are defined in projected space with five viewpoints, say, v1,v2, v3, v4, and v5. For example, similarity features between two documents (viewpoints v1, and v2) are computed concerning the other three multi-viewpoints (v3, v4, and v5), unlike a single viewpoint in traditional cosine metric.FindingsHealthcare problems with tweets data. Topic models play a crucial role in the classification of health-related tweets with finding topics (or health clusters) instead of finding term frequency and inverse document frequency (TF–IDF) for unlabelled tweets.Originality/valueTopic models play a crucial role in the classification of health-related tweets with finding topics (or health clusters) instead of finding TF-IDF for unlabelled tweets.

Information retrieval system is a system that is widely used to retrieve information. This research will discuss how the system finds back the information stored in database tables. Tables in the database are arranged to store all forms of data entered by database users so that later the stored data can be used again. Re-accessing the database's information must go through a mechanism known as a database management system (DBMS). One of the most widely used DBMS is MySQL. By using a DBMS, information and data can be manipulated according to user needs. Data manipulation in the database is done in a special language, namely SQL (Structure Query Language). Mastery of SQL commands is an obligation for database users so that the manipulated data can produce the required information. However, many database users still do not understand how the actual SQL command syntax manages and manipulates data into information. This is, of course, very risky if the solution is not immediately sought because it will hinder the process of retrieving information from the data stored in the database. For this problem to be resolved, it is necessary to design a system that can help database users translate their wishes into SQL command syntax. This paper will discuss how a command in Indonesian can be translated into SQL command syntax. The method used to solve this research problem is rule-based. There are two stages in the main process: the pre-processing stage, which consists of a word tokenization process, and a translation stage, including a keyword grouping process. This keyword grouping process consists of the keyword group analysis phase, table and column analysis, identification of SQL commands, and mapping of SQL commands. From all stages that have been passed and testing of 7 scenarios with ten (10) commands for each scenario, the accuracy is 81.42%. The inaccuracy in the testing process is more a problem of displaying data from two or more tables, for example, using the join table command. This problem can be addressed by adding new rules for the use of table joins.

PurposeBlockchain technology is increasingly positioned as a promising and disruptive technology. Such a promise has attracted companies to explore how blockchain technology can be used to gain significant benefits. Process models play a cardinal role when seeking to improve business processes as they are the foundation of process analysis and redesign. This paper examines how blockchain-oriented processes can be conceptually modelled with activity- (BPMN) and artifact-centric (CMMN) modelling paradigms.Design/methodology/approachThis paper discusses how commonly occurring patterns, specific to block-chain-based applications, can be modelled with BPMN and CMMN. Furthermore, the advantages and disadvantages of both notations for accurately representing blockchain-specific patterns are discussed.FindingsThe main finding of this paper is that neither BPMN nor CMMN can adequately and accurately represent certain patterns specific for blockchain-oriented processes. BPMN, while supporting most of the patterns, does not provide sufficient support to represent tokenization. CMMN, on the other hand, does not provide support to distinguish between activities executed and data stored on-chain versus off-chain.Originality/valueThe paper provides insight into the strengths and weaknesses of BPMN and CMMN for modelling processes to be supported by blockchain. This will serve to aid analysts to produce better process models for communication purposes and, thereby, facilitate development of blockchain-based solutions.

In recent years, the use of deep learning in language models has gained much attention. Some research projects claim that they can generate text that can be interpreted as human writing, enabling new possibilities in many application areas. Among the different areas related to language processing, one of the most notable in applying this type of modeling is programming languages. For years, the machine learning community has been researching this software engineering area, pursuing goals like applying different approaches to auto-complete, generate, fix, or evaluate code programmed by humans. Considering the increasing popularity of the deep learning-enabled language models approach, we found a lack of empirical papers that compare different deep learning architectures to create and use language models based on programming code. This paper compares different neural network architectures like Average Stochastic Gradient Descent (ASGD) Weight-Dropped LSTMs (AWD-LSTMs), AWD-Quasi-Recurrent Neural Networks (QRNNs), and Transformer while using transfer learning and different forms of tokenization to see how they behave in building language models using a Python dataset for code generation and filling mask tasks. Considering the results, we discuss each approach&rsquo, s different strengths and weaknesses and what gaps we found to evaluate the language models or to apply them in a real programming context.

Machine translation is one of the most popular and hardest tasks in Natural Language Processing. This paper proposes a self-attention based model for machine translation, named Re-Transformer, by transforming the Transformer [1]. Different from prevailing approach through module or GPU stacking to improve the system performance, Re-Transformer modifies the basic architecture; there are four refinements as follows. First, Re-Transformer adopts sub-word Tokenization in corpus preprocessing to overcome rare words. In the encoder layer, dual Self-Attention stacks and less Point Wise Feed Forward layer are adopted to obtain better comprehension of an input sentence. In decoder, reduced the stack of “Decoder” are used to speedup training and inferring speed meanwhile keeps the same level of BLEU. The experiment results show that Re-Transformer with BLEU metric score 31.36, 38.45 (four-layer decoder) and 32.14, 55.62 (two-layer decoder) improves around 4 and 17 points of BLEU metric against the Transformer over the WMT 2014 English-German and English-French Translation Corpus.

Previously, rapid disease detection and prevention was difficult. This is because disease modeling and prediction was dependent on a manually obtained dataset that includes use of survey. With the increased use of social media platforms like Twitter, Facebook, Instagram, etc., data mining and sentiment analysis can help avoid diseases. Sentiment analysis is a powerful tool for analyzing people’s perceptions, emotions, value assessments, attitudes, and feelings as expressed in texts. The purpose of this research is to use machine learning techniques to classify and predict the spatial distribution of positive and negative sentiments of Covid-19 pandemic. This study research has employed machine learning to classify spatial distribution of Covid-19 twitter sentiments as positive or negative. The data for this study were geo-tagged tweets concerning COVID-19 which were live streamed using streamR package. The key terms used for streaming the data were: Corona, Covid-19, sanitizer, virus, lockdown, quarantine, and social distance. The classification used Naive Bayes algorithms with ngram approaches. N-Gram model is a probabilistic language model used to predict next item in a sequence in the form (n - 1) order Markov. It relies on the Markov assumption—the probability of a word depends only on the previous word without looking too far into the past. The steps followed in this research include: cleaning and preprocessing the data, text tokenization using n-gram i.e. 1-gram, 2-gram, and 3-gram, tweets were converted or weighted into a matrix of numeric vectors using Term Frequency Inverse-Document. Also, data were divided 80:20 between train and test data. A confusion matrix was utilized to evaluate the classification accuracy, precision, and recall performance of the various algorithms tested. Prediction was done using the best performing Naive Bayes algorithm. The results of this research showed that under Multinomial Naive Bayes, unigram accuracy was 92.02%, bigram accuracy was 97.37%, and trigram accuracy was 94.40%. Unigram had 89.34% accuracy, bigram had 96.80%, and trigram had 94.90% accuracy using Bernoulli Naive Bayes. Unigram accuracy was 90.43%, bigram accuracy was 95.67%, and trigram accuracy was 92.89% using Gaussian Naive Bayes. Bigram tokenization outperformed unigram and trigram tokenization. Bigram Multinomial Naive Bayes was used to predict test data since it was the most accurate in classifying train data. Prediction accuracy was 84.92%, precision 85.50%, recall 81.02%, and F1 measure 83.20%. TF-IDF was employed to increase prediction accuracy, obtaining 87.06%. These were then plotted on a globe map. The study indicates that machine learning can identify patterns and emotions in public tweets, which may then be used to steer targeted intervention programs aimed at limiting disease spread.

Many machine learning algorithms and almost all deep learning architectures are incapable of processing plain texts in their raw form. This means that their input to the algorithms must be numerical in order to solve classification or regression problems. Hence, it is necessary to encode these categorical variables into numerical values using encoding techniques. Categorical features are common and often of high cardinality. One-hot encoding in such circumstances leads to very high dimensional vector representations, raising memory and computability concerns for machine learning models. This paper proposes a deep-learned embedding technique for categorical features encoding on categorical datasets. Our technique is a distributed representation for categorical features where each category is mapped to a distinct vector, and the properties of the vector are learned while training a neural network. First, we create a data vocabulary that includes only categorical data, and then we use word tokenization to make each categorical data a single word. After that, feature learning is introduced to map all of the categorical data from the vocabulary to word vectors. Three different datasets provided by the University of California Irvine (UCI) are used for training. The experimental results show that the proposed deep-learned embedding technique for categorical data provides a higher F1 score of 89% than 71% of one-hot encoding, in the case of the Long short-term memory (LSTM) model. Moreover, the deep-learned embedding technique uses less memory and generates fewer features than one-hot encoding.

Abstract—The number of hotel reviews are huge and growing day by day. Many travelers rely on this review to take down their decision to book a hotel. Gather valid and useful information from a huge amount of reviews are needs that must be met. Hence, a summarize tool for hotel review is built to create a representative summary. Hotel review data growing a trend of sentiment in a range of time due to the condition and improvement at that time, so the analysis of sentiment trend is done to choose the appropriate representative review data. Then, there are two method to summarize the selected review data. First, summary was obtained from extractive method with selecting most related sentence base on its Term Frequency-Inverse Document Frequency (TF-IDF) score. Second, phrase summary style is built by pairing adjective to the nearest noun and considering the polarity. From those method, obtained Recall-Oriented Understudy for Gisting Evaluation (ROUGE)-1 recall and Bilingual Evaluation Understudy (BLEU) score respectively {0.2101and 0.7820} for first method and {0.0670 and 0.03672} for the second method. All the reviews are crawled from TripAdvisor website and have been pre-processed by segmenting/tokenization, case folding, and tagging.

This paper attempts to analyze the Korean sentence classification system. Sentence classification is the task of classifying an input sentence based on predefined categories. However, spelling or space error contained in the input sentence causes problems in morphological analysis and tokenization. This paper proposes a novel approach of Integrated Eojeol (Korean syntactic word separated by space) Embedding to reduce the effect of poorly analyzed morphemes on sentence classification. The paper also proposes two noise insertion methods that further improve classification performance. Our evaluation results indicate that by applying the proposed methods on the existing sentence classifiers, the sentence classification accuracy on erroneous sentences is increased by 8% to 15%.

Automatic grading requires the adaption of the latest technologies. It has become essential especially when most of the courses became online courses (MOOCs). The objectives of the current work are (1) Reviewing the literature on the text semantic similarity and automatic exam correction systems, (2) Proposing an automatic exam correction framework (HMB-AECF) for MCQs, essays, and equations that is abstracted into five layers, (3) Suggesting equations similarity checker algorithm named “HMB-MMS-EMA”, (4) Presenting an expression matching dataset named “HMB-EMD-v1”, (5) Comparing the different approaches to convert textual data into numerical data (Word2Vec, FastText, Glove, and Universal Sentence Encoder (USE)) using three well-known Python packages (Gensim, SpaCy, and NLTK), and (6) Comparing the proposed equations similarity checker algorithm (HMB-MMS-EMA) with a Python package (SymPy) on the proposed dataset (HMB-EMD-v1). Eight experiments were performed on the Quora Questions Pairs and the UNT Computer Science Short Answer datasets. The best-achieved highest accuracy in the first four experiments was 77.95% without fine-tuning the pre-trained models by the USE. The best-achieved lowest root mean square error (RMSE) in the second four experiments was 1.09 without fine-tuning the used pre-trained models by the USE. The proposed equations similarity checker algorithm (HMB-MMS-EMA) reported 100% accuracy over the SymPy Python package which reported 71.33% only on “HMB-EMD-v1”.

Blockchain teknolojisi, Bitcoin veya diğer kripto para birimlerinden ayrı olarak muhasebe ve denetim alanında geniş bir uygulanabilirliğe sahiptir. Ancak blokzincir teknolojisinin alanda hızlı bir şekilde yer edinmesine rağmen, günümüzde gerçekçi bir şekilde nasıl uygulanacağı konusunda fikir birliği bulunmamaktadır. Bu çalışmanın amacı, mevcut uluslararası denetim literatürünü haritalamak ve uygulamadaki mevcut karmaşıklığa açıklık getirmektir. Web of Science Core Collection'dan indirilen 112 çalışma Bibliometrix R-Package "Biblioshiny" yardımı ile analiz edilmiştir. Bu çalışmanın sonucunda, konunun uluslararası alanda farklı araştırma dallarına ayrıldığı tespit edilmiştir; Şifreli Özel ve Güvenli Bilgi Paylaşımı, Dijital Defter, Akıllı Sözleşmeler, Sürekli Denetim, Denetim İzi ve Tokenizasyon. Bununla birlikte, denetimde blokzincir teknolojisinin benimsenmesini hızlandırmak için ele alınması gereken araştırma boşlukları tespit edilmiştir., Blockchain technology has a wide scope of applicability for accounting and auditing purposes, separate from Bitcoin or other cryptocurrencies. However, although blockchain technology is identified as a popular topic within the field, there currently exists a lack of consensus on how it will be realistically applied. This paper aims to map the presently existing international auditing literature and offer clarity to an otherwise heterogeneous field. The mapping analysis will aid in examining the underlying theme employed within studies. Through the use of the Bibliometrix R-Package "Biblioshiny", a sample of 112 studies were downloaded from the Web of Science Core Collection and analyzed. This study's results show that the topic has gained increased international popularity and separated into distinct research streams; Encrypted Private and Secure Information Sharing, Distributed Ledger, Smart Contracts, Continuous Audit, Audit Trail and Tokenization. However, there still exists a research gap that must be addressed to hasten the adoption of blockchain technology in auditing.

Data in the Cloud, which applies to data as a cloud service provider (CSP), transmits stores, or manages it. The company will enforce the same definition of data usage while the data is resident within the enterprise and thus extend the required cryptographic security criteria to data collected, exchanged, or handled by CSP. The CSP Service Level Agreements cannot override the cryptographic access measures. When the data is transferred securely to CSP, it can be securely collected, distributed, and interpreted. Data at the rest position applies to data as it is processed internally in organized and in the unstructured ways like databases and file cabinets. The Data at the Rest example includes the use of cryptography for preserving the integrity of valuable data when processed. For cloud services, computing takes multiple forms from recording units, repositories, and many unstructured items. This paper presents a secure model for Data at rest. The TF-Sec model suggested is planned for use with Slicing, Tokenization, and Encryption. The model encrypts the given cloud data using AES 256 encryption, and then the encrypted block is sliced into the chunks of data fragments using HD-Slicer. Then it applies tokenization algorithm TKNZ to each chunk of data, applies erasure coding technique to tokens, applies the data dispersion technique to scramble encrypted data fragments, and allocates to storage nodes of the multiple CSP. In taking the above steps, this study aims to resolve the cloud security problems found and to guarantee the confidentiality of their data to cloud users due to encryption of data fragments would be of little benefit to a CSP.

Sentiment Analysis concerns the analysis of ideas, emotions, evaluations, values, attitudes and feelings about products, services, companies, individuals, tasks, events, titles and their characteristics. With the increase in applications on the Internet and social networks, Sentiment Analysis has become more crucial in the field of text mining research and has since been used to explore users’ opinions on various products or topics discussed on the Internet. Developments in the fields of Natural Language Processing and Computational Linguistics have contributed positively to Sentiment Analysis studies, especially for sentiments written in non-structured or semi-structured languages. In this paper, we present a literature review on the pre-processing task on the field of sentiment analysis and an analytical and comparative study of different researches conducted in Arabic social networks. This study allowed as concluding that several works have dealt with the generation of stop words dictionary. In this context, two approaches are adopted: first, the manual one, which gives rise to a limited list, and second, the automatic, where the list of stop words is extracted from social networks based on defined rules. For stemming two, algorithms have been proposed to isolate prefixes and suffixes from words in dialects. However, few works have been interested in dialects directly without translation. The Moroccan dialect in particular is considered as the 5th dialect studied among Arabic dialects after Jordanian, Egyptian, Tunisian and Algerian dialects. Despite the significant lack in studies carried out on Arabic dialects, we were able to extract several conclusions about the difficulties and challenges encountered through this comparative study, as well as the possible ways and tracks to study in any dialects sentiment analysis pre-processing solution.

Findings of the Association for Computational Linguistics: EMNLP 2020, In traditional NLP, we tokenize a given sentence as a preprocessing, and thus the tokenization is unrelated to a target downstream task. To address this issue, we propose a novel method to explore a tokenization which is appropriate for the downstream task. Our proposed method, optimizing tokenization (OpTok), is trained to assign a high probability to such appropriate tokenization based on the downstream task loss. OpTok can be used for any downstream task which uses a vector representation of a sentence such as text classification. Experimental results demonstrate that OpTok improves the performance of sentiment analysis and textual entailment. In addition, we introduce OpTok into BERT, the state-of-the-art contextualized embeddings and report a positive effect.

Cyberbullying is disturbing and troubling online misconduct. It appears in various forms and is usually in a textual format in most social networks. Intelligent systems are necessary for automated detection of these incidents. Some of the recent experiments have tackled this issue with traditional machine learning models. Most of the models have been applied to one social network at a time. The latest research has seen different models based on deep learning algorithms make an impact on the detection of cyberbullying. These detection mechanisms have resulted in efficient identification of incidences while others have limitations of standard identification versions. This paper performs an empirical analysis to determine the effectiveness and performance of deep learning algorithms in detecting insults in Social Commentary. The following four deep learning models were used for experimental results, namely: Bidirectional Long Short-Term Memory (BLSTM), Gated Recurrent Units (GRU), Long Short-Term Memory (LSTM), and Recurrent Neural Network (RNN). Data pre-processing steps were followed that included text cleaning, tokenization, stemming, Lemmatization, and removal of stop words. After performing data pre-processing, clean textual data is passed to deep learning algorithms for prediction. The results show that the BLSTM model achieved high accuracy and F1-measure scores in comparison to RNN, LSTM, and GRU. Our in-depth results shown which deep learning models can be most effective against cyberbullying when directly compared with others and paves the way for future hybrid technologies that may be employed to combat this serious online issue.

In the derived approach, an analysis is performed on Twitter data for World Cup soccer 2014 held in Brazil to detect the sentiment of the people throughout the world using machine learning techniques. By filtering and analyzing the data using natural language processing techniques, sentiment polarity was calculated based on the emotion words detected in the user tweets. The dataset is normalized to be used by machine learning algorithms and prepared using natural language processing techniques like word tokenization, stemming and lemmatization, part-of-speech (POS) tagger, name entity recognition (NER), and parser to extract emotions for the textual data from each tweet. This approach is implemented using Python programming language and Natural Language Toolkit (NLTK). A derived algorithm extracts emotional words using WordNet with its POS (part-of-speech) for the word in a sentence that has a meaning in the current context, and is assigned sentiment polarity using the SentiWordNet dictionary or using a lexicon-based method. The resultant polarity assigned is further analyzed using na&iuml, ve Bayes, support vector machine (SVM), K-nearest neighbor (KNN), and random forest machine learning algorithms and visualized on the Weka platform. Na&iuml, ve Bayes gives the best accuracy of 88.17% whereas random forest gives the best area under the receiver operating characteristics curve (AUC) of 0.97.

Public opinion is a valuable source for evaluating the performance of transportation services, which can guide management and policy adjustment. In contrast to traditional surveying methods, which are often inefficient and high-cost, social media has become a popular channel to collect public opinion due to its large data volume and accessibility characteristics. However, existing research lack efficient methods to extract and interpret Twitter data for transportation services evaluation. Therefore, this paper presents a comprehensive framework that enables high-efficient extraction and analysis of public opinions on transportation services from Twitter. The transportation system of Miami-Dade County is chosen as the case study to describe the framework development and validation process. First, Twitter data in a defined area over a certain period are collected and preprocessed to clean erroneous and redundant information. Then, the Twitter data that relate to personal opinions on transportation services (POTS) are filtered hieratically using text classification models trained by manually labeled dataset. Next, timeline analysis using Post Intensity (PI) and Average Sentiment Value (ASV) is implemented on the selected Twitter data to explore the public perceptions towards the transportation-related events. Topic modeling and tokenization techniques are also adopted to extract relevant semantic content needed for data analysis. Significantly, the developed framework can improve the efficiency of Twitter data extraction and analysis. Many manual steps are involved in the development process of the framework while can be avoided when generalized to other applications. The framework can be used as a tool for stakeholders to enable a holistic understanding of public opinions on transportation services and increase the degree of public participation in transportation management.

JSON parser, that has been developed, is made of two modules, each dedicated to process structure of either an object or an array. In both modules, building step of document model, that represents its structure, is combined with tokenization step. Design and development of algorithm is aimed at improving speed of parsing of documents.

In modern multilingual societies, there is a demand for multilingual Automatic Speech Recognition (ASR) and Spoken Term Detection (STD). Multilingual Spoken Term Detection refers to the process of retrieving appropriate audio files from a vast multilingual database using audio queries. This paper presents an overview of various efforts on multilingual spoken term detection, even for low resourced languages. A detailed discussion on different methodologies, along with a comparison, has been made. Various approaches for multilingual STD are organized based on feature representations, tokenization techniques, matching techniques and availability of resources. Different languages and corresponding datasets employed for the task of multilingual STD have been listed for quick referencing. A discussion of different benchmarking platforms for multilingual STD has also been included. The paper aims to provide a quick overview of different techniques and datasets widely used in multilingual STD research.

Tixl (TXL) is a crypto money project that uses Initial Coin Offering (ICO) to fund its development phase. Various problems underlying the existence of Tixl (TXL) and the proposed solutions are closely related to the condition of society in the information technology era. This study aims to determine the various factors underlying the use of ICO as a way to raise funds, examine Tixl (TXL) as a crypto money project and its position in public life which is reviewed in the perspective of the maṣlaḥa mursala, and token sales transactions in ICO organized by Tixl gmmbH in the perspective of Islamic treaty law. This research is a library research using a conceptual approach and qualitative analysis techniques. This paper shows that there are three factors underlying the use of ICOs as a way to raise funds, namely a change in orientation in the business model, tokenization, and characteristics of business activities. As a crypto money project, Tixl (TXL) cannot provide benefits that are comprehensive or kully, even the harm caused is greater than the benefits. Transaction of token sales in ICO held by Tixl (TXL) is a transaction that is damaged because it is contrary to maqāṣid al-sharī'ah.

This paper presents a concept of real estate tokenization, which includes legal, technological, and organizational aspects. The research introduces a theory of a Title Token – a digital record of ownership on the blockchain. It is discussed the principle of technological neutrality, where the traditional land registry is not necessarily abandoned in favor of blockchains, but instead, people gain the right to choose. Nowadays, public administrations use central-server databases, giving no alternatives for citizens. Recognition of the right of citizens to choose which technology to apply for managing their property rights creates a basis for free competition and the development of new technologies for better public services. Decentralized distributed ledgers are the key to decentralization. They enable more secure automation of legal procedures. On the contrary, centralization is a source of many issues in governance: abuse of power, corruption, inefficient governance, and high costs, slowness and complexity of bureaucratic procedures. With automation and reduction of intermediaries, the role of the government does not decrease but significantly changes, i.e. land cadaster bodies should not be monopolistic providers on the market. The paper introduces a theoretical basis for developing a new type of property registries.

The type–token distinction and the notion of ‘tokenization’ are proposed as analytical tools that may help us understand better the emergence of numbers and mathematical thinking from the non-mathematical cultural practices of the Upper Palaeolithic such as painting, decorating portable objects, or making ornaments from beads, as described in recent studies in cognitive archaeology. While the type–token distinction has been a salient element in recent debates in the philosophy of mathematics, it seems not yet to have been registered in those areas of cognitive archaeology concerned with numbers—and yet it may help to identify the circumstances under which the conceptual potential of cultural artefacts develops. The concept of tokenization permits us to identify a plausible link between non-numerical cultural resources and the emergence of numerical thought patterns. This approach offers a way of appreciating the role of artefacts without raising the notoriously difficult question of their original ‘meaning’.

This paper discusses the potential impacts of the so-called `initial coin offerings', and of several developments based on distributed ledger technology (`DLT'), on corporate governance. While many academic papers focus mainly on the legal qualification of DLT and crypto-assets, and most notably in relation to the potential definition of the latter as securities/financial instruments, the authors analyze some of the use cases based on DLT technology and their potential for significant changes of the corporate governance analyses. This article studies the consequences due to the emergence of new kinds of firm stakeholders, i.e. the crypto-assets holders, on the governance of small and medium-sized enterprises (`SMEs') as well as of publicly traded companies. Since early 2016, a new way of raising funds has rapidly emerged as a major issue for FinTech founders and financial regulators. Frequently referred to as initial coin offerings, Initial Token Offerings (`ITO'), Token Generation Events (`TGE') or simply `token sales', we use in our paper the terminology Initial Crypto-asset Offerings (`ICO'), as it describes more effectively than `initial coin offerings' the vast diversity of assets that could be created and which goes far beyond the payment instrument issue., Comment: URL des Documents de travail : https://centredeconomiesorbonne.univ-paris1.fr/documents-de-travail-du-ces/

Medical images are playing an important role in the medical domain. A mature medical visual question answering system can aid diagnosis, but there is no satisfactory method to solve this comprehensive problem so far. Considering that there are many different types of questions, we propose a model called CGMVQA, including classification and answer generation capabilities to turn this complex problem into multiple simple problems in this paper. We adopt data augmentation on images and tokenization on texts. We use pre-trained ResNet152 to extract image features and add three kinds of embeddings together to deal with texts. We reduce the parameters of the multi-head self-attention transformer to cut the computational cost down. We adjust the masking and output layers to change the functions of the model. This model establishes new state-of-the-art results: 0.640 of classification accuracy, 0.659 of word matching and 0.678 of semantic similarity in ImageCLEF 2019 VQA-Med data set. It suggests that the CGMVQA is effective in medical visual question answering and can better assist doctors in clinical analysis and diagnosis.

The field of sentiment analysis is widely utilized for analyzing the text data and then extracting the sentiment component out of that. The online commercial websites generates a huge amount of textual data via customer’s reviews, comments, feedbacks and tweets every day. Aspect level analysis of this data provides a great help to retailers in better understanding of customer’s expectations and then shaping their policies accordingly. However, a number of algorithms are existing these days to do aspect level sentiment detection on specified domains, but a few consider bipolar words (words which changes polarity according to context) while doing analyses. In this paper, a novel approach has been presented that utilize aspect level sentiment detection, which focuses on the features of the item. The work has been implemented and tested on Amazon customer reviews (crawled data) where aspect terms are identified first for each review. The system performs pre-processing operations like stemming, tokenization, casing, stop-word removal on the dataset to extract meaningful information and finally gives a rank for its classification in negativity or positivity.

Data stream increasing demands development new intellectual tools and methods for BigData processing. Existing anomalies detection approaches based on distance, density and ranking. But these approaches do not take into account data stream features. Unfortunately, in the context of streaming data, those methods more or less have some drawbacks and are not directly applied to streaming data, such as poor adaptability and extensibility, inability to detection novel anomaly, high model updating cost and slow updating speed so on. Besides, existing methods based on numeric or categorical data and do not support Ukrainian text data. Purpose of research: creating new stream data anomalies detection approach with Ukrainian and Russian language text supporting. Actually, several software libraries have Ukrainian language text support: Pymorphy2, OpenCorpora, LanguageTool, ВЕСУМ dictionary. Using data preprocessing (normalization, tokenization and noise reduction) and text abstracting for anomalies detection are proposed. Abstracting method developed on base combination of LSA and TextRank methods. For abstracting cosine similarity is used. For method evaluation was prepared Ukrainian dataset from news portal http://korrespondent.net/. Average cosine similarity between original dataset and dataset with 40% of volume is 0,9033, and for 40% of volume - 0,7812. Anomalies detection speed increase for “20%” dataset in 1.7 time in compare original dataset processing, and in 1.47 time for “40%” dataset. Anomalies detection method based on Isolation forest method. Input data: TF-IDF metrics, message lengths, TF-IDF metrics of original messages and “20%” / “40%” messages are similar. Text stream data anomalies detection approach is presented. Method includes preprocessing and Abstracting stage. Abstracting method developed on base combination of LSA and TextRank methods. Anomalies detection method based on a Isolation Forest method and data stream model. Ukrainian and Russian language text processing is supported. The processing speed of original and abstract data stream is compared.