Sequence Of Characters Research Articles

Emotional Analysis of Twitter Posts During the First Phase of the COVID-19 Pandemic in Greece: Infoveillance Study.

BackgroundThe effectiveness of public health measures depends upon a community’s compliance as well as on its positive or negative emotions.ObjectiveThe purpose of this study was to perform an analysis of the expressed emotions in English tweets by Greek Twitter users during the first phase of the COVID-19 pandemic in Greece.MethodsThe period of this study was from January 25, 2020 to June 30, 2020. Data collection was performed by using appropriate search words with the filter-streaming application programming interface of Twitter. The emotional analysis of the tweets that satisfied the inclusion criteria was achieved using a deep learning approach that performs better by utilizing recurrent neural networks on sequences of characters. Emotional epidemiology tools such as the 6 basic emotions, that is, joy, sadness, disgust, fear, surprise, and anger based on the Paul Ekman classification were adopted.ResultsThe most frequent emotion that was detected in the tweets was “surprise” at the emerging contagion, while the imposed isolation resulted mostly in “anger” (odds ratio 2.108, 95% CI 0.986-4.506). Although the Greeks felt rather safe during the first phase of the COVID-19 pandemic, their positive and negative emotions reflected a masked “flight or fight” or “fear versus anger” response to the contagion.ConclusionsThe findings of our study show that emotional analysis emerges as a valid tool for epidemiology evaluations, design, and public health strategy and surveillance.

Sequential alignment attention model for scene text recognition

Scene text recognition has been a hot research topic in computer vision due to its various applications. The state-of-the-art solutions usually depend on the attention-based encoder-decoder framework that learns the mapping between input images and output sequences in a purely data-driven way. Unfortunately, there often exists severe misalignment between feature areas and text labels in real-world scenarios. To address this problem, this paper proposes a sequential alignment attention model to enhance the alignment between input images and output character sequences. In this model, an attention gated recurrent unit (AGRU) is first devised to distinguish the text and background regions, and further extract the localized features focusing on sequential text regions. Furthermore, CTC guided decoding strategy is integrated into the popular attention-based decoder, which not only helps to boost the convergence of the training but also enhances the well-aligned sequence recognition. Extensive experiments on various benchmarks, including the IIIT5k, SVT, and ICDAR datasets, show that our method substantially outperforms the state-of-the-art methods.

Named Entity Recognition with Gating Mechanism and Parallel BiLSTM

In this paper, we propose a novel neural network for named entity recognition, which is improved in two aspects. On the one hand, our model uses a parallel BiLSTM structure to generate character-level word representations. By inputting character sequences of words into several independent and parallel BiLSTMs, we can obtain word representations from different representation subspaces, because the parameters of these BiLSTMs are randomly initialized. This method can enhance the expression abilities of character-level word representations. On the other hand, we use a two-layer BiLSTM with gating mechanism to model sentences. Since the features extracted by each layer in a multi-layer LSTM from texts contain different types of information, we use the gating mechanism to assign appropriate weights to the outputs of each layer, and take the weighted sum of these outputs as the final output for named entity recognition. Our model only changes the structure, does not need any feature engineering or external knowledge source, which is a complete end-to-end NER model. We used the CoNLL-2003 English and German datasets to evaluate our model and got better results compared with baseline models.

Accessible authentication: dyslexia and password strategies

PurposeThe purpose of this paper is to reveal the lived experiences of dyslexics in engaging with all kinds of alphanumeric authentication mechanisms.Design/methodology/approachA significant proportion of the world’s population experiences some degree of dyslexia, which can lead to spelling, processing, sequencing and retention difficulties. Passwords, being essentially sequences of alphanumeric characters, make it likely that dyslexics will struggle with these, even more so than the rest of the population. Here, this study explores the difficulties people with dyslexia face, their general experiences with passwords, the coping strategies they use and the advice they can provide to developers and others who struggle with passwords. This paper collects empirical data through semi-structured interviews with 13 participants. Thematic analysis was used to provide an in-depth view of each participant’s experience.FindingsThe main contribution of this paper is to provide evidence related to the inaccessibility dimensions of passwords as an authentication mechanism, especially for dyslexics and to recommend a solution direction.Research limitations/implicationsThere is a possible volunteer bias, as this study is dealing with self-reported data including historical and reflective elements and this paper is seeking information only from those with self-declared or diagnosed dyslexia. Furthermore, many expressed interest or curiosity in the relationship between dyslexia and password difficulties, for some a motivation for their participation. Finally, given that the participants told us that dyslexics might hide, it is possible that the experiences of those who do hide are different from those who chose to speak to us and thus were not hiding.Originality/valueA few authors have written about the difficulties dyslexics face when it comes to passwords, but no one has asked dyslexics to tell them about their experiences. This paper fills that gap.

Serial encryption using the functions of real variable

Context. Using the functions of a real variable in cryptosystems as keys allow increasing their cryptographic strength since it is more difficult to select such keys. Therefore, the development of such systems is relevant. Objective. Cryptosystems with symmetric keys are proposed for encrypting and decrypting a sequence of characters represented as a one-dimensional numerical array of ASCII codes. These keys are functions of a real variable that satisfies certain restrictions. They can be both continuous and discrete. Method. Two cryptosystem options are proposed. In the first embodiment, the transmitting and receiving sides select two functions, the first transmitted symbol, the area of the function definition, and the step of changing the function argument. Discrete messages are encrypted by calculating the first-order integral disproportion of the encrypted array using a function. The corresponding value of the second function is added to the obtained cipher of each symbol for scrambling to complicate the analysis of the intercepted message. On the receiving side, the second function is subtracted and decryption performed by the inverse transformation of the formula for integral disproportion. In the second version, sequential encryption is performed when the cipher obtained using one of the key functions in the first stage is encrypted again by calculating the disproportion using the second function, the key. Accordingly, in two stages, decryption is performed. Results. Examples of encryption and decryption of a sequence of text characters are presented. It is shown that the same character is encoded differently depending on its position in the message. In the given examples it is presented the difficulty of key functions parameters choosing and the cryptographic strength of the proposed cryptosystem. Conclusions. Variants of the cryptosystem using the first-order integral disproportion function are proposed, in which the functions of a real variable serve as keys. To “crack” such a system, it is necessary not only to select the form of each function but also to find the values of its parameters with high accuracy. The system has high cryptographic strength.

A Signal Processing Method for Text Language Identification

Language identification is a critical step prior to any natural language processing. In this paper, a signal processing method for Language Identification is proposed. Sequence of characters in a word and the order of words in stream identify the language. The sequence of characters in a stream provides a signature to recognize the language without understanding its meaning. The signature can be extracted using signal processing techniques via converting texts into time series. Although several research and commercial software have been developed to identify text language, they need a standard dictionary for each language. We proposed a dictionary independent method consisting of three main steps, I) preprocessing, II) clustering and finally III) classification. First, the texts are converted to time series using UTF-8 codes. Second, to group similar languages, the obtained series are clustered. Third, each cluster is decomposed into 32 sub-bands using a Wavelet packet, and 32 features are extracted from each sub-band. Also, a multilayer perceptron neural network is used to classify the extracted features. The proposed method was tested on our dataset with 31000 texts from 31 different languages. The proposed method achieved 72.20% accuracy for language identification.

MANGO: A Mask Attention Guided One-Stage Scene Text Spotter

Recently end-to-end scene text spotting has become a popular research topic due to its advantages of global optimization and high maintainability in real applications. Most methods attempt to develop various region of interest (RoI) operations to concatenate the detection part and the sequence recognition part into a two-stage text spotting framework. However, in such framework, the recognition part is highly sensitive to the detected results (e.g., the compactness of text contours). To address this problem, in this paper, we propose a novel Mask AttentioN Guided One-stage text spotting framework named MANGO, in which character sequences can be directly recognized without RoI operation. Concretely, a position-aware mask attention module is developed to generate attention weights on each text instance and its characters. It allows different text instances in an image to be allocated on different feature map channels which are further grouped as a batch of instance features. Finally, a lightweight sequence decoder is applied to generate the character sequences. It is worth noting that MANGO inherently adapts to arbitrary-shaped text spotting and can be trained end-to-end with only coarse position information (e.g., rectangular bounding box) and text annotations. Experimental results show that the proposed method achieves competitive and even new state-of-the-art performance on both regular and irregular text spotting benchmarks, i.e., ICDAR 2013, ICDAR 2015, Total-Text, and SCUT-CTW1500.

Dynamic Modeling Cross- and Self-Lattice Attention Network for Chinese NER

Word-character lattice models have been proved to be effective for Chinese named entity recognition (NER), in which word boundary information is fused into character sequences for enhancing character representations. However, prior approaches have only used simple methods such as feature concatenation or position encoding to integrate word-character lattice information, but fail to capture fine-grained correlations in word-character spaces. In this paper, we propose DCSAN, a Dynamic Cross- and Self-lattice Attention Network that aims to model dense interactions over word-character lattice structure for Chinese NER. By carefully combining cross-lattice and self-lattice attention modules with gated word-character semantic fusion unit, the network can explicitly capture fine-grained correlations across different spaces (e.g., word-to-character and character-to-character), thus significantly improving model performance. Experiments on four Chinese NER datasets show that DCSAN obtains stateof-the-art results as well as efficiency compared to several competitive approaches.

Efficient License Plate Recognition via Holistic Position Attention

License plate recognition (LPR) is a fundamental component of various intelligent transportation systems, and is always expected to be accurate and efficient enough in real-world applications. Nowadays, recognition of single character has been sophisticated benefiting from the power of deep learning, and extracting position information for forming a character sequence becomes the main bottleneck of LPR. To tackle this issue, we propose a novel holistic position attention (HPA) in this paper that consists of position network and shared classifier. Specifically, the position network explicitly encodes the character position into the maps of HPA, and then the shared classifier performs the character recognition in a unified and parallel way. Here the extracted features are modulated by the attention maps before feeding into the classifier to yield the final recognition results. Note that our proposed method is end-to-end trainable, character recognition can be concurrently performed, and no post-processing is needed. Thus our LPR system can achieve good effectiveness and efficiency simultaneously. The experimental results on four public datasets, including AOLP, Media Lab, CCPD, and CLPD, well demonstrate the superiority of our method to previous state-of-the-art methods in both accuracy and speed.

End-to-end keyword search system based on attention mechanism and energy scorer for low resource languages

Keyword search (KWS) means searching for keywords given by the user from continuous speech. Conventional KWS systems are based on Automatic Speech Recognition (ASR), where the input speech has to be first processed by the ASR system before keyword searching. In the recent decade, as deep learning and deep neural networks (DNN) become increasingly popular, KWS systems can also be trained in an end-to-end (E2E) manner. The main advantage of E2E KWS is that there is no need for speech recognition, which makes the training and searching procedure much more straightforward than the traditional ones. This article proposes an E2E KWS model, which consists of four parts: speech encoder–decoder, query encoder–decoder, attention mechanism, and energy scorer. Firstly, the proposed model outperforms the baseline model. Secondly, we find that under various supervision, character or phoneme sequences, speech or query encoders can extract the corresponding information, resulting in different performances. Moreover, we introduce an attention mechanism and invent a novel energy scorer, where the former can help locate keywords. The latter can make final decisions by considering speech embeddings, query embeddings, and attention weights in parallel. We evaluate our model on low resource conditions with about 10-hour training data for four different languages. The experiment results prove that the proposed model can work well on low resource conditions.

Position-Encoding Convolutional Network to Solving Connected Text Captcha

Abstract Text-based CAPTCHA is a convenient and effective safety mechanism that has been widely deployed across websites. The efficient end-to-end models of scene text recognition consisting of CNN and attention-based RNN show limited performance in solving text-based CAPTCHAs. In contrast with the street view image and document, the character sequence in CAPTCHA is non-semantic. The RNN loses its ability to learn the semantic context and only implicitly encodes the relative position of extracted features. Meanwhile, the security features, which prevent characters from segmentation and recognition, extensively increase the complexity of CAPTCHAs. The performance of this model is sensitive to different CAPTCHA schemes. In this paper, we analyze the properties of the text-based CAPTCHA and accordingly consider solving it as a highly position-relative character sequence recognition task. We propose a network named PosConv to leverage the position information in the character sequence without RNN. PosConv uses a novel padding strategy and modified convolution, explicitly encoding the relative position into the local features of characters. This mechanism of PosConv makes the extracted features from CAPTCHAs more informative and robust. We validate PosConv on six text-based CAPTCHA schemes, and it achieves state-of-the-art or competitive recognition accuracy with significantly fewer parameters and faster convergence speed.

Robustness of Differentiable Neural Computer Using Limited Retention Vector-based Memory Deallocation in Language Model

Recurrent neural network (RNN) architectures have been used for language modeling (LM) tasks that require learning long-range word or character sequences. However, the RNN architecture is still suffered from unstable gradients on long-range sequences. To address the issue of long-range sequences, an attention mechanism has been used, showing state-of-the-art (SOTA) performance in all LM tasks. A differentiable neural computer (DNC) is a deep learning architecture using an attention mechanism. The DNC architecture is a neural network augmented with a content-addressable external memory. However, in the write operation, some information unrelated to the input word remains in memory. Moreover, DNCs have been found to perform poorly with low numbers of weight parameters. Therefore, we propose a robust memory deallocation method using a limited retention vector. The limited retention vector determines whether the network increases or decreases its usage of information in external memory according to a threshold. We experimentally evaluate the robustness of a DNC implementing the proposed approach according to the size of the controller and external memory on the enwik8 LM task. When we decreased the number of weight parameters by 32.47%, the proposed DNC showed a low bits-per-character (BPC) degradation of 4.30%, demonstrating the effectiveness of our approach in language modeling tasks.

Convolutional neural network with joint stepwise character/word modeling based system for scene text recognition

Text recognition in the wild is a challenging task in the field of computer vision and machine learning. Existing optical character recognition engines cannot perform well in the natural scene. In this context, deep learning models have emerged as a powerful state-of-the-art technique in the classification and recognition process. This study proposes a new Convolutional Neural Network based system for scene text reading. We investigate how to combine the character recognition module followed by the word recognition module to achieve the overall system goal. The first module analyzes characters within multi-scale images by relaying on the power of the convolutional network and the fully connected network for character recognition. The second module relies on the Viterbi search to find the closest word to a given characters sequence. For the sake of more precision, a bigram based linguistic module is applied. The proposed system achieves the state-of-the-art performance on three standard scene text recognition benchmarks: chars74k, ICDAR 2003 and ICDAR 2013. In particular, this performance is proven on both of character and word recognition accuracy as well as speed aspects via a comparative study between different deep learning architectures.

Finding Better Subwords for Tibetan Neural Machine Translation

Subword segmentation plays an important role in Tibetan neural machine translation (NMT). The structure of Tibetan words consists of two levels. First, words consist of a sequence of syllables, and then a syllable consists of a sequence of characters. According to this special word structure, we propose two methods for Tibetan subword segmentation, namely syllable-based and character-based methods. The former generates subwords based on the Tibetan syllables, and the latter is based on Tibetan characters. In addition, we carry out experiments with these two subword segmentation methods on low-resource Tibetan-to-Chinese NMT, respectively. The experimental results show that both of them can improve translation performance, in which the subword segmentation based on character sequences can achieve better results. Overall, our proposed character-based subword segmentation is more simple and effective. Moreover, it can achieve better experimental results without paying much attention to the linguistic features of Tibetan.

Character-based handwritten text transcription with attention networks

The paper approaches the task of handwritten text recognition (HTR) with attentional encoder–decoder networks trained on sequences of characters, rather than words. We experiment on lines of text from popular handwriting datasets and compare different activation functions for the attention mechanism used for aligning image pixels and target characters. We find that softmax attention focuses heavily on individual characters, while sigmoid attention focuses on multiple characters at each step of the decoding. When the sequence alignment is one-to-one, softmax attention is able to learn a more precise alignment at each step of the decoding, whereas the alignment generated by sigmoid attention is much less precise. When a linear function is used to obtain attention weights, the model predicts a character by looking at the entire sequence of characters and performs poorly because it lacks a precise alignment between the source and target. Future research may explore HTR in natural scene images, since the model is capable of transcribing handwritten text without the need for producing segmentations or bounding boxes of text in images.

Lexical Patterns Based on Maximal Frequent Secuences for Automatic Keyphrase Extraction

This paper presents a method for the automatic keyphrase extraction task using lexical patterns. First, the patterns are obtained from a set of data and converted into regular expression search patterns, allowing to consider sequences of characters that define a phrase without depending on its syntactic or semantic characteristics and thus obtain a list of possible candidates. Besides, to select the best, only those that obtained a high weight will be considered, in the following four weights: Boolean (B), Precision (P), Recall (R), and F-Measure (F); which corresponds to the result obtained from each evaluated pattern, therefore a list is generating of the best 5,10 and 15 keyphrases for each document. The evaluation of the method was realized by length (L) and combination (C), where the combination takes the best candidates for each length (1 to 4). The method was tested in corpus of scientific articles using the SemEval-2010 data set for task 5.

Incorporating Hierarchical Characters into Phylogenetic Analysis.

Popular optimality criteria for phylogenetic trees focus on sequences of characters that are applicable to all the taxa. As studies grow in breadth, it can be the case that some characters are applicable for a portion of the taxa and inapplicable for others. Past work has explored the limitations of treating inapplicable characters as missing data, noting that this strategy may favor trees where internal nodes are assigned impossible states, where the arrangement of taxa within subclades is unduly influenced by variation in distant parts of the tree, and/or where taxa that otherwise share most primary characters are grouped distantly. Approaches that avoid the first two problems have recently been proposed. Here, we propose an alternative approach which avoids all three problems. We focus on data matrices that use reductive coding of traits, that is, explicitly incorporate the innate hierarchy induced by inapplicability, and as such our approach extend to hierarchical characters, in general. In the spirit of maximum parsimony, the proposed criterion seeks the phylogenetic tree with the minimal changes across any tree branch, but where changes are defined in terms of dissimilarity metrics that weigh the effects of inapplicable characters. The approach can accommodate binary, multistate, ordered, unordered, and polymorphic characters. We give a polynomial-time algorithm, inspired by Fitch's algorithm, to score trees under a family of dissimilarity metrics, and prove its correctness. We show that the resulting optimality criteria is computationally hard, by reduction to the NP-hardness of the maximum parsimony optimality criteria. We demonstrate our approach using synthetic and empirical data sets and compare the results with other recently proposed methods for choosing optimal phylogenetic trees when the data includes hierarchical characters. [Character optimization, dissimilarity metrics, hierarchical characters, inapplicable data, phylogenetic tree search.].

Handwritten Text Recognition using Fully Convolutional Network

Handwritten text recognition from images is challenging because there are many variations in handwriting as each person has a different writing style. This research implements multilevel recognition to solve this problem. In the first level, a Lexicon Convolutional Neural Network (CNN) model is used to recognize words (containing a set of word) that often appear in the text. If a word is not recognized by the Lexicon CNN model, which is designed for a limited number of words, then it goes to the next level. A character sequence recognition consisting of predicting the number of characters using CNN, cropping each character using a sliding window, and character recognition using Fully Convolutional Network (FCN), is applied at the second level. Experiments show that the system performance is promising. The experiment conducted using NIST Special Database 19 as a training dataset and a handwritten text on screen as a testing dataset. The best accuracy of word recognition, character number prediction, and character recognition is 99.98%, 98.56%, and 83.52%, respectively.

Perceptual expertise with Chinese characters predicts Chinese reading performance among Hong Kong Chinese children with developmental dyslexia.

This study explores the theoretical proposal that developmental dyslexia involves a failure to develop perceptual expertise with words despite adequate education. Among a group of Hong Kong Chinese children diagnosed with developmental dyslexia, we investigated the relationship between Chinese word reading and perceptual expertise with Chinese characters. In a perceptual fluency task, the time of visual exposure to Chinese characters was manipulated and limited such that the speed of discrimination of a short sequence of Chinese characters at an accuracy level of 80% was estimated. Pair-wise correlations showed that perceptual fluency for characters predicted speeded and non-speeded word reading performance. Exploratory hierarchical regressions showed that perceptual fluency for characters accounted for 5.3% and 9.6% variance in speeded and non-speeded reading respectively, in addition to age, non-verbal IQ, phonological awareness, morphological awareness, rapid automatized naming (RAN) and perceptual fluency for digits. The findings suggest that perceptual expertise with words plays an important role in Chinese reading performance in developmental dyslexia, and that perceptual training is a potential remediation direction.

New Construction of Family of MLCS Algorithms.

The multiple longest common subsequence (MLCS) problem involves finding all the longest common subsequences of multiple character sequences. This problem is encountered in a variety of areas, including data mining, text processing, and bioinformatics, and is particularly important for biological sequence analysis. By taking the MLCS problem and algorithms for its solution as research domain, this study analyzes the domain of multiple longest common subsequence algorithms, extracts features that algorithms in the domain do and do not have in common, and creates a domain feature model for the MLCS problem by using generic programming, domain engineering, abstraction, and related technologies. A component library for the domain is designed based on the feature model for the MLCS problem, and the partition and recur (PAR) platform is used to ensure that highly reliable MLCS algorithms can be quickly assembled through component assembly. This study provides a valuable reference for obtaining rapid solutions to problems of biological sequence analysis and improves the reliability and assembly flexibility of assembling algorithms.