Text Pairs Research Articles

Compressor Fault Diagnosis Knowledge: A Benchmark Dataset for Knowledge Extraction From Maintenance Log Sheets Based on Sequence Labeling

Compressor fault diagnosis requires expert knowledge. Using the sequence labeling technology, this expert knowledge can be automatically extracted from compressor maintenance log sheets. Previous studies indicate that sequence labeling methods often need a substantial amount of annotation data for knowledge extraction, Unfortunately, the annotation data are very scarce in the field of compressor fault diagnosis. In this paper, we introduce a benchmark dataset for extraction of knowledge suitable for air compressor fault diagnosis. First, we collected 11,418 pieces of information from air compressor maintenance log sheets. Fault description, service requests, causes and troubleshooting solutions were stored in a dataset for data preprocessing and masking. In addition, 6196 valid text pairs were developed after the “noises” in the raw dataset were cleaned. Second, five kinds of entities and sequences, such as equipment, faults, service requests, causes and troubleshooting solutions, were annotated by three subject experts. The annotation consistency was assessed with F1 scores. Furthermore, our proposed baseline model (or the BERT-BI-LSTM-CRF model) was compared against other five sequence labeling models (BI-LSTM-CRF, Lattice LSTM, BERT NER, ZEN, and ERNIE). The BERT-BI-LSTM-CRF model gives superior performance in extracting expert knowledge from the subject dataset. Although the baseline model is not the most cutting-edge model in the sequence labeling and named entity recognition fields, it indeed presents a great potential for compressor fault diagnosis. The dataset is available at https://github.com/chentao1999/CFDK .

Deep neural review text interaction for recommendation systems

Users’ reviews contain valuable information which are not taken into account in most recommender systems. According to the latest studies in this field, using review texts could not only improve the performance of recommendation, but it can also alleviate the impact of data sparsity and help to tackle this problem. In this paper, we present a neural recommender model which recommends items by leveraging user reviews. In order to predict user rating for each item, our proposed model, named MatchPyramidRecommenderSystem (MPRS), represents each user and item with their corresponding review texts. Thus, the problem of recommendation is viewed as a text matching problem such that the matching score obtained from matching user and item texts could be considered as a good representative of their joint extent of similarity. To solve the text matching problem, inspired by MatchPyramid (Pang et al., 2016), we employed an interaction-based approach according to which a matching matrix is constructed given a pair of input texts. The matching matrix, which has the property of hierarchical matching patterns, is then fed into a Convolutional Neural Network (CNN) to compute the matching score for the given user–item pair. Our experiments on the small data categories of Amazon review dataset show that our proposed model gains from 1.76% to 21.72% relative improvement compared to DeepCoNN model (Zheng et al., 2017), and from 0.83% to 3.15% relative improvement compared to TransNets model (Catherine and Cohen, 2017). Also, on two large categories, namely AZ-CSJ and AZ-Mov, our model achieves relative improvements of 8.08% and 7.56% compared to the DeepCoNN model, and relative improvements of 1.74% and 0.86% compared to the TransNets model, respectively.

AlgoLabel: A Large Dataset for Multi-Label Classification of Algorithmic Challenges

While semantic parsing has been an important problem in natural language processing for decades, recent years have seen a wide interest in automatic generation of code from text. We propose an alternative problem to code generation: labelling the algorithmic solution for programming challenges. While this may seem an easier task, we highlight that current deep learning techniques are still far from offering a reliable solution. The contributions of the paper are twofold. First, we propose a large multi-modal dataset of text and code pairs consisting of algorithmic challenges and their solutions, called AlgoLabel. Second, we show that vanilla deep learning solutions need to be greatly improved to solve this task and we propose a dual text-code neural model for detecting the algorithmic solution type for a programming challenge. While the proposed text-code model increases the performance of using the text or code alone, the improvement is rather small highlighting that we require better methods to combine text and code features.

Semantic Relationships Between Representational Gestures and Their Lexical Affiliates Are Evaluated Similarly for Speech and Text.

This research examined whether the semantic relationships between representational gestures and their lexical affiliates are evaluated similarly when lexical affiliates are conveyed via speech and text. In two studies, adult native English speakers rated the similarity of the meanings of representational gesture-word pairs presented via speech and text. Gesture-word pairs in each modality consisted of gestures and words matching in meaning (semantically-congruent pairs) as well as gestures and words mismatching in meaning (semantically-incongruent pairs). The results revealed that ratings differed by semantic congruency but not language modality. These findings provide the first evidence that semantic relationships between representational gestures and their lexical affiliates are evaluated similarly regardless of language modality. Moreover, this research provides an open normed database of semantically-congruent and semantically-incongruent gesture-word pairs in both text and speech that will be useful for future research investigating gesture-language integration.

Joint Sentiment Part Topic Regression Model for Multimodal Analysis

The development of multimodal media compensates for the lack of information expression in a single modality and thus gradually becomes the main carrier of sentiment. In this situation, automatic assessment for sentiment information in multimodal contents is of increasing importance for many applications. To achieve this, we propose a joint sentiment part topic regression model (JSP) based on latent Dirichlet allocation (LDA), with a sentiment part, which effectively utilizes the complementary information between the modalities and strengthens the relationship between the sentiment layer and multimodal content. Specifically, a linear regression module is developed to share implicit variables between image–text pairs, so that one modality can predict the other. Moreover, a sentiment label layer is added to model the relationship between sentiment distribution parameters and multimodal contents. Experimental results on several datasets verify the feasibility of our proposed approach for multimodal sentiment analysis.

Exit God: Border Crossings in Jīrṇoddhāra Procedure

Exit God: Border Crossings in Jīrṇoddhāra Procedure
 Jīrṇoddhāra, the disposal and replacement of an old idol or temple, necessitates the temporary departure of the deity from that place. The ritual for this departure will be followed as it is reported in two texts: the Piṅgalāmata and the Tantrasamuccaya. The Piṅgalāmata is useful in understanding the process of jīrṇoddhāra: the text defines the process, explains the need for it, rationalises the ritual procedure, and gives a brief account of it. The description in the Tantrasamuccaya gives finer detail on how the procedure should be carried out. Unlike other records of the subject, that of the Tantrasamuccaya does not glide over the most important moments of transition for the deity, but stops to consider them in depth, giving attention to his displacement in terms of saṅkocana (contraction), niṣkrāmaṇa (expulsion), and sūtracchidā (the severing of the cord). While this pair of texts offers the material in contrasting forms, the two sets of teaching are found to be congruent one with the other, making them companion pieces that give us a good view into what it means to invite the deity to leave. We find that in order for the deity to cross concentric borders to leave his base, the human ritual officiant must make a crossing too, and, indeed, the borders themselves will be moved around.

MDA: Multimodal Data Augmentation Framework for Boosting Performance on Sentiment/Emotion Classification Tasks

Multimodal data analysis has drawn increasing attention with the explosive growth of multimedia data. Although traditional unimodal data analysis tasks have accumulated abundant labeled datasets, there are few labeled multimodal datasets due to the difficulty and complexity of multimodal data annotation, nor is it easy to directly transfer unimodal knowledge to multimodal data. Unfortunately, there is little related data augmentation work in multimodal domain, especially for image–text data. In this article, to address the scarcity problem of labeled multimodal data, we propose a Multimodal Data Augmentation framework for boosting the performance on multimodal image–text classification task. Our framework learns a cross-modality matching network to select image–text pairs from existing unimodal datasets as the multimodal synthetic dataset, and uses this dataset to enhance the performance of classifiers. We take the multimodal sentiment analysis and multimodal emotion analysis as the experimental tasks and the experimental results show the effectiveness of our framework for boosting the performance on multimodal classification task.

Cultural and politico-religious challenges impacting critical reading of text for Iranian postgraduates in Australia

Abstract This study investigates the impact of cultural and politico-religious dominance on the practice of critical reading (CR) of texts by a group of Iranian postgraduate students in Australia. Four postgraduate students were interviewed individually four times (each time for reading one text) for critical understanding of two pairs of Persian texts, each with opposing viewpoints, on current socio-political and nationalistic debates of Iran. They were then involved in a focus group discussion for further critique of each other’s viewpoints. Findings indicate two major Persian constructs that influence CR: hefz-e zaaher ‘keeping up appearances’ and ta’sob/gheyrat, approximating to ‘one’s honor combined with prejudice and bigotry’. Findings also reveal that participants’ CR is contributed by heavy emotional attachment to nationalistic views engendered by Persian poetry. Chafe (1982), too, found that emotional attachment in appraising text was true with American English speakers. Finally, the focus group discussion had a slight impact on encouraging CR. Overall, it seems that participants’ repositioned journeys in Australia have influenced their perspective.

Grover on $$\,SIMON\,$$

For any symmetric key cryptosystem with $n$-bit secret key, the key can be recovered in $O(2^{n/2})$ exploiting Grover search algorithm, resulting in the effective key length to be half. In this direction, subsequent work has been done on AES and some other block ciphers. On the other hand, lightweight ciphers like SIMON was left unexplored. In this backdrop, we present Grover's search algorithm on all the variants of SIMON and enumerate the quantum resources to implement such attack in terms of NOT, CNOT and Toffoli gates. We also provide the T-depth of the circuits and the number of qubits required for the attack. We show that the number of qubits required for implementing Grover on SIMON $2n/mn$ is $O(2nr+mn)$, where $r$ is the number of chosen plaintext-cipher text pairs. We run a reduced version of SIMON in IBMQ quantum simulator and the 14-qubits processor as well. We found that where simulation supports theory, the actual implementation is far from the reality due to the infidelity of the gates and short decoherence time of the qubits. The complete codes for all version of SIMON have also been presented.

Effect of Previous Translations on Retranslation: A Case Study of Russian-Dutch Literary Translation

As Outi Paloposki and Kaisa Koskinen (2010) correctly stated in their article on the “fine line between retranslating and revising”, the exact relationship of a text with the previous translation(s) cannot always be determined, even if the most recent translation is presented (in the paratext to the edition) as a ‘retranslation’, ‘re-edition’ or ‘revision’. Indeed, in practice the “labels” of “(re)translation, (…) revisions, adaptations and retellings” are “hard to separate and cover different contents” in different circumstances (Van Coillie 2014).
 In this paper I will try to study the effect of a first or previous translation on the process of retranslating. In order to do that I will compare a number of literary translations with their predecessors. Literary translations are specifically chosen here because they are probably more often retranslated than other types of texts. Moreover, the style of the translated text is presumably more important than for other genres, which allows me to mutually compare the translations not only at lexical and syntactic, but also at stylistic level.
 In order to visualize the effect of a previous translation on a retranslation I will compare a number of Dutch translations of Russian literary works with their retranslations. On the one hand, I will compare three recent retranslations with older translations of the same work, ordered and published by the same publishing house, and explicitly announced to the reader as a ‘refreshed’ and reworked translation of the previous one. In these particular cases the retranslators were fully aware of the existence of another translation and the explicit reference to ‘retranslation’ virtually forced them to use the previous translation as a starting point. In order to avoid the influence of (1) the changing translation strategies throughout time, and (2) the possible idiosyncratic peculiarities of the specific translation strategy of one particular (but not representative) translator, I will use translations made by three different translators and published in approximately the same period.
 On the other hand, I will investigate two Dutch translations of the same Russian literary work, that were made independently from each other and were published virtually on the same day, as the translators were unaware of each other’s translation effort. In this case the ‘previous’ (the term is not really applicable in this case) translation could not have had any effect on the ‘retranslation’, which makes it an interesting case to compare with the three genuine ‘retranslations’.
 Special attention will be drawn to the differences in the translator’s decisions at lexical, syntactical and stylistic level. More specifically, the analysis will include a quantitative and qualitative approach. I will establish the amount of overlap in lexical, syntactical and stylistic choices in the four pairs of texts, and will try to find whether the retention or substitution of certain terms and linguistic features can be explained.

ECCParaCorp: a cross-lingual parallel corpus towards cancer education, dissemination and application

BackgroundThe increasing global cancer incidence corresponds to serious health impact in countries worldwide. Knowledge-powered health system in different languages would enhance clinicians’ healthcare practice, patients’ health management and public health literacy. High-quality corpus containing cancer information is the necessary foundation of cancer education. Massive non-structural information resources exist in clinical narratives, electronic health records (EHR) etc. They can only be used for training AI models after being transformed into structured corpus. However, the scarcity of multilingual cancer corpus limits the intelligent processing, such as machine translation in medical scenarios. Thus, we created the cancer specific cross-lingual corpus and open it to the public for academic use.MethodsAiming to build an English-Chinese cancer parallel corpus, we developed a workflow of seven steps including data retrieval, data parsing, data processing, corpus implementation, assessment verification, corpus release, and application. We applied the workflow to a cross-lingual, comprehensive and authoritative cancer information resource, PDQ (Physician Data Query). We constructed, validated and released the parallel corpus named as ECCParaCorp, made it openly accessible online.ResultsThe proposed English-Chinese Cancer Parallel Corpus (ECCParaCorp) consists of 6685 aligned text pairs in Xml, Excel, Csv format, containing 5190 sentence pairs, 1083 phrase pairs and 412 word pairs, which involved information of 6 cancers including breast cancer, liver cancer, lung cancer, esophageal cancer, colorectal cancer, and stomach cancer, and 3 cancer themes containing cancer prevention, screening, and treatment. All data in the parallel corpus are online, available for users to browse and download (http://www.phoc.org.cn/ECCParaCorp/).ConclusionsECCParaCorp is a parallel corpus focused on cancer in a cross-lingual form, which is openly accessible. It would make up the imbalance of scarce multilingual corpus resources, bridge the gap between human readable information and machine understanding data resources, and would contribute to intelligent technology application as a preparatory data foundation e.g. cancer-related machine translation, cancer system development towards medical education, and disease-oriented knowledge extraction.

SimiT: A Text Similarity Method Using Lexicon and Dependency Representations

Semantic textual similarity methods are becoming increasingly crucial in text mining research areas such as text retrieval and summarization. Existing methods of text similarity have often been computed by their shallow or syntactic representation rather than considering their semantic content and meanings. This paper focuses mainly on computing the similarity between sentences without a supervised learning approach, only considering their word-level coherence which is calculated by a hybrid method of dependency parser and lexicon embeddings. Hence, we concentrate on structural similarity between text pairs by regarding their dependency parser embeddings. Our hybrid method also pays attention to the semantic information of words implied in the sentences. In the evaluation, we compare our method with the state-of-the-art semantic similarity measures in a well-known dataset. Our method outperforms most of the studies in the literature and the overall performance achieves better results when combining the similarity scores of both embedding models.

ОПРЕДЕЛЕНИЕ НАБОРА УСЛОВИЙ ДЛЯ АВТОМАТИЧЕСКОГО ПОИСКА ОПТИМАЛЬНОГО ВАРИАНТА ГИБРИДНОГО МАШИННОГО ПЕРЕВОДА ТЕКСТА НА ГРАФЕМНОМ УРОВНЕ

The article is devoted to the Algorithmic Search for Optimal Solutions for evaluating andimproving the Quality of Hybrid Machine Translation of Text. The Object of the Research is Textson any Alphabetical Languages with different Bases (Alphabets), as well as their Translations intoother Alphabetical Languages.Currently, existing Methods and Means of Hybrid Machine Translationare characterized by a wide variety of Quality Assessment Algorithms, but the Disadvantageof these Methods is that most of them do not have Clear Criteria, Limitations and Schemes of Assessments,eventually, the Result of the Translation in most cases does not correspond to the Levelof Publication. The Aim of the Work is to determine a Set of Conditions for automatic search forthe Best Option of Hybrid Machine Translation of Text at the Level of Graphemes.The Main Tasksto be solved during the Research are the Search for Qualitative and Quantitative Conditions, includingthe maximum, minimum and average values of the Lengths of Translations, Reverse Translationsand Editorial Distances between Pairs of Texts that have the Same Meaning. The ScientificNovelty lies in use the Graphical Representation of the Model of Alphabetic Languages at theLevel of Graphemes in the Form of a Cartesian Coordinate System with a Dimension equal to aUnit Editorial Distance (by Levenstein). When solving the de Goui’s Theorem, the current Rules ofStandardization PR 50.1.027–2014 "Rules for the Provision of Translation and Special Types ofLinguistic Services", the Method of Decanonicalization and the Model "Original Text – Translation– Reverse Translation" were used. As a Result, Actual and Practically Applicable Solutionsfor the Problems under consideration are obtained. In this regard; this Work may be interest to awide range of Specialists engaged in Machine Translation and Translation Studies.

Algebraic Analysis of a Simplified Encryption Algorithm GOST R 34.12-2015

In January 2016, a new standard for symmetric block encryption was established in the Russian Federation. The standard contains two encryption algorithms: Magma and Kuznyechik. In this paper we propose to consider the possibility of applying the algebraic analysis method to these ciphers. To do this, we use the simplified algorithms Magma ⊕ and S-KN2. To solve sets of nonlinear Boolean equations, we choose two different approaches: a reduction and solving of the Boolean satisfiability problem (by using the CryptoMiniSat solver) and an extended linearization method (XL). In our research, we suggest using a security assessment approach that identifies the resistance of block ciphers to algebraic cryptanalysis. The algebraic analysis of an eight-round Magma (68 key bits were fixed) with the CryptoMiniSat solver demanded four known text pairs and took 3029.56 s to complete (the search took 416.31 s). The algebraic analysis of a five-round Magma cipher with weakened S-boxes required seven known text pairs and took 1135.61 s (the search took 3.36 s). The algebraic analysis of a five-round Magma cipher with disabled S-blocks (equivalent value substitution) led to getting only one solution for five known text pairs in 501.18 s (the search took 4.92 s). The complexity of the XL algebraic analysis of a four-round S-KN2 cipher with three text pairs was 236.33 s (took 1.191 Gb RAM).

Trekstuaaliset tilat

Translation studies and textual scholarship have much common ground. In this paper, we set to explore how three of their central concepts, translation, transmission and text, can be approached from a multidisciplinary point of view. Combining these three central terms, we call this approach trextual. This paper is based on the workshop on trextuality held at the KäTu Symposium in Tampere in 2019. In the workshop, the discussion built upon an introduction, which examined the foundations of trextuality, and four papers, which addressed genetic translation criticism, retranslation, polyphony and translation of intertextuality, as well as multi-modality of audiovisual translation, respectively. The workshop revealed that the juxtaposing of two text-oriented disciplines, translation studies and textual studies, and comparing their similarities and differences reveals unexplored areas and weaknesses in their axiomatic fundaments. In particular, the papers presented in the workshop invited to reconsider transmission, definitions of text, and source text-target text pairs in different contexts. The workshop provided a starting point for further trextual studies exploring such cross-disciplinary questions.

Learning paths in synthesis writing: Which learning path contributes most to which learning outcome?

This paper presents a secondary analysis of data collected during an intervention study in which students learnt to synthesise pairs of texts presenting opposite views on controversial issues. The original intervention study included two treatments and examined the effects of two instruction conditions when instructional materials and tasks were held constant. The participants were 114 undergraduate psychology students. The object of the instruction was a guide on strategies for writing an argumentative synthesis text. However, the instruction varied between explicit strategy instruction, consisting of explaining each of the process’s four phases (exploring and identifying arguments and counterarguments, contrasting positions, drawing an integrative conclusion, and organising and revising the final draft), modelled via videos, versus self-study of the written strategy guide. After the initial instruction session, the students in both groups practiced collaboratively writing synthesis texts over two sessions with access to the strategy guide. The primary study compared the individually written pre- and posttest syntheses and found statistically significant differences favouring explicit instruction in both dependent variables: the argumentation coverage and the level of integration. The secondary analysis reported in the current paper involved scoring additional written syntheses produced during two practice sessions and then analysing the data for all time points (pretest, posttest, and the two practice sessions) using structural equation modelling (SEM) to test whether explicit instruction directly or indirectly affected the two indicators of good argumentative synthesis texts—argument coverage and integration—via the following collaborative practice. The results suggested two different learning paths for both dependent variables: explicit instruction is effective for both variables, while collaborative practice only has an additional indirect effect on argument coverage.

Studying texts in a non-native language: A further investigation of factors involved in the L2 recall cost.

With academic internationalisation at full speed, English is increasingly used as a medium of instruction in higher education. The question arises of whether unbalanced bilinguals remember study materials in a non-native language (L2) as well as in a first language (L1). In previous studies, we found a disadvantage for students recalling short, expository texts in L2 compared with L1, but no such disadvantage for a true/false recognition test, not even on delayed tests after a month. As no additional forgetting occurs, the quality of the memory trace seems to be equally strong in both languages and the recall cost might be caused by a lack of production skill in L2. To test this hypothesis, we ran experiments in L1-L1, L2-L1, and L2-L2 conditions with free and cued recall (short open questions). We replicate the L2 free recall cost reported earlier and show that it is due to the encoding in L2 rather than to an L2 production cost. In contrast, we found no significant difference in a new pair of texts with short, cued recall questions, though there was a trend in the expected direction. A summary of the effect sizes obtained so far shows a considerable variety in magnitudes (with rather big confidence intervals), suggesting that the cost of studying in L2 depends on several factors such as study time, test requirement, and language proficiency level.

Matching with Text Data: An Experimental Evaluation of Methods for Matching Documents and of Measuring Match Quality

Matching for causal inference is a well-studied problem, but standard methods fail when the units to match are text documents: the high-dimensional and rich nature of the data renders exact matching infeasible, causes propensity scores to produce incomparable matches, and makes assessing match quality difficult. In this paper, we characterize a framework for matching text documents that decomposes existing methods into (1) the choice of text representation and (2) the choice of distance metric. We investigate how different choices within this framework affect both the quantity and quality of matches identified through a systematic multifactor evaluation experiment using human subjects. Altogether, we evaluate over 100 unique text-matching methods along with 5 comparison methods taken from the literature. Our experimental results identify methods that generate matches with higher subjective match quality than current state-of-the-art techniques. We enhance the precision of these results by developing a predictive model to estimate the match quality of pairs of text documents as a function of our various distance scores. This model, which we find successfully mimics human judgment, also allows for approximate and unsupervised evaluation of new procedures in our context. We then employ the identified best method to illustrate the utility of text matching in two applications. First, we engage with a substantive debate in the study of media bias by using text matching to control for topic selection when comparing news articles from thirteen news sources. We then show how conditioning on text data leads to more precise causal inferences in an observational study examining the effects of a medical intervention.

Enhanced Text Matching Based on Semantic Transformation

Text matching is the core of natural language processing (NLP) system. It's considered as a touchstone of the NLP, and it aims to find whether text pairs are equal in semantics. However, the semantic gap in text matching is still an open problem to solve. Inspired by successes of cycle-consistent adversarial network (CycleGAN) in image domain transformation, we propose an enhanced text matching method based on the CycleGAN combined with the Transformer network. Based on the proposed method, the text semantics in a source domain is transferred to a similar or different target domain, and the semantic distance between text pairs is decreased. Meanwhile, we demonstrate our method in paraphrase identification and question answer matching. The matching degree is computed by a standard text matching model to evaluate the transforming influence on narrowing the text semantic gap. The experiments show that our method achieves text domain adaptation, and the effects on different matching models are remarkable.

PATIENT PREFERENCES FOR LEVEL OF HEALTH LITERACY IN APPS FOR CHRONIC DISEASE SELF-MANAGEMENT

In spite of expert recommendations that written material should be provided at a level of health literacy that matches that of the person receiving it, there have been few studies of matching. In this study we evaluated the utility of a new strategy to assess patients’ preference for information at different difficulties and assessed the relation of their preference to measured health literacy and health locus of control (LOC). We measured health literacy in participants then asked them to choose between pairs of texts with the same content but at the 3rd, 6th, or 8th-grade levels. Statistical analyses assessed the relation of participants’ health literacy to their preference as well as personal characteristics. Participants (n = 155) were 84 women and 71 men aged from 40 to 82 years (mean=57; 136 African Americans and 19 whites). Participants had clear preferences: those with lower levels of health literacy preferred texts at the 3rd grade level and those with higher levels preferred more difficult texts. Preference was not related to age, gender or race but was to education and health literacy (p < 0.05). Persons who chose more difficult texts reported higher levels of internal health locus of control (t [144] = 2.68, p = 0.01). A predictive analytic model using education and preference resulted in 80% correct classification of participants. Using this model may be a simple way to match information presentation to patients’ level of health literacy. Further research on this strategy is needed.