Chinese Corpus Research Articles

Evaluating the Performance of Large Language Models (LLMs) in Answering and Analysing the Chinese Dental Licensing Examination.

This study aimed to simulate diverse scenarios of students employing LLMs for CDLE examination preparation, providing a detailed evaluation of their performance in medical education. A stratified random sampling strategy was implemented to select and subsequently revise 200 questions from the CDLE. Seven LLMs, recognised for their exceptional performance in the Chinese domain, were selected as test subjects. Three distinct testing scenarios were constructed: answering questions, explaining questions and adversarial testing. The evaluation metrics included accuracy, agreement rate and teaching effectiveness score. Wald χ2 tests and Kruskal-Wallis tests were employed to determine whether the differences among the LLMs across various scenarios and before and after adversarial testing were statistically significant. The majority of the tested LLMs met the passing threshold on the CDLE benchmark, with Doubao-pro 32k and Qwen2-72b (81%) achieving the highest accuracy rates. Doubao-pro 32k demonstrated the highest 98% agreement rate with the reference answers when providing explanations. Although statistically significant differences existed among various LLMs in their teaching effectiveness scores based on the Likert scale, all these models demonstrated a commendable ability to deliver comprehensible and effective instructional content. In adversarial testing, GPT-4 exhibited the smallest decline in accuracy (2%, p = 0.623), while ChatGLM-4 demonstrated the least reduction in agreement rate (14.6%, p = 0.001). LLMs trained on Chinese corpora, such as Doubao-pro 32k, demonstrated superior performance compared to GPT-4 in answering and explaining questions, with no statistically significant difference. However, during adversarial testing, all models exhibited diminished performance, with GPT-4 displaying comparatively greater robustness. Future research should further investigate the interpretability of LLM outputs and develop strategies to mitigate hallucinations generated in medical education.

A DeBERTa-Based Semantic Conversion Model for Spatiotemporal Questions in Natural Language

To address current issues in natural language spatiotemporal queries, including insufficient question semantic understanding, incomplete semantic information extraction, and inaccurate intent recognition, this paper proposes NL2Cypher, a DeBERTa (Decoding-enhanced BERT with disentangled attention)-based natural language spatiotemporal question semantic conversion model. The model first performs semantic encoding on natural language spatiotemporal questions, extracts pre-trained features based on the DeBERTa model, inputs feature vector sequences into BiGRU (Bidirectional Gated Recurrent Unit) to learn text features, and finally obtains globally optimal label sequences through a CRF (Conditional Random Field) layer. Then, based on the encoding results, it performs classification and semantic parsing of spatiotemporal questions to achieve question intent recognition and conversion to Cypher query language. The experimental results show that the proposed DeBERTa-based conversion model NL2Cypher can accurately achieve semantic information extraction and intent understanding in both simple and compound queries when using Chinese corpus, reaching an F1 score of 92.69%, with significant accuracy improvement compared to other models. The conversion accuracy from spatiotemporal questions to query language reaches 88% on the training set and 92% on the test set. The proposed model can quickly and accurately query spatiotemporal data using natural language questions. The research results provide new tools and perspectives for subsequent knowledge graph construction and intelligent question answering, effectively promoting the development of geographic information towards intelligent services.

The Beijing Sentence Corpus II: A cross-script comparison between traditional and simplified Chinese sentence reading

We introduce a sentence corpus with eye-movement data in traditional Chinese (TC), based on the original Beijing Sentence Corpus (BSC) in simplified Chinese (SC). The most noticeable difference between TC and SC character sets is their visual complexity. There are reaction time corpora in isolated TC character/word lexical decision and naming tasks. However, up to now natural TC sentence reading corpus with recorded eye movements has not been available for general public. We report effects of word frequency, visual complexity, and predictability on eye movements on fixation location and duration based on 60 native TC readers. In addition, because the current BSC-II sentences are nearly identical to the original BSC sentences, we report similarities and differences of the linguistic influences on eye movements for the two varieties of written Chinese. The results shed light on how visual complexity affects eye movements. Together, the two sentence corpora comprise a useful tool to establish cross-script similarities and differences in TC and SC.

PhonemeVec: A Phoneme-Level Contextual Prosody Representation For Speech Synthesis

Recently, fine-grained prosody representations have emerged and attracted growing attention to address the one-to-many problem in text-to-speech (TTS). In this paper, we propose the PhonemeVec, a pre-trained prosody representations with considering the contextual information. To obtain the contextual prosody representations, we improve the data2vec framework according to the characteristics of prosody to extract the PhonemeVec from the low-band mel-spectrogram, and pre-train on a 960 hours Chinese corpus with high quality and diverse pronunciation. PhonemeVec is subsequently integrated into FastSpeech2, supervising the prosody modeling of the text encoder. Experiments conducted on the Blizzard Challenge 2019 dataset show that the integration of PhonemeVec results in the synthesis of more natural speech. Additionally, objective evaluations confirm that the application of PhonemeVec reduces the distortions between the generated speech and original recordings in terms of duration and F0. Audio samples can be found in http://home.ustc.edu.cn/~wsmzzz/PhonemeVec/demo.html.

Consistency Verification Method of Chinese and Russian Traditional Culture Translation Text Corpus Based on CNN-BiGRU

China and Russia have completely different cultural backgrounds, historical traditions and ways of thinking, which makes it difficult to directly correspond or accurately convey many culturally unique concepts, customs and expressions during the translation process. For example, some idioms, sayings or cultural symbols may have completely different meanings or no equivalents in the two cultures. Therefore, there are problems such as poor verification resulting in the consistency verification of Chinese and Russian traditional cultural external publicity translation text corpus. In order to solve this problem, this paper designs a consistency verification method for Chinese and Russian traditional cultural external publicity translation text corpus based on CNN-BiGRU. Through the part-of-speech correspondence of the Chinese and Russian traditional cultural external publicity translation text corpus, it is represented by vectors, the feature weights of the Chinese and Russian traditional cultural external publicity translation text corpus are calculated, the collocation information between the feature items is determined, and the maximum likelihood of the features is calculated using data smoothing technology. The rules are adjusted to extract the characteristics of the Chinese and Russian traditional cultural external publicity translation text corpus. Preprocess the key feature source documents of each Chinese and Russian traditional culture external publicity translation text corpus, determine the source documents and their corresponding key features of the traditional culture external publicity translation text corpus, and use TF-IDF and IDF to calculate the key feature sources of the corpus. Based on the rarity of document data, the BERT model is introduced as an encoder to determine the key features of the Chinese and Russian traditional cultural external publicity translation text corpus, integrating the CNN-BiGRU algorithm to design a consistency verification algorithm for Chinese and Russian traditional cultural external publicity translation text corpus. Set up the convolution layer, weight sharing layer, pooling layer, and BiGRU verification layer to build a CNN-BiGRU model to achieve consistency verification of the Chinese and Russian traditional cultural external publicity translation text corpus. Experimental results show that this method can improve the consistency verification effect of Chinese and Russian traditional cultural external publicity translation text corpus.

Speech Emotion Recognition Algorithm of Intelligent Robot Based on ACO-SVM

Despite the significant advancements in computer speech emotion recognition technology, the deployment of intelligent robots in this domain continues to encounter challenges related to inefficiency and emotional ambiguity. Aiming at the shortcomings of existing algorithms in the accuracy and processing of complex emotion states, a novel emotion recognition model of intelligent robot speech based on ACO algorithm and multi-level support vector machine is proposed. This paper first identifies the efficiency and accuracy problems existing in the current emotion recognition system, and clarified the research gap, that is, how to effectively improve the recognition performance of the model under diverse emotion states. To solve this problem, genetic algorithm is used to fine-tune SVM model parameters, and multi-level SVM classification architecture is constructed to enhance the accuracy of emotion recognition. In the feature selection stage, ACO algorithm is introduced to explore the optimal combination of emotion features, aiming at improving the efficiency and robustness of emotion recognition. The experimental results showed that the average recognition rate of the multilevel SVM model optimized by genetic algorithm was as high as 81.26%, which was 8.78 percentage points higher than that before optimization. When the ACO algorithm was combined with the optimized multi-stage SVM, the average recognition rate of German and Chinese speech emotion corpus was further improved to 85.62%, with an improvement of 4.38 percentage points. In particular, the optimized algorithm performed significantly better than other emotion categories in recognizing complex emotions such as "anger" and "sadness." This paper not only offers an innovative solution for the emotion recognition technology within the realm of intelligent robots, but also extends its methodology to have significant implications for related fields, thereby strongly advancing the in-depth application and development of intelligent bionic algorithms in speech emotion recognition. In addition, through this series of optimization strategies, this paper significantly improves the accuracy of emotion recognition, laying a solid foundation for intelligent robots to achieve more natural and efficient human-computer interaction.

A hybrid BiLSTM-CNN deep learning model for Chinese sentiment analysis of online car reviews

Although sentiment analysis with datasets in English has achieved significant progress, there are still relatively few studies on sentiment analysis in the area of Chinesecar review texts. In addition, the existing Chinese text sentiment analysis methods cannot simultaneously extract the contextual features and the local semantic information from the review texts. To address the above issues, a hybrid deep learning model namely BiLSTM-CNN was designed and validated with Chinese car review texts. We trained word vectors with the CBOW model by combining the Chinese Wikipedia corpus with other open-source car-related corpora. Such word vectors include car-specific vocabulary, which improves sentimental classification accuracy. Experimental results show that the performance indices of the deep learning models (CNN, LSTM, BiLSTM) are much better than the KNN, SVM, Naïve Bayes, and RF model, which is attributed to deep learning's powerful feature extraction ability and nonlinear fitting ability. Furthermore, when compared to deep learning models such as CNN, LSTM, BiLSTM, and LSTM-CNN, the suggested hybrid BiLSTM-CNN model outperformed the others. The results may provide references for consumers to buy a car and for car companies to optimize their products.

Identifying artificial intelligence–generated content in online Q&A communities through interpretable machine learning

This study aims to construct a comprehensive feature system for identifying artificial intelligence–generated content (AIGC) in online Q&A communities, thus uncovering the key factors and mechanisms influencing the identification of AIGC. First, based on the theory of systemic functional linguistics (SFL) and information quality (IQ), this article extracts vocabulary, content, structure, and emotional features from the text, and identifies the AIGC through nine mainstream machine learning algorithms. Subsequently, three widely used resampling strategies are exploited to address the category imbalance problem. The grid search optimisation algorithm fine-tunes different combinations of parameters to improve the performance of the identification classifier. Finally, SHAP values are introduced to evaluate and elucidate the global feature importance and feature influence mechanism. A Chinese corpus from the Zhihu Q&A community is constructed to verify the validity of these methods. The experimental results show that the eXtreme Gradient Boosting (XGBoost) model optimised with hybrid sampling and grid search parameters exhibits excellent performance in identifying AI-generated text, which achieves an F1-score of 0.9935, an improvement of 0.11 percentage points over the original model. In addition, all four dimensions of features constructed in this article contribute to AI-generated text identification, and the results of feature interpretability analysis show the greatest impact of features that focus on content readability. The study facilitates the identification and labelling of AIGC in online Q&A communities, thereby enhancing transparency and accountability of information shared online.

Comparison of Judgment Document Information Between China and Russia Based on Corpus

The study of legal comparisons has become particularly important as the number of foreign-related legal disputes has increased dramatically with the growing number of cross-border trade. This study discovers the differences and similarities between Chinese and Russian judgment documents and provides a linguistic basis for the academic theoretical researches and the call for linguists to enter the courtrooms by establishing a corpus of Chinese and Russian judgement documents, quantifying and analysing the Chinese and Russian judgement documents based on the Legal Discourse Information Theory. It can be found that the macro-structure of civil ruling documents is clearer but does not provide more micro-information. By adding language elements, it can be seen that there are more information points and information components in Russian criminal judgment documents. The information points in Chinese civil judgment documents are more concentrated than in Russian civil judgment documents, while they are more similar in terms of the proportion and type of information components.

Development and evaluation of a large language model of ophthalmology in Chinese

BackgroundLarge language models (LLMs), such as ChatGPT, have considerable implications for various medical applications. However, ChatGPT’s training primarily draws from English-centric internet data and is not tailored explicitly to the...

Faithfulness Fails to Fix it All: A Corpus-Driven Investigation into English Translation Issues in Abstracts of Chinese Civil Engineering Research Articles

The primary concern in translating research article abstracts into English is to ensure accuracy and efficiency. This study established two corpora: CCEJAC (Chinese Civil Engineering Journal Abstract Corpus) and ICEJAC (International Civil Engineering Journal Abstract Corpus). The paper applied Hyland’s five-move model and established ten sub-corpora for comparative studies of equivalent moves in CCEJAC and ICEJAC. It was observed that CCEJAC and ICEJAC differ in both lexical and syntactic levels. Specifically, differences were found in the use of general academic vocabulary and bundles, as well as in sentence length and focus. It can be concluded that translation versions in CCEJAC prioritize preserving the literal meaning and the original sentence structures, which may end in undesirable outcomes. These contrasts are the main factors contributing to English translation issues. To address these, this study thoroughly investigated the English translation issues in each move in CCEJAC. The aim is to refine English versions of abstracts to be more concise and practical, in a bid to further facilitate academic communication between Chinese and international scholars in the field of civil engineering.

The extraction method used for English–Chinese machine translation corpus based on bilingual sentence pair coverage

Abstract To improve the effect of corpus extraction in bilingual English–Chinese machine translation (ECMT), this article combines a machine learning algorithm with the ECMT corpus extraction method based on bilingual sentence pair coverage. Aiming at the propagation characteristics of medium- and short-range speeches, and employing multi-band speech waveforms in the troposphere, the actual requirements of remote machine translation are combined with the free-space, 1546, and the dual-path models, respectively, to analyze three typical air-to-ground wave propagation models. A simulation is conducted for both predictions and comparisons. In addition, a radio wave propagation model within the line of sight in the troposphere is established. The results suggest that the improved dual-path model is consistent with the propagation characteristics of radio waves in the troposphere. Moreover, the experiments show that the proposed ECMT corpus extraction method based on bilingual sentence pair coverage can play an important role in more accurate English–Chinese translations.

Research on the Training and Application Methods of a Lightweight Agricultural Domain-Specific Large Language Model Supporting Mandarin Chinese and Uyghur

In the field of Natural Language Processing (NLP), the lack of support for minority languages, especially Uyghur, the scarcity of Uyghur language corpora in the agricultural domain, and the lightweight nature of large language models remain prominent issues. This study proposes a method for constructing a bilingual (Uyghur and Chinese) lightweight specialized large language model for the agricultural domain. By utilizing a mixed training approach of Uyghur and Chinese, we extracted Chinese corpus text from agricultural-themed books in PDF format using OCR (Optical Character Recognition) technology, converted the Chinese text corpus into a Uyghur corpus using a rapid translation API, and constructed a bilingual mixed vocabulary. We applied the parameterized Transformer model algorithm to train the model for the agricultural domain in both Chinese and Uyghur. Furthermore, we introduced a context detection and fail-safe mechanism for the generated text. The constructed model possesses the ability to support bilingual reasoning in Uyghur and Chinese in the agricultural domain, with higher accuracy and a smaller size that requires less hardware. It (our work) addresses issues such as the scarcity of Uyghur corpora in the agricultural domain, mixed word segmentation and word vector modeling in Uyghur for widespread agricultural languages, model lightweighting and deployment, and the fragmentation of non-relevant texts during knowledge extraction from small-scale corpora. The lightweight design of the model reduces hardware requirements, facilitating deployment in resource-constrained environments. This advancement promotes agricultural intelligence, aids in the development of specific applications and minority languages (such as agriculture and Uyghur), and contributes to rural revitalization.

A refined and concise model of indices for quantitatively measuring lexical richness of Chinese university students’ EFL writing

In the existing literature, scholars have proposed various indices to measure the lexical richness (LR) of English as a foreign language (EFL) writing. However, there are currently issues of redundant indices and inconsistent usage. Attempting to address the research question of which indices are the most sensitive and effective ones to distinguish between different grade levels of Chinese university students’ EFL writing, this study aims to put forward a refined and concise model of indices that can truthfully reflect LR in EFL writing. A total of 180 compositions were selected from a Chinese EFL learner corpus: <i>Spoken and written English corpus of Chinese learners</i>. Scores of 28 LR indices of these compositions were computed using the software <i>Lexical Complexity Analyzer</i>,<i> MATTR</i>,<i> </i>and<i> Coh-Metrix</i>. One-way ANOVA or Welch’s ANOVA, depending on the variable’s homogeneity of variances, was conducted for each index. Two criteria were applied to determine which index of a measure should be included in the refined model: whether the difference of an index is significant among different grade levels and the effect size of ANOVA. Based on the quantitative results of ANOVAs and qualitative human judgment based on literature, six indices of the six LR measures were included in the refined model: lexical density, lexical sophistication-I, verb sophistication-II, number of different words-expected sequence 50, corrected TTR, and squared verb variation-I. This refined model addresses the issues of redundancy and inconsistency in previous studies, providing a more accurate and efficient tool for assessing LR in EFL writing.

An adaptive multi-neural network model for named entity recognition of Chinese mechanical equipment corpus

Mining entities from open Chinese mechanical equipment texts have become prevailing in the intelligent manufacturing field. However, compared to named entities in other domains, it is very hard to determine entity boundaries in mining Chinese mechanical equipment texts because there is no unified standard for entity simplification, and digits and units are mixed in the text. To address the issue, this paper presents an entity boundary-define strategy and constructs a mechanical equipment-oriented corpus called MECorpus using open Chinese mechanical equipment texts by combining domain knowledge. A multi-neural network collaboration model Adaptive-BERT-BiLSTM-CRF-Rating (ABBCR) is then proposed for mechanical equipment named entity recognition. The novelty of ABBCR is characterised by its adaptive input mechanism and rating score ability for identified entities. Various experiments about ABBCR model selection, evaluation and application are conducted on MECorpus. Experimental results show that the ABBCR model provides high-quality mechanical equipment entities for constructing the mechanical equipment knowledge graph. ABBCR combined with the large language model has proved to be a promising method to manage complex mechanical equipment expertise.

Improved XLNet modeling for Chinese named entity recognition of edible fungus.

The diversity of edible fungus species and the extent of mycological knowledge pose significant challenges to the research, cultivation, and popularization of edible fungus. To tackle this challenge, there is an urgent need for a rapid and accurate method of acquiring relevant information. The emergence of question and answer (Q&A) systems has the potential to solve this problem. Named entity recognition (NER) provides the basis for building an intelligent Q&A system for edible fungus. In the field of edible fungus, there is a lack of a publicly available Chinese corpus suitable for use in NER, and conventional methods struggle to capture long-distance dependencies in the NER process. This paper describes the establishment of a Chinese corpus in the field of edible fungus and introduces an NER method for edible fungus information based on XLNet and conditional random fields (CRFs). Our approach combines an iterated dilated convolutional neural network (IDCNN) with a CRF. First, leveraging the XLNet model as the foundation, an IDCNN layer is introduced. This layer addresses the limited capacity to capture features across utterances by extending the receptive field of the convolutional kernel. The output of the IDCNN layer is input to the CRF layer, which mitigates any labeling logic errors, resulting in the globally optimal labels for the NER task relating to edible fungus. Experimental results show that the precision achieved by the proposed model reaches 0.971, with a recall of 0.986 and an F1-score of 0.979. The proposed model outperforms existing approaches in terms of these evaluation metrics, effectively recognizing entities related to edible fungus information and offering methodological support for the construction of knowledge graphs.

Comparative Study of Metaphors and Metonymies of Russian and Chinese Related to Word "Hand"

Language comparison is a powerful tool in linguistics. It not only deepens the understanding of the linguistic characteristics of the language being compared, but also helps to analyze the characteristics of cross-language and cross-cultural human thinking. The purpose of this thesis is to supplement the deficiencies of previous studies, conduct a comprehensive comparison of "hand" metaphors and metonyms in Chinese and Russian, use corpus methods to estimate their generality in the two languages, and analyze the Russian language reflected by these metaphors and metonyms. and characteristics of Chinese culture. Literature research method: This paper sorts out the metaphor and metonymy of "hand" in Russian and Chinese on the basis of researching metaphors, idioms, word-formation literature, dictionaries and other materials. This study concludes that first, compared with Russian words for "hand", metonymy and metaphor are more frequently used in Chinese. Second, there are more "hand" word metaphors than metonymy in the Chinese corpus, but there is no big difference in the number of metonymy and metaphor in the Russian corpus. Third, the metonymy of "hand" words in Chinese and Russian is relatively similar, and the difference between metaphors is more obvious, which shows that culture has a greater influence on metaphors, such as the characteristics of the showman's culture, gestures and other non-verbal communication habits, etc.

Nonlinear Regularization Decoding Method for Speech Recognition.

Existing end-to-end speech recognition methods typically employ hybrid decoders based on CTC and Transformer. However, the issue of error accumulation in these hybrid decoders hinders further improvements in accuracy. Additionally, most existing models are built upon Transformer architecture, which tends to be complex and unfriendly to small datasets. Hence, we propose a Nonlinear Regularization Decoding Method for Speech Recognition. Firstly, we introduce the nonlinear Transformer decoder, breaking away from traditional left-to-right or right-to-left decoding orders and enabling associations between any characters, mitigating the limitations of Transformer architectures on small datasets. Secondly, we propose a novel regularization attention module to optimize the attention score matrix, reducing the impact of early errors on later outputs. Finally, we introduce the tiny model to address the challenge of overly large model parameters. The experimental results indicate that our model demonstrates good performance. Compared to the baseline, our model achieves recognition improvements of 0.12%, 0.54%, 0.51%, and 1.2% on the Aishell1, Primewords, Free ST Chinese Corpus, and Common Voice 16.1 datasets of Uyghur, respectively.

Metricizing policy texts: Comprehensive dataset on China’s Agri-policy intensity spanning 1982–2023

Due to the lack of direct assessment metrics, existing studies on the intensity of agricultural policies often utilize indicators such as Gross Domestic Product (GDP) of agriculture or the quantity of agricultural policies as measures. Optimizing methods for analyzing the intensity of agricultural policies will significantly impact parameter selection in agricultural policy research and the evaluation of policy effectiveness. In this study, we constructed a Chinese Agricultural Policy Corpus using agricultural policies released by various governmental agencies at the national level in China from 1982 to April 2023. We quantified the values of agricultural domain terms in the corpus and evaluated the intensity of each agricultural policy document. The validation results of this study indicate a strong correlation between the intensity of agricultural policies and agricultural GDP. The trend in agricultural GDP changes lags behind policy intensity by 2.5 years (at a 95% confidence level), thus validating the rationality of our constructed corpus, agricultural policy scoring dataset, and methodology.

Language Dissemination Paths and Modes Aided by Computer Technology

The expansion of technology and computer science, as well as advancements in language instruction and learning methodologies, has enabled computer-assisted language learning technologies to tackle this challenge. In the field of Chinese learning, a few language learning computerized systems in the country and abroad concentrate mainly on language, grammar acquisition only have one or two assessment indicators as basis of evaluation, that definite functional flaws provide a general assessment to learners' pronunciation. In this manuscript, Language Dissemination Paths and Modes Aided by Computer Technology (LDPM-QICCNN-KOA) are proposed. The input data are collected from Chinese Corpus dataset. Then the data is given into unscented trainable kalman filter for preprocessing the input data. Then the preprocessed data are provided to QICCNN for Language Dissemination. In general, the based Quantum-inspired Complex Convolutional Neural Network doesn’t express adapting optimization approaches to determine optimal parameters to ensure exact identification. Hence, KOA utilized to enhance Quantum-inspired Complex Convolutional Neural Network, which accurately done the Language Dissemination Paths and Modes. The proposed LDPM-QICCNN-KOA method is executed on python. Then performance of proposed technique is analyzed with other existing methods. The proposed technique attains 26.36%, 20.69% and 35.29% higher accuracy; 19.23%, 23.56%, and 33.96% higher F1-Score; 26.28%, 31.26%, and 19.66% higher precision when comparing with the existing methods such as research on network oral English teaching system depend on machine learning (LDPM-DBN), nonlinear network speech recognition structure in deep learning algorithm (LDPM-DNN), research on open oral English scoring system depend on neural network (LDPM-BPNN).