Is this you? Claim this profile to correct it, add a bio and choose the work people see first.
Claim this profileWorks265 from public data
- Learning Deep Transformer Models for Machine Translation782
It is claimed that a truly deep Transformer model can surpass the Transformer-Big counterpart by 1) proper use of layer normalization and 2) a novel way of passing the combination of previous layers to the next.
- A Survey on LLM-Generated Text Detection: Necessity, Methods, and Future Directions200
This survey examines various LLM-generated text detection paradigms, shedding light on challenges like out-of-distribution problems, potential attacks, real-world data issues and ineffective evaluation frameworks, and outlines intriguing directions for future research in LLM-generated text detection to advance responsible artificial intelligence (AI).
- Modeling Localness for Self-Attention Networks196
This work cast localness modeling as a learnable Gaussian bias, which indicates the central and scope of the local region to be paid more attention in self-attention networks, to maintain the strength of capturing long distance dependencies while enhance the ability of capturing short-range dependencies.
- Is ChatGPT a Highly Fluent Grammatical Error Correction System? A Comprehensive Evaluation181
This work designs zero-shot chain-of-thought (CoT) and few-shot CoT settings using in-context learning for ChatGPT to showcase its capabilities in GEC and demonstrates that it has excellent error detection capabilities and can freely correct errors to make the corrected sentences very fluent.
- UM-Corpus: A Large English-Chinese Parallel Corpus for Statistical Machine Translation138
The acquisition of a large scale and high quality parallel corpora for English and Chinese for Statistical Machine Translation (SMT) is described, designed to embrace eight different domains.
- Convolutional Self-Attention Networks137
Novel convolutional self-attention networks are proposed, which offer SANs the abilities to strengthen dependencies among neighboring elements, and model the interaction between features extracted by multiple attention heads.
- Norm-Based Curriculum Learning for Neural Machine Translation135
This paper aims to improve the efficiency of training an NMT by introducing a novel norm-based curriculum learning method that uses the norm (aka length or module) of a word embedding as a measure of the difficulty of the sentence, the competence of the model, and the weight of the sentences.
- Recent Progress in the Discovery and Design of Antimicrobial Peptides Using Traditional Machine Learning and Deep Learning125
The biology background of AMP is introduced, various feature encoding methods used to represent the features of peptide sequences are presented, and the most popular DL techniques are explained and highlighted, to survey the latest AMP prediction methods achieved by DL approaches.
- Context-Aware Self-Attention Networks124
This work proposes to contextualize the transformations of the query and key layers, which are used to calculates the relevance between elements, and leverage the internal representations that embed both global and deep contexts, thus avoid relying on external resources.
- Uncertainty-Aware Curriculum Learning for Neural Machine Translation111
This work proposes uncertainty-aware curriculum learning, motivated by the intuition that the higher the uncertainty in a translation pair, the more complex and rarer the information it contains; and the end of the decline in model uncertainty indicates the completeness of current training stage.
- Knowledge modeling via contextualized representations for LSTM-based personalized exercise recommendation101
Two approaches that use a contextualized representation of KCs are proposed, one with a content-based approach and another with a Long Short Term Memory (LSTM) network plus a personalization mechanism.
- Latent Space Chain-of-Embedding Enables Output-free LLM Self-Evaluation71
This research track proposes the Chain-of-Embedding in the latent space to enable LLMs to perform output-free self-evaluation and provides interesting insights into LLM response correctness from the perspective of hidden state changes inside LLMs.
- Context-aware Self-Attention Networks for Natural Language Processing67
This work casts local context modeling as a learnable Gaussian bias, which indicates the central and scope of the local region to be paid more attention and leverage the internal representations that embed sentence-level information as the global context to contextualize the transformations of query and key layers.
- 64
- Is Your Model Really A Good Math Reasoner? Evaluating Mathematical Reasoning with Checklist63
It is argued that if a model really understands a problem, it should be robustly applied across a diverse array of tasks and MathCheck, a well-designed checklist for testing task generalization and reasoning robustness, is introduced, as well as an automatic tool to generate checklists efficiently.
- LEPOR: A Robust Evaluation Metric for Machine Translation with Augmented Factors59
Experiments show that this novel metric yields the state-of-the-art correlation with human judgments compared with classic metrics BLEU, TER, Meteor-1.3 and two latest metrics (AMBER and MP4IBM1), which proves it a robust one by employing a feature-rich and model-independent approach.
- CultureVLM: Characterizing and Improving Cultural Understanding of Vision-Language Models for over 100 Countries57
This paper constructs CultureVerse, a large-scale multimodal benchmark covering 19, 682 cultural concepts, 188 countries/regions, 15 cultural concepts, and 3 question types, and proposes CultureVLM, a series of VLMs fine-tuned on the authors' dataset to achieve significant performance improvement in cultural understanding.
- 56
- Rejuvenating Low-Frequency Words: Making the Most of Parallel Data in Non-Autoregressive Translation55
This work directly exposes the raw data into NAT by leveraging pretraining to rejuvenate more alignments for low-frequency target words, and combines these complementary approaches as a new training strategy for further boosting NAT performance.
- Self-Paced Learning for Neural Machine Translation54
Experimental results over multiple translation tasks demonstrate that the proposed NMT model yields better performance than strong baselines and those models trained with human-designed curricula on both translation quality and convergence speed.
- Stock Market Prediction from WSJ: Text Mining via Sparse Matrix Factorization54
The problem of predicting directional movements of stock prices based on news articles is revisited and a unified latent space model is proposed to characterize the "co-movements" between stock prices and news articles.
- Understanding Aha Moments: from External Observations to Internal Mechanisms53
This work systematically study "aha moments" in LRMs, and demonstrates that the "aha moment" is externally manifested in a more frequent use of anthropomorphic tones for self-reflection and an adaptive adjustment of uncertainty based on problem difficulty.
- Salute the Classic: Revisiting Challenges of Machine Translation in the Age of Large Language Models51
This study revisits six core challenges of Neural Machine Translation, offering insights into their ongoing relevance in the context of advanced Large Language Models, and identifies three new challenges in LLM-based translation: inference efficiency, translation of low-resource languages during pretraining, and human-aligned evaluation.
- Meta-Curriculum Learning for Domain Adaptation in Neural Machine Translation51
Experimental results on 10 different low-resource domains show that meta-curriculum learning can improve the translation performance of both familiar and unfamiliar domains.
- Leveraging Local and Global Patterns for Self-Attention Networks51
The extensive analyses verify that the two types of contexts are complementary to each other, and the proposed hybrid attention mechanism gives highly effective improvements in their integration.
- 50
- Guiding Variational Response Generator to Exploit Persona49
This paper proposes to adopt the personality-related characteristics of human conversations into variational response generators, by designing a specific conditional variational autoencoder based deep model with two new regularization terms employed to the loss function, to guide the optimization towards the direction of generating both persona-aware and relevant responses.
- UniTE: Unified Translation Evaluation47
This paper proposes monotonic regional attention to control the interaction among input segments, and unified pretraining to better adapt multi-task training and can universally surpass various state-of-the-art or winner methods across tasks.
- Progressive Multi-Granularity Training for Non-Autoregressive Translation47
This study empirically shows that NAT models are prone to learn fine-grained lower-mode knowledge, such as words and phrases, compared with sentences, and proposes progressive multi-granularity training for NAT, which improves phrase translation accuracy and model reordering ability, resulting in better translation quality against strong NAT baselines.
- Who Wrote This? The Key to Zero-Shot LLM-Generated Text Detection Is GECScore45
This paper proposes a simple yet effective black-box zero-shot detection approach based on the observation that, from the perspective of LLMs, human-written texts typically contain more grammatical errors than LLM-generated texts.
- A Paradigm Shift: The Future of Machine Translation Lies with Large Language Models45
This paper provides an overview of the significant enhancements in MT that are influenced by LLMs and advocate for their pivotal role in upcoming MT research and implementations, and highlights several new MT directions, emphasizing the benefits of LLMs in scenarios such as Long-Document Translation, Stylized Translation, and Interactive Translation.
- Graph-based Semi-Supervised Model for Joint Chinese Word Segmentation and Part-of-Speech Tagging45
Empirical results reveal that the proposed graph-based semisupervised joint model of Chinese word segmentation and part-of-speech tagging can yield better results than the supervised baselines and other competitive semi-supervised CRFs in this task.
- Can LLMs Learn Uncertainty on Their Own? Expressing Uncertainty Effectively in A Self-Training Manner41
An uncertainty-aware instruction tuning (UaIT) method is proposed, aligning LLMs’ perception with the probabilistic uncertainty of the generation to instigate a heightened awareness of self-uncertainty in LLMs, enabling them to express uncertainty more effectively.
- Challenges of Neural Machine Translation for Short Texts41
Two challenges in NMT for STs associated with translation error types above, respectively are summarized: the imbalanced length distribution in training set intensifies model inference calibration over STs, leading to more over-translation cases on STs; and the lack of contextual information forces NMT to have higher data uncertainty on short sentences, and thus NMT model is troubled by considerable mistranslation errors.
- DelTA: An Online Document-Level Translation Agent Based on Multi-Level Memory38
Results indicate that DelTA significantly outperforms strong baselines in terms of translation consistency and quality across four open/closed-source LLMs and two representative document translation datasets, achieving an increase in consistency scores by up to 4.58 percentage points and in COMET scores by up to 3.16 points on average.
- Shared-Private Bilingual Word Embeddings for Neural Machine Translation38
Experiments demonstrate that the proposed shared-private bilingual word embeddings provide a significant performance boost over the strong baselines with dramatically fewer model parameters.
- Chinese Named Entity Recognition with Conditional Random Fields in the Light of Chinese Characteristics38
This paper firstly conducts a study of the Chinese characteristics and makes a discussion of the different feature sets; then a promising comparison result is shown with the optimized features and concise model.
- Assessing the Ability of Self-Attention Networks to Learn Word Order36
Experimental results reveal that: 1) SAN trained on word reordering detection indeed has difficulty learning the positional information even with the position embedding; and 2) SANtrained on machine translation learns better positional information than its RNN counterpart, in which position embeddedding plays a critical role.
- TemplateGEC: Improving Grammatical Error Correction with Detection Template34
A novel method, TemplateGEC, is proposed, which capitalizes on the capabilities of both Seq2Edit and Sequ2Seq frameworks in error detection and correction respectively, and reveals the potential of the method in performing human-in-the-loop GEC.
- Towards Bidirectional Hierarchical Representations for Attention-based Neural Machine Translation33
A hierarchical attentional neural translation model which focuses on enhancing source-side hierarchical representations by covering both local and global semantic information using a bidirectional tree-based encoder to alleviate the out-of-vocabulary (OOV) problem.
- A Systematic Comparison of Data Selection Criteria for SMT Domain Adaptation33
An in-depth analysis of three different sentence selection techniques for statistical machine translation (SMT) and achieves the goal to consistently boost the overall translation performance that can ensure optimal quality of a real-life SMT system.
- Document Graph for Neural Machine Translation32
Experiments on various NMT benchmarks, including IWSLT English–French, Chinese-English, WMT English–German and Opensubtitle English–Russian, demonstrate that using document graphs can significantly improve the translation quality.
- 30
- Constructing Better Classifier Ensemble Based on Weighted Accuracy and Diversity Measure30
A novel measure used to evaluate the quality of the classifier ensemble, assisting in the ensemble selection task, and the empirical results demonstrate that the WAD measure is superior to others in most cases.
- On the Complementarity between Pre-Training and Back-Translation for Neural Machine Translation29
Experimental results show that PT and BT are nicely complementary to each other, establishing state-of-the-art performances on the WMT16 English-Romanian and English-Russian benchmarks.
- Improving tree-based neural machine translation with dynamic lexicalized dependency encoding28
A lexicalized dependency model is proposed, in which the source-side lexical representations are learned in a head-dependent fashion following a dependency graph, and a variant recurrent neural network (RNN) is proposed to jointly consider the long-distance dependencies and the sequential information of words.
- On the Copying Behaviors of Pre-Training for Neural Machine Translation26
Extensive experiments on both in-domain and out-of-domain benchmarks show that the copying penalty method consistently improves translation performance by controlling copying behaviors for pre-training based NMT models.
- Quality Estimation for Machine Translation Using the Joint Method of Evaluation Criteria and Statistical Modeling26
The designed methods of this paper yielded promising results especially the statistical models of CRF and NB, which achieved the highest F-score 0.8297 in binary classification of Task 2.
- 26
- 24
- Embedding Trajectory for Out-of-Distribution Detection in Mathematical Reasoning23
A trajectory-based method TV score is proposed, which uses trajectory volatility for OOD detection in mathematical reasoning and outperforms all traditional algorithms on GLMs under mathematical reasoning scenarios and can be extended to more applications with high-density features in output spaces, such as multiple-choice questions.
- 23
- Unsupervised Neural Dialect Translation with Commonality and Diversity Modeling22
This paper leverages pivot-private embedding, layer coordination, as well as parameter sharing to sufficiently model commonality and diversity among source and target, ranging from lexical, through syntactic, to semantic levels, in unsupervised translation models merely accessing to monolingual data.
- Factored Statistical Machine Translation for Grammatical Error Correction22
Focusing on all possible error types in a real-life environment, a factored statistical machine translation (SMT) model is proposed for grammatical error correction (GEC), and empirical results show an improvement over a baseline phrase-based SMT model.
- 21
- 21
- TransGEC: Improving Grammatical Error Correction with Translationese20
Experimental results on the CoNLL14 and BEA19 English, NLPCC18 Chinese, Falko-MERLIN German, and RULEC-GEC Russian GEC benchmarks show that the proposed translationese approach consistently improves correction accuracy over strong baselines, and is helpful for overcoming mainstream correction difficulties.
- A flexible example annotation schema19
Work is presented on the task of constructing an example base from a given bilingual corpus based on the annotation schema of Translation Corresponding Tree, which represents the syntactic structure of source language sentence and more importantly is the facility to specify the correspondences between string and the representation tree.
- Bilingual recursive neural network based data selection for statistical machine translation18
This paper proposes bilingual semi-supervised recursive neural network data selection methods to differentiate domain-relevant data from out-domain data and shows that the proposed methods outperform the state-of-the-art data selection approaches.
- AutoCBT: An Autonomous Multi-agent Framework for Cognitive Behavioral Therapy in Psychological Counseling17
Experimental results indicate that AutoCBT can provide higher-quality automated psychological counseling services and incorporate dynamic routing and supervisory mechanisms inspired by real psychological counseling to construct a CBT-oriented autonomous multi-agent framework, demonstrating its general applicability.
- iSentenizer- μ : Multilingual Sentence Boundary Detection Model17
iSentenizer-μ, under the incremental learning framework, is adaptable to text of different topics and Roman-alphabet languages, by merging new data into existing model to learn the new knowledge incrementally by revision instead of retraining.
- UM-Checker: A Hybrid System for English Grammatical Error Correction17
The NLP 2 CT Grammatical Error Detection and Correction system for the CoNLL 2013 shared task, with a focus on the errors of article or determiner, noun number, preposition, verb form and subject-verb agreement, achieves the 5 and 3 rd F1 scores on official test set among all 17 participating teams based on goldstandard edits before and after revision.
- Improving Grammatical Error Correction with Multimodal Feature Integration16
A novel framework that integrates both speech and text features to enhance GEC is proposed that achieves significant improvements over strong baselines and achieves a new state-of-the-art result on the Falko-MERLIN test set.
- GuoFeng: A Benchmark for Zero Pronoun Recovery and Translation16
A benchmark testset for target evaluation on Chinese-English ZP translation is proposed, which covers five challenging genres, which reveal different characteristics of ZPs for comprehensive evaluation.
- 15
- kNN-TL: k-Nearest-Neighbor Transfer Learning for Low-Resource Neural Machine Translation15
A k-Nearest-Neighbor Transfer Learning (kNN-TL) approach for low-resource NMT, which leverages the parent knowledge throughout the entire developing process of the child model, and which outperforms strong baselines on four low- resource translation tasks.
- User Retention: A Causal Approach with Triple Task Modeling15
This is the first work to model user retention by estimating the revisit rate from a causal perspective, and the superiority of UR-IPW over previous methods is demonstrated.
- HeTROPY: Explainable learning diagnostics via heterogeneous maximum-entropy and multi-spatial knowledge representation15
A novel heterogeneous attention interpreter with a maximum entropy regularizer on top of a student knowledge model to achieve explainable learning diagnostics and is adoptable to various types of e-learning systems to assist teachers to gain insights into student learning states and diagnose learning problems.
- Chinese Named Entity Recognition with Graph-based Semi-supervised Learning Model15
The experiment shows that the unlabeled corpus can enhance the state-of-theart conditional random field (CRF) learning model and has potential to improve the tagging accuracy even though the margin is a little weak and not satisfying in current experiments.
- Toward Better Chinese Word Segmentation for SMT via Bilingual Constraints15
The experiments on a Chinese-to-English machine translation task reveal that the proposed model can bring positive segmentation effects to translation quality.
- Machine translation using constraint-based synchronous grammar15
A synchronous grammar based on the formalism of context-free grammar developed by generalizing the first component of production that models the source text, which was used as the basis for building a machine translation system for Portuguese to Chinese translation.
- 14
- A Description of Tunable Machine Translation Evaluation Systems in WMT13 Metrics Task14
This paper is to describe the machine translation evaluation systems used for participation in the WMT13 shared Metrics Task and two automatic MT evaluation systems nLEPOR_baseline and LEPOR_v3.1.
- CRFs-Based Chinese Word Segmentation for Micro-Blog with Small-Scale Data14
The proposed model has been implemented for the Second CIPS-SIGHAN Joint Conference on Chinese Language Processing Bakeoff (Bakeoff-2012) and achieves a very high Fmeasure of 93.38% within the test set of 5,000 micro-blog sentences.
- 13
- RoBLEURT Submission for the WMT2021 Metrics Task13
Experimental results show that the submission to Shared Metrics Task: RoBLEURT (Robustly Optimizing the training of BLEURT) reaches state-of-the-art correlations with the WMT2020 human annotations upon 8 out of 10 to-English language pairs.
- 13
- Phrase Tagset Mapping for French and English Treebanks and Its Application in Machine Translation Evaluation13
To bridge the phrase level tag sets of multilingual treebanks, a phrase mapping between the French Treebank and the English Penn Treebank is designed and one of the potential applications of this mapping work is explored in the machine translation evaluation task.
- Co-regularizing character-based and word-based models for semi-supervised Chinese word segmentation13
This paper presents a semi-supervised Chinese word segmentation (CWS) approach that co-regularizes character-based and word-based models and reveals that the model results in better gains over the state-of-the-art semi- supervised models reported in the literature.
- Machine Translation Based on Constraint-Based Synchronous Grammar13
It turns out that such grammar is more expressive to model the translational equivalences of parallel texts for machine translation, and in this paper, the use of CSG as a basis for building a machine translation (MT) system for Portuguese to Chinese translation is proposed.
- Rethinking the Exploitation of Monolingual Data for Low-Resource Neural Machine Translation12
This study carries out a systematic investigation of the effects of techniques such as back-translation and self-supervised objectives on linguistic properties through the use of probing tasks, including source language comprehension, bilingual word alignment, and translation fluency on bitexts of varying sizes.
- Revisiting Commonsense Reasoning in Machine Translation: Training, Evaluation and Challenge12
This paper confirms the effectiveness of incorporating pretrained knowledge into NMT models and subsequently utilizing these models as robust testbeds for investigating CR in NMT and proposes a novel entity-aware evaluation method that takes into account both the NMT candidate and important entities in the candidate, which is more aligned with human judgement.
- A Relationship: Word Alignment, Phrase Table, and Translation Quality12
The deduced formula is feasible, which not only can be used to predict the size of the phrase table, but also can be a valuable reference for investigating the relationship between the translation performance and phrase tables based on different links of word alignment.
- A Joint Chinese Named Entity Recognition and Disambiguation System12
An integrated approach for named entity recognition and disambiguation in Chinese using Named entity recognition (NER), entity linking and document clustering models based on hidden Markov model (HMM).
- Yu Sheng: Human-in-Loop Classical Chinese Poetry Generation System11
Yu Sheng is a web-based poetry generation system that is featured a human-in-loop generation framework, providing various customization options for users with different backgrounds to engage in the process of poetry composition.
- Can LMs Generalize to Future Data? An Empirical Analysis on Text Summarization11
This work proposes TempoSum, a novel benchmark that contains data samples from 2010 to 2022, to understand the temporal generalization ability of abstractive summarization models and shows that parametric knowledge stored in summarization Models significantly affects the faithfulness of the generated summaries on future data.
- Human-in-the-loop Machine Translation with Large Language Model11
A human-in-the-loop pipeline that guides LLMs to produce customized outputs with revision instructions is proposed that demonstrates the effectiveness of the pipeline in tailoring in-domain translations and improving translation performance compared to direct translation instructions.
- 11
- 11
- A Multi-Agent Learning Paradigm for Medical Data Mining Diagnostic Workbench11
The philosophy behind i +DiaMAS is to assist physicians in diagnosing new case objectively and reliably by providing practical diagnostic rules that acquired from heterogeneous and distributed historical medical data, where their new discoveries and correct diagnostic results can benefit other physicians.
- 10
- 10
- Machine Translation Evaluation: A Survey10
The state-of-the-art machine translation (MT) evaluation survey that contains both manual and automatic evaluation methods is introduced and the different classifications from manual to automatic evaluation measures are introduced.
- Edit Distance: A New Data Selection Criterion for Domain Adaptation in SMT10
Based on the investigation of the state-ofthe-art similarity metrics, edit distance is proposed as a new data selection criterion for this topic and it is shown that the proposed approach achieves a significant improvement over the baseline system as well as the best rival model.
- Machine translation based on translation corresponding tree structure10
A representation schema called translation corresponding tree (TCT) has been applied to a Portu- guese to Chinese example-based machine translation system and can be effectively represented with this annotation schema and organized in the bilingual knowledge database or example base.
- Beyond Simple Fusion: Adaptive Gated Fusion for Robust Multimodal Sentiment Analysis9
Visualization analysis of feature representations demonstrates that AGFN enhances generalization by learning from a broader feature distribution, achieved by reducing the correlation between feature location and prediction error, thereby decreasing reliance on specific locations and creating more robust multimodal feature representations.
- 9
- Cohesive Relations in Text Comprehension and Production: An Exploratory Study Comparing Translation and Post-Editing9
An exploratory study of data related to one text included in MS13, contained in CRITT Translation Process Research Database, with a view to investigating the impact of type of cohesive chain on cognitive effort in Portuguese-Chinese translation and post-editing tasks.
- Combining Domain Adaptation Approaches for Medical Text Translation9
A number of simple and effective techniques to adapt statistical machine translation systems in the medical domain and these systems achieve the best BLEU scores for Czech-English, EnglishGerman, French-English language pairs and the second best Blemish scores for reminding pairs are explored.
- Unsupervised Chunking Based on Graph Propagation from Bilingual Corpus9
A novel approach for unsupervised shallow parsing model trained on the unannotated Chinese text of parallel Chinese-English corpus, with exploitation of graph-based label propagation for bilingual knowledge transfer, along with an application of using the projected labels as features in un supervised model.
- Unsupervised Quality Estimation Model for English to German Translation and Its Application in Extensive Supervised Evaluation9
An unsupervised MT evaluation metric using universal part-of-speech tagset without relying on reference translations is proposed, which shows that the designed methods yield higher correlation scores with human judgments.
- Lexicon expansion for latent variable grammars9
A graph-based lexicon expansion approach that builds k-nearest-neighbor similarity graphs over the words of labeled and unlabeled data, for propagating lexical emission probabilities and is very effective in parsing out-of-vocabulary (OOV) words.
- 9
- Systematic Processing of Long Sentences in Rule Based Portuguese-Chinese Machine Translation9
An approach to parse and translate long sentences efficiently in application to Rule based Portuguese-Chinese Machine Translation and Constraint Synchronous Grammar is used to model both source and target languages simultaneously at the parsing stage to reduce ambiguities and the parsing efficiency.
- CSAT: A Chinese Segmentation and Tagging Module Based on the Interpolated Probabilistic Model9
An integrated application with segmentation and tagging ability has been studied and implemented and it is used as a preprocessing module in the Chinese to Portuguesemachine translation system.
- 9
- 8
- Let's Focus on Neuron: Neuron-Level Supervised Fine-tuning for Large Language Model8
Neuron-Level Fine-Tuning (NeFT) is introduced, a novel approach that refines the granularity of parameter training down to the individual neuron, enabling more precise and computationally efficient model updates.
- 8
- 8
- Test-time Adaptation for Machine Translation Evaluation by Uncertainty Minimization8
This paper aims to address the inference bias of neural metrics through uncertainty minimization during test time, without requiring additional data, by employing the prediction uncertainty of the current data as a signal to update a small fraction of parameters during testTime.
- Difficulty-Aware Machine Translation Evaluation8
A novel difficulty-aware MT evaluation metric is proposed, expanding the evaluation dimension by taking translation difficulty into consideration, and shows that the proposed method outperforms commonly used MT metrics in terms of human correlation.
- 8
- Domain Adaptation for Medical Text Translation using Web Resources8
This paper describes adapting statistical machine translation systems to medical domain using in-domain and general-domain data as well as webcrawled in- domain resources and proposes an alternative filtering approach to clean the crawled data and to further optimize the domain-specific SMT system.
- A Universal Phrase Tagset for Multilingual Treebanks8
A refined universal phrase tagset that contains 9 commonly used phrase categories is designed that can generally reduce the costs in the parsing models and even improve the parsing accuracy.
- Time series for blind biosignal classification model8
A blind biosignal classification model (B(2)SC Model), which uses a dynamic time warping algorithm with clustering to discover the similarity between two biosignals, and consequently classifies disease without prior knowledge of the source signal type.
- 8
- Path Drift in Large Reasoning Models: How First-Person Commitments Override Safety7
The findings highlight the need for trajectory-level alignment oversight in long-form reasoning beyond token-level alignment, and introduce a three-stage Path Drift Induction Framework comprising cognitive load amplification, self-role priming, and condition chain hijacking.
- 7
- 7
- 7
- 7
- Probing Semantic Alignment, Lexical Invariance, and Syntactic Influence in LLM Metaphor Processing6
A diagnostic analysis that examines the limits of behavioral evidence by probing three complementary dimensions: semantic attribute alignment, lexical invariance, and syntactic sensitivity reveals that LLM-generated interpretations can exhibit semantic drift relative to reference attributes.
- 6
- 6
- 6
- 6
- A Supportive Attribute-Assisted Discretization Model for Medical Classification6
Empirical experiments were conducted to compare the classification accuracy achieved by several prestigious classifiers with SAAD and other state-of-the-art discretization approaches, demonstrating the effectiveness and usefulness of the proposed approach in enhancing the diagnostic accuracy.
- iCPE: A Hybrid Data Selection Model for SMT Domain Adaptation6
A hybrid data selection model named iCPE is proposed, which combines three state-of-the-art similarity metrics: Cosine tf-idf, Perplexity and Edit distance at both corpus level and model level, which performs better over the baseline system trained on entire data as well as the best rival method.
- A Maximum Entropy (ME) Based Translation Model for Chinese Characters Conversion6
This paper proposes a Chinese conversion model based on Maximum Entropy (ME), a Machine Learning (ML) technique that is comparable to the state of the art conversion system.
- 6
- 6
- How well can state-of-the-art machine translation systems render a 16th-century Chinese novel?5
This study evaluates the performance of state-of-the-art machine translation systems in rendering Journey to the West, a culturally rich 16th-century Chinese novel, into Portuguese and finds that NovelTrans-J consistently outperforms all other participants, particularly in terms of cultural appropriateness.
- Rethinking Prompt-based Debiasing in Large Language Models5
This study systematically analyzed bias benchmarks using the BBQ and StereoSet benchmarks on both open-source models as well as commercial GPT model to highlight a potential "false prosperity" in prompt-base efforts and emphasize the need to rethink bias metrics to ensure truly trustworthy AI.
- 5
- 5
- 5
- 5
- 5
- A Study of Chinese Word Segmentation Based on the Characteristics of Chinese5
An augmented and optimized set of features is developed for Chinese word segmentation based on the analysis of the characteristics of Chinese and several categories of ambiguities in Chinese to explore potential solutions.
- 5
- ToolRM: Towards Agentic Tool-Use Reward Modeling4
This work introduces ToolRM, a family of lightweight reward models tailored for general tool-use scenarios that generalizes to broader critique tasks, including Best-of-N sampling and self-correction, and proposes a novel pipeline that constructs high-quality pairwise preference data using rule-based scoring and multidimensional sampling.
- Can ChatGPT Really Understand Modern Chinese Poetry?4
A comprehensive framework for evaluating ChatGPT's understanding of modern poetry and establishes a solid foundation for future research on LLMs and their application to poetry-related tasks is introduced.
- Towards Fine-Grained Information: Identifying the Type and Location of Translation Errors4
This paper proposes Fine-Grained Translation Error Detection (FG-TED) task, aiming at identifying both the position and the type of translation errors on given source-hypothesis sentence pairs, and builds an FG-TED model, which can identify both error type and position concurrently.
- Attention Mechanism with Energy-Friendly Operations4
A novel attention model is built by replacing multiplications with either selective operations or additions by reaching the conclusion that the energy costs of several energy-friendly operations are far less than their multiplication counterparts.
- 4
- 4
- Effective Hypotheses Re-ranking Model in Statistical Machine Translation4
A novel approach to improve the final translation result by dynamically augmenting the translation scores of hypotheses that derived from the N-best translation candidates is presented.
- Experiments with POS-based restructuring and alignment-based reordering for statistical machine translation4
The methods which are based on the part-of-speech (POS) and auto alignment information to improve the quality of machine translation result and the word alignment and use an alignment-based reordering method to improved the alignment.
- An Improvement in Cross-Language Document Retrieval Based on Statistical Models4
A proposed method integrated with three statistical models including Translation model, Query generation model and Document retrieval model for cross-language document retrieval that can efficiently solve the problem of translation ambiguity and query expansion for disambiguation which are critical in Cross-Language Information Retrieval.
- 4
- Query Translation for Cross-Language Information Retrieval by Parsing Constraint Synchronous Grammar4
- Application of Translation Corresponding Tree (TCT) Annotation Schema for Chinese to Portuguese Machine Translation4
In order to resolve the weakness of the Structure Based EBMT system and linguistic problems between Chinese and Portuguese, Tang and Wong propose a new Portuguese to Chinese Machine Translation method based on a novel technology called Translation Corresponding Tree.
- CoCoReviewBench: A Completeness- and Correctness-Oriented Benchmark for AI Reviewers3
CoReviewBench is introduced, which curates 3,900 papers from ICLR and NeurIPS to enable reliable and fine-grained evaluation of AI reviewers, and shows that AI reviewers remain limited in correctness and are prone to hallucinations.
- DetectRL-X: Towards Reliable Multilingual and Real-World LLM-Generated Text Detection3
DetectRL-X is introduced, a comprehensive multilingual benchmark designed to evaluate advanced detectors across 8 dimensions and analyzed how domains, generators, attack strategies, text length, and refinement operations influence performance in different languages, underscoring DetectRL-X as an effective benchmark for strengthening multilingual and language-specific detectors.
- Neuron-Aware Data Selection In Instruction Tuning For Large Language Models3
NAIT evaluates the impact of IT data on LLMs performance by analyzing the similarity of neuron activation patterns between the IT dataset and the target domain capability, and captures neuron activation patterns from in-domain datasets of target domain capabilities to construct reusable and transferable neuron activation features.
- 3
- Domain-Aware k-Nearest-Neighbor Knowledge Distillation for Machine Translation3
A novel domain-aware k NN-KD method, which filters out domain-relevant neighborhood knowledge for learning in the distillation process, and exclusively utilizes the neighborhood knowledge of the original model, eliminating the need for establishing any additional datas-tores.
- NovelTrans: System for WMT24 Discourse-Level Literary Translation3
NovelTrans from NLP 2 CT and DeepTranx for the WMT24 Discourse-Level Literary Translation Task in Chinese-English, Chinese-German, and Chinese-Russian language pairs under un-constrained conditions achieves an xCOMET score of 79.14, higher than performing a direct chapter-level translation on the authors' dataset.
- A benchmark dataset and evaluation methodology for Chinese zero pronoun translation3
A benchmark testset and evaluation metric for target evaluation on Chinese ZP translation, which covers five challenging genres, which reveal different characteristics of ZPs for comprehensive evaluation.
- 3
- 3
- Towards the Generation of Bilingual Chinese-English Multi-word Expressions from Large Scale Parallel Corpora: An Experimental Approach3
A preliminary attempt at successful efforts to mine bilingual MWEs from 300+k Chinese-English patents is reported here.
- Data Selection via Semi-supervised Recursive Autoencoders for SMT Domain Adaptation3
The proposed data selection model is trained to capture the domain specific features and used for detecting sentences, which are relevant to a specific domain, from a large general-domain corpus.
- Chinese Unknown Word Recognition for PCFG-LA Parsing3
A novel method, using graph-based semisupervised learning (SSL), is proposed to improve the syntax parsing of unknown words and propagate lexical emission probabilities to unknown words by building the similarity graphs over the words of labeled and unlabeled data.
- TQDL: Integrated Models for Cross-Language Document Retrieval3
An integrated approach for Cross-Language Information Retrieval (CLIR), which integrated with four statistical models: Translation model, Query generation model, Document retrieval model and Length Filter model, which can efficiently solve the problem of translation ambiguity and query expansion for disambiguation.
- 3
- A Template Based Hybrid Model for Chinese Personal Name Disambiguation3
The template makes use of the features of personal role such as discriminating personal name (nickname, stage name), together with the specific context of most frequent words, personal name nearest words named entities, date and time that are effective for this disambiguation task.
- Chinese Tagging Based on Maximum Entropy Model3
Their models for the evaulation are based on the maximum entropy approach, and the authors concentrated on the word segmentation task for the bakeoff and the best official results on all the corpora for this task are 0.9083 F-score on CITYU, 0.8985 on CKIP,0.9077 on CTB, 0,8995 on NCC and 0,9146 on SXU.
- EvoPolicyGym: Evaluating Autonomous Policy Evolution in Interactive Environments2
This work introduces Autonomous Policy Evolution, a controlled evaluation setting in which a harness-model agent repeatedly edits an executable policy system under a fixed interaction budget, and instantiates this setting in EvoPolicyGym, a benchmark built from compact interactive RL environments that evaluates how agents iteratively improve explored policies.
- Chain-of-Procedure: Hierarchical Visual-Language Reasoning for Procedural QA2
This work presents Chain-of-Procedure (CoP), a hierarchical reasoning framework that first retrieves relevant instructions using visual cues, then performs step refinement through semantic decomposition, and finally generates the next step.
- SelfMem: Self-Optimizing Memory for AI Agents2
SelfMem provides an environment with memory tools and feedback signals that allow the agent to explore, evaluate, and refine its own memory strategy, following the principle of teaching an agent to fish rather than giving it a fish.
- 2
- 2
- 2
- 2
- 2
- 2
- 2
- MoNMT: Modularly Leveraging Monolingual and Bilingual Knowledge for Neural Machine Translation2
This paper proposes a modular strategy that facilitates the cooperation of monolingual and bilingual knowledge in translation tasks, while avoiding the issue of catastrophic forgetting and exhibiting superior model generalization and robustness.
- 2
- Machine Translation2
This book will not make you feel bored when reading, so you may not need to get it at once in a day, and machine translation really offers what everybody wants.
- Augmented Parsing of Unknown Word by Graph-Based Semi-Supervised Learning2
A novel method using graph-based semi-supervised learning (SSL) to improve the syntax parsing of unknown words by contributing 2.28% and 1.72% absolute improvement on the accuracy of POS tagging and syntax parsing for Penn Chinese Treebank respectively.
- Automatic Machine Translation Evaluation with Part-of-Speech Information2
This paper explores the evaluation only using Part-of-Speech information of the words, which means the method is based only on the consilience of the POS strings of the hypothesis translation and reference, which acts as the similar function with the synonyms in addition to its syntactic or morphological behaviour of the lexical item in question.
- 2
- A Simplified Chinese Parser with Factored Model2
A factored model to parse the Simplified Chinese using TCT Treebank as experimental data and demonstrates that the head propagation table improves the parsing performance.
- Adapting Multilingual Parsing Models to Sinica Treebank2
This paper gives the first result of adapting existing multilingual parsing models to the Sinica Treebank and shows that the parsing accuracy can be improved by the suggested approach.
- 2
- 2
- 2
- Discovery of the Latent Supportive Relevance in Medical Data Mining2
It is believed that, among an attributes space, each attribute may have certain relevance with another attribute, therefore to take the attributes relevant correlation into consideration in data pre-processing is a vital factor for the ideal pre- processing methods.
- Integration of Named Entity Information for Chinese Word Segmentation Based on Maximum Entropy2
The model is constructed based on Maximum Entropy to handle the UNE identification as tagging problem and shows that the overall accuracy of the segmentation is improved after integrating theUNE identification module into the word segmenter.
- Overcoming Data Sparseness Problem in Statistical Corpus Based Sense Disambiguation2
The problem of data sparseness is considered as a common problem for Statistical Corpus based Sense Disambiguation approaches and large amounts of data are required in the corpus to guarantee that all senses of an ambiguous word are presented.
- From Scenes to Elements: Multi-Granularity Evidence Retrieval for Verifiable Multimodal RAG1
This work introduces GranuVistaVQA, a multimodal benchmark featuring real-world landmarks with element-level annotations across multiple viewpoints, capturing the partial observation challenge where individual images contain only subsets of entities.
- Seeing the Poem: Image-Semantic Detection of AI-Generated Modern Chinese Poetry with MLLMs1
Experimental results demonstrate that the LLM detectors based on the proposed image-semantic guided poetry detection method outperform baseline detectors based on plain text, and even surpass the best-performing traditional detector, RoBERTa.
- Loong: A Human-Like Long Document Translation Agent with Observe-and-Act Adaptive Context Selection1
A human-like long document translation agent called Loong, which leverages a 3E memory module (Essence-Exemplar-Entity) to store summaries, sentence pairs, and entity records as historical context and performs deep reasoning to adaptively identify the optimal context for translation guidance.
- EmoS: A High-Fidelity Multimodal Benchmark for Fine-grained Streaming Emotional Understanding1
Empirical results show that fine-tuning MLLMs (multimodal large language models) on EmoS yields significant gains over zero-shot baselines, laying the foundation for the training and evaluation of future emotion recognition models and empathy models.
- G-IdiomAlign: A Gloss-Pivoted Benchmark for Cross-Lingual Idiom Alignment1
G-IdiomAlign is presented, a gloss-pivoted benchmark where each idiom is anchored by an English gloss from Wiktionary, and a high-confidence reference alignment set is constructed for reproducible evaluation.
- 1
- 1
- 1
- Umplc: the first longitudinal learner corpus of Portuguese1
The University of Macau Portuguese Learner Corpus (UMPLC), an annotated longitudinal learner corpus of Portuguese, is introduced, which enables a wide range of linguistic studies, and contributes to the development of pedagogical applications in the context of teaching and learning of Portuguese as a second language.
- 1
- 1
- 1
- 1
- 1
- 1
- Regenerator Placement in Survivable Optical Networks Using Deep Tensor Neural Network1
The feasibility of using deep tensor neural network to optimize regenerator placement in optical networks is demonstrated and brings significant improvements over the results achieved by the currently deploying techniques and can be used in the real-life networks.
- 1
- 1
- Benefiting feature selection by the discovery of false irrelevant attributes1
This paper proposes a feature selection methodology from a new angle that selects not only the relevant features, but also targeting at the potentially useful false irrelevant attributes by measuring their supportive importance to other attributes.
- iTagger: Part-of-Speech Tagging Based on SBCB Learning Algorithm1
A novel POS tagging system, iTagger, which is developed based on Selecting Base Classifiers on Bagging (SBCB) learning algorithm and compared against two state-of-the-art tagging methods, Hidden Markov Model (HMM) and Maximum Entropy.
- An Experimental Platform for Cross-Language Document Retrieval1
This paper presents a proposed Cross-Language Document Retrieval experimental platform integrated with preprocessing of training data, document translation, query generation, document retrieval and precision evaluation modules that was effective and achieved very good performance.
- Influence of Part-of-Speech and Phrasal Category Universal Tag-set in Tree-to-Tree Translation Models1
A unified tagset for all languages at Part-of-Speech and Phrasal Category level in tree-to-tree models is presented.
- 1
- Unearth the Hidden Supportive Information for an Intelligent Medical Diagnostic System1
An intelligent diagnostic supporting system, which automatically extracts useful knowledge from massive medical data to support real medical diagnosis and the empirical results on several real-life datasets from UCI repository demonstrate the goodness of the diagnostic system.
- 1
- Digital camera based visual-to-textual translation system1
A novel technique of detecting and extracting text regions on camera-captured and video images for realizing more accurate text recognition is proposed and the higher accuracy of camera-based visual translation is witnessed from the obtained testing results.
- From Texts to Scores: Tracing the Emergence of Essay Quality Representations in Large Language Models–
Evidence is found that LLMs encode structured representations related to essay quality and new insights are offered into the interpretability of LLM-based AES systems.
- –
- Worlds Within Words: Translating Culture in Ancient Chinese Texts with Multi-Agent Coordination–
This work forms CLW translation as a selective explicitation task and proposes MACAT, a dynamically identifies culturally salient phrases and injects concise explanatory knowledge when necessary, a quality-aware reranking module for candidate selection and a multi-round evaluation agent that assesses translations across terminological precision, readability, fidelity, cultural preservation, and cultural explicitation.
- –
- –
- CLIF: Concept-Level Influence Functions for Transparent Bottleneck Models–
A novel approach using influence functions to enhance interpretability in NLP models at both the sample and concept levels and demonstrates that model performance can be restored to baseline levels without retraining, confirming the value of influence functions for efficient data debugging.
- –
- Beyond the sycophantic vs. cold divide: towards flexible, emotionally intelligent agents to foster healthier human-AI interactions–
To support users’ healthy emotional regulation, flexible, emotionally aware AI systems are needed.
- –
- MC-PDD: Masked Corpus-Level Pretraining Data Detection for Black-Box Large Language Models–
M masked Corpus-level Pretraining Data Detection (MC-PDD) is introduced, a novel method inspired by the masked language modeling paradigm that enables practical applications such as model auditing and data copyright verification using only standard API access.
- –
- –
- AHAlogy: Agent-Supported In-Class Analogical Learning for Primary School Students' Divergent Thinking–
AHAlogy, a curriculum-aligned large language model (LLM) agent designed for in-class use within a principle-guided analogical STEAM curriculum, augmented the analogical curriculum without restructuring the teacher-led lesson.
- CHAKT: A Robust Knowledge Tracing Model with Uncertainty-Aware and Hierarchical Attention Mechanisms–
- –
- –
- Privacy-preserving RAG via multi-agent semantic rewriting: Achieving confidentiality without compromising contextual fidelity–
A multi-agent framework that sanitizes retrieved content through semantic rewriting by employing three specialized agents for privacy extraction, semantic analysis, and reconstruction, which collaboratively removes sensitive identifiers while preserving the semantic core is proposed.
- –
- –
- –
- –
- BioPro: Towards Difference-Aware Gender Fairness for Vision-Language Models–
This work proposes BioPro (Bias Orthogonal Projection), an entirely training-free framework that effectively reduces gender bias in neutral cases while maintaining gender faithfulness in explicit ones, providing a promising direction toward achieving selective fairness in VLMs.
- –
- –
- –
- –
- –
- –
- –
- Learning from "Silly" Questions Improves Large Language Models, But Only Slightly–
Surprisingly, results indicate that rules can significantly improve model performance in certain tasks, while potentially diminishing performance on others, and highlights the importance of considering task diversity and rule applicability when constructing SFT datasets to achieve more comprehensive performance improvements.
- Annotation System for the Construction of Synchronous Grammar Tree Alignment Relationships–
A system for annotating and constructing synchronous grammar tree structures in a semi-automatic way is proposed and provides import functions in obtaining monolingual skeletal bracketing syntactic tree and Translation Corresponding Tree structures for the creation of synchronous rules in order to have wider applicability.
- –
- Rules Design in Word Segmentation of Chinese Micro-Blog–
An HMM based segmentation model integrated with a pre and post correction module for Chinese micro-blog texts is proposed and can achieve an F-measure of 90.98% on test set of 5,000 sentences.
- TransLite - Development of Lightweight Machine Translation System based on Constraint Synchronous Grammar–
PCT TransLite is proposed, a lightweight system that only relies on parsing Constraint Synchronous Grammar (CSG) in the whole translation process, and has been already used in constructing CSG rules for real MT systems, for education purposes, and got positive feedback.
- –
- –
- Portuguese Chinese machine aided translation system–
A computer aided translation system based on multiple translation engines, including MT and Translation Memory, with a focus on the languages of Portuguese and Chinese is presented and it has a centralized knowledge base for better management.
- –
- Cross Sentence Alignment for Structurally Dissimilar Corpus Based on Singular Value Decomposition–
The proposed approach is based on Singular Value Decomposition techniques and similarity measurement, which covers the problem in aligning structurally dissimilar corpus and enhances the accuracy of the alignment result.
- Treebank Creation Based on TCT Annotation Schema–
An automatic tool to create Treebank in TCT annotation schema is proposed to enhance tagging performance and ensure sentence tree correctness in Example Based Machine Translation research.
- Cross-Level Sentence Alignment–
A new model for sentence alignment system of structurally different languages such as Chinese and Portuguese is described, which applies the maximum entropy model to align the named entities without perform the word segmentation for Chinese.
- –
- –
- –
- –
- A New Adaptive Self-Organizing Map–
A new model based on Kohenen SOM is proposed that has integrated a series of evolutionary working mechanisms of neurons that can conquer the problems of instability in competitive learning.
Publication data from OpenAlex, with missing venues and authors filled in from Crossref; citation counts are the higher of OpenAlex and Semantic Scholar, last synced 2026-10-14. One-sentence summaries under some papers are written by Semantic Scholar’s model. Citation counts may be lower than on Google Scholar, which indexes more sources.
Report an error
Wrong papers, two people merged into one, or a profile that should not be here? Tell us and we will fix or hide it.