Pre-trained BERT Model Research Articles

Research on Semantic Similarity of Short Text Based on Bert and Time Warping Distance

The research on semantic similarity of short text plays an important role in machine translation, emotion analysis, information retrieval and other AI business applications. However, according to existing short text similarity research, the characteristics of ambiguous vocabularies are difficult to be effectively analyzed, the solution of the problem caused by words order needs to be further optimized as well. This paper proposes a short text semantic similarity calculation method that combines BERT and time warping distance algorithm, in order to solve the problem of vocabulary ambiguity. The model first uses the pre trained Bert model to extract the semantic features of the short text from the whole level, and obtains a 768 dimensional short text feature vector. Then, it transforms the extracted feature vector into a point sequence in space, uses the CTW algorithm to calculate the time warping distance between the curves connected by the point sequence, and finally uses the weight function designed by the analysis, according to the smaller the time warpage distance is, the higher the degree of small similarity is, to calculate the similarity between short texts. The experimental results show that this model can mine the feature information of ambiguous words, and calculate the similarity of short texts with lexical ambiguity effectively. Compared with other models, it can distinguish the semantic features of ambiguous words more accurately.

Knowledge Based Deep Inception Model for Web Page Classification

Web Page Classification is decisive for information retrieval and management task and plays an imperative role for natural language processing (NLP) problems in web engineering. Traditional machine learning algorithms excerpt covet features from web pages whereas deep leaning algorithms crave features as the network goes deeper. Pre-trained models such as BERT attains remarkable achievement for text classification and continue to show state-ofthe-art results. Knowledge Graphs can provide rich structured factual information for better language modelling and representation. In this study, we proposed an ensemble Knowledge Based Deep Inception (KBDI) approachfor web page classification by learning bidirectional contextual representation using pre-trained BERT incorporating Knowledge Graph embeddings and fine-tune the target task by applying Deep Inception network utilizing parallel multi-scale semantics. Proposed ensemble evaluates the efficacy of fusing domain specific knowledge embeddings with the pre-trained BERT model. Experimental interpretation exhibit that the proposed BERT fused KBDI model outperforms benchmark baselines and achieve better performance in contrast to other conventional approaches evaluated on web page classification datasets.

Fine-grained spatial information extraction in radiology as two-turn question answering

ObjectivesRadiology reports contain important clinical information that can be used to automatically construct fine-grained labels for applications requiring deep phenotyping. We propose a two-turn question answering (QA) method based on a transformer language model, BERT, for extracting detailed spatial information from radiology reports. We aim to demonstrate the advantage that a multi-turn QA framework provides over sequence-based methods for extracting fine-grained information. MethodsOur proposed method identifies spatial and descriptor information by answering queries given a radiology report text. We frame the extraction problem such that all the main radiology entities (e.g., finding, device, anatomy) and the spatial trigger terms (denoting the presence of a spatial relation between finding/device and anatomical location) are identified in the first turn. In the subsequent turn, various other contextual information that acts as important spatial roles with respect to a spatial trigger term are extracted along with identifying the spatial and other descriptor terms qualifying a radiological entity. The queries are constructed using separate templates for the two turns and we employ two query variations in the second turn. ResultsWhen compared to the best-reported work on this task using a traditional sequence tagging method, the two-turn QA model exceeds its performance on every component. This includes promising improvements of 12, 13, and 12 points in the average F1 scores for identifying the spatial triggers, Figure, and Ground frame elements, respectively. DiscussionOur experiments suggest that incorporating domain knowledge in the query (a general description about a frame element) helps in obtaining better results for some of the spatial and descriptive frame elements, especially in the case of the clinical pre-trained BERT model. We further highlight that the two-turn QA approach fits well for extracting information for complex schema where the objective is to identify all the frame elements linked to each spatial trigger and finding/device/anatomy entity, thereby enabling the extraction of more comprehensive information in the radiology domain. ConclusionExtracting fine-grained spatial information from text in the form of answering natural language queries holds potential in achieving better results when compared to more standard sequence labeling-based approaches.

Modelling Drug-Target Binding Affinity using a BERT based Graph Neural network.

Understanding the interactions between novel drugs and target proteins is fundamentally important in disease research as discovering drug-protein interactions can be an exceptionally time-consuming and expensive process. Alternatively, this process can be simulated using modern deep learning methods that have the potential of utilising vast quantities of data to reduce the cost and time required to provide accurate predictions. We seek to leverage a set of BERT-style models that have been pre-trained on vast quantities of both protein and drug data. The encodings produced by each model are then utilised as node representations for a graph convolutional neural network, which in turn are used to model the interactions without the need to simultaneously fine-tune both protein and drug BERT models to the task. We evaluate the performance of our approach on two drug-target interaction datasets that were previously used as benchmarks in recent work.Our results significantly improve upon a vanilla BERT baseline approach as well as the former state-of-the-art methods for each task dataset. Our approach builds upon past work in two key areas; firstly, we take full advantage of two large pre-trained BERT models that provide improved representations of task-relevant properties of both drugs and proteins. Secondly, inspired by work in natural language processing that investigates how linguistic structure is represented in such models, we perform interpretability analyses that allow us to locate functionally-relevant areas of interest within each drug and protein. By modelling the drug-target interactions as a graph as opposed to a set of isolated interactions, we demonstrate the benefits of combining large pre-trained models and a graph neural network to make state-of-the-art predictions on drug-target binding affinity.

Aspect-Based Sentiment Analysis in Hindi Language by Ensembling Pre-Trained mBERT Models

Sentiment Analysis is becoming an essential task for academics, as well as for commercial companies. However, most current approaches only identify the overall polarity of a sentence, instead of the polarity of each aspect mentioned in the sentence. Aspect-Based Sentiment Analysis (ABSA) identifies the aspects within the given sentence, and the sentiment that was expressed for each aspect. Recently, the use of pre-trained models such as BERT has achieved state-of-the-art results in the field of natural language processing. In this paper, we propose two ensemble models based on multilingual-BERT, namely, mBERT-E-MV and mBERT-E-AS. Using different methods, we construct an auxiliary sentence from this aspect and convert the ABSA problem to a sentence-pair classification task. We then fine-tune different pre-trained BERT models and ensemble them for a final prediction based on the proposed model; we achieve new, state-of-the-art results for datasets belonging to different domains in the Hindi language.

Sentence Compression Using BERT and Graph Convolutional Networks

Sentence compression is a natural language-processing task that produces a short paraphrase of an input sentence by deleting words from the input sentence while ensuring grammatical correctness and preserving meaningful core information. This study introduces a graph convolutional network (GCN) into a sentence compression task to encode syntactic information, such as dependency trees. As we upgrade the GCN to activate a directed edge, the compression model with the GCN layers can distinguish between parent and child nodes in a dependency tree when aggregating adjacent nodes. Furthermore, by increasing the number of GCN layers, the model can gradually collect high-order information of a dependency tree when propagating node information through the layers. We implement a sentence compression model for Korean and English, respectively. This model consists of three components: pre-trained BERT model, GCN layers, and a scoring layer. The scoring layer can determine whether a word should remain in a compressed sentence by relying on the word vector containing contextual and syntactic information encoded by BERT and GCN layers. To train and evaluate the proposed model, we used the Google sentence compression dataset for English and a Korean sentence compression corpus containing about 140,000 sentence pairs for Korean. The experimental results demonstrate that the proposed model achieves state-of-the-art performance for English. To the best of our knowledge, this sentence compression model based on the deep learning model trained with a large-scale corpus is the first attempt for Korean.

Automated Essay Scoring Using Recurrence over BERT (RoBERT)

초록·키워드 목차 오류제보하기 This study aimed to build a system that could automate students" English essay evaluation by using Recurrence over BERT (RoBERT), a state-of-the-art deep learning model. English essay evaluation is inherently time-consuming. It may reflect teacher bias. English teachers are usually burdened with the task of evaluating many essays in a short period of time. Automated essay scoring (AES) can solve these problems. It has the advantage of being able to evaluate essays in a short time and without bias. In this paper, the RoBERT model was trained and evaluated on Essay Set #8 of the Automated Student Assessment Prize (ASAP) dataset. The 5-fold cross validation evaluation method was used for fair comparison with the previously suggested AES models. As a result, the RoBERT model showed the highest agreement with the human raters’ resolved scores in 5 out of 6 trait scores than the previous evaluation models. The advantage of it is that it can use the pre-trained BERT model and deal with long inputs, overcoming the input size limit of the BERT model. It was confirmed that the RoBERT model works well for trait-specific evaluation of long essays. Thus, the RoBERT model can be used as an auxiliary means to automate the evaluation of students" essays and reduce the excessive work of English teachers. #automated essay scoring #trait-specific essay scoring #essay evaluation #Recurrence over BERT (RoBERT) #hierarchical transformers Ⅰ. INTRODUCTIONⅡ. LITERATURE REVIEWⅢ. MethodologyREFERENCES

Bridge inspection named entity recognition via BERT and lexicon augmented machine reading comprehension neural model

As an important data source in the field of bridge management, bridge inspection reports contain large-scale fine-grained data, including information on bridge members and structural defects. However, due to insufficient research on automatic information extraction in this field, valuable bridge inspection information has not been fully utilized. Particularly, for Chinese bridge inspection entities, which involve domain-specific vocabularies and have obvious nesting characteristics, most of the existing named entity recognition (NER) solutions are not suitable. To address this problem, this paper proposes a novel lexicon augmented machine reading comprehension-based NER neural model for identifying flat and nested entities from Chinese bridge inspection text. The proposed model uses the bridge inspection text and predefined question queries as input to enhance the ability of contextual feature representation and to integrate prior knowledge. Based on the character-level features encoded by the pre-trained BERT model, bigram embeddings and weighted lexicon features are further combined into a context representation. Then, the bidirectional long short-term memory neural network is used to extract sequence features before predicting the spans of named entities. The proposed model is verified by the Chinese bridge inspection named entity corpus. The experimental results show that the proposed model outperforms other mainstream NER models on the bridge inspection corpus. The proposed model not only provides a basis for automatic bridge inspection information extraction but also supports the downstream tasks such as knowledge graph construction and question answering systems.

Relation classification via BERT with piecewise convolution and focal loss.

Recent relation extraction models’ architecture are evolved from the shallow neural networks to natural language model, such as convolutional neural networks or recurrent neural networks to Bert. However, these methods did not consider the semantic information in the sequence or the distance dependence problem, the internal semantic information may contain the useful knowledge which can help relation classification. Focus on these problems, this paper proposed a BERT-based relation classification method. Compare with the existing Bert-based architecture, the proposed model can obtain the internal semantic information between entity pair and solve the distance semantic dependence better. The pre-trained BERT model after fine tuning is used in this paper to abstract the semantic representation of sequence, then adopt the piecewise convolution to obtain semantic information which influence the extraction results. Compare with the existing methods, the proposed method can achieve a better accuracy on relational extraction task because of the internal semantic information extracted in the sequence. While, the generalization ability is still a problem that cannot be ignored, and the numbers of the relationships are difference between different categories. In this paper, the focal loss function is adopted to solve this problem by assigning a heavy weight to less number or hard classify categories. Finally, comparing with the existing methods, the F1 metric of the proposed method can reach a superior result 89.95% on the SemEval-2010 Task 8 dataset.

P-078. Profile of cesarean delivery care in hypertensive pregnant women

As an important data source in the field of bridge management, bridge inspection reports contain large-scale fine-grained data, including information on bridge members and structural defects. However, due to insufficient research on automatic information extraction in this field, valuable bridge inspection information has not been fully utilized. Particularly, for Chinese bridge inspection entities, which involve domain-specific vocabularies and have obvious nesting characteristics, most of the existing named entity recognition (NER) solutions are not suitable. To address this problem, this paper proposes a novel lexicon augmented machine reading comprehension-based NER neural model for identifying flat and nested entities from Chinese bridge inspection text. The proposed model uses the bridge inspection text and predefined question queries as input to enhance the ability of contextual feature representation and to integrate prior knowledge. Based on the character-level features encoded by the pre-trained BERT model, bigram embeddings and weighted lexicon features are further combined into a context representation. Then, the bidirectional long short-term memory neural network is used to extract sequence features before predicting the spans of named entities. The proposed model is verified by the Chinese bridge inspection named entity corpus. The experimental results show that the proposed model outperforms other mainstream NER models on the bridge inspection corpus. The proposed model not only provides a basis for automatic bridge inspection information extraction but also supports the downstream tasks such as knowledge graph construction and question answering systems.

Transformers-sklearn: a toolkit for medical language understanding with transformer-based models

BackgroundTransformer is an attention-based architecture proven the state-of-the-art model in natural language processing (NLP). To reduce the difficulty of beginning to use transformer-based models in medical language understanding and expand the capability of the scikit-learn toolkit in deep learning, we proposed an easy to learn Python toolkit named transformers-sklearn. By wrapping the interfaces of transformers in only three functions (i.e., fit, score, and predict), transformers-sklearn combines the advantages of the transformers and scikit-learn toolkits.MethodsIn transformers-sklearn, three Python classes were implemented, namely, BERTologyClassifier for the classification task, BERTologyNERClassifier for the named entity recognition (NER) task, and BERTologyRegressor for the regression task. Each class contains three methods, i.e., fit for fine-tuning transformer-based models with the training dataset, score for evaluating the performance of the fine-tuned model, and predict for predicting the labels of the test dataset. transformers-sklearn is a user-friendly toolkit that (1) Is customizable via a few parameters (e.g., model_name_or_path and model_type), (2) Supports multilingual NLP tasks, and (3) Requires less coding. The input data format is automatically generated by transformers-sklearn with the annotated corpus. Newcomers only need to prepare the dataset. The model framework and training methods are predefined in transformers-sklearn.ResultsWe collected four open-source medical language datasets, including TrialClassification for Chinese medical trial text multi label classification, BC5CDR for English biomedical text name entity recognition, DiabetesNER for Chinese diabetes entity recognition and BIOSSES for English biomedical sentence similarity estimation.In the four medical NLP tasks, the average code size of our script is 45 lines/task, which is one-sixth the size of transformers’ script. The experimental results show that transformers-sklearn based on pretrained BERT models achieved macro F1 scores of 0.8225, 0.8703 and 0.6908, respectively, on the TrialClassification, BC5CDR and DiabetesNER tasks and a Pearson correlation of 0.8260 on the BIOSSES task, which is consistent with the results of transformers.ConclusionsThe proposed toolkit could help newcomers address medical language understanding tasks using the scikit-learn coding style easily. The code and tutorials of transformers-sklearn are available at https://doi.org/10.5281/zenodo.4453803. In future, more medical language understanding tasks will be supported to improve the applications of transformers_sklearn.

PSSM-Distil: Protein Secondary Structure Prediction (PSSP) on Low-Quality PSSM by Knowledge Distillation with Contrastive Learning

Protein secondary structure prediction (PSSP) is an essential task in computational biology. To achieve the accurate PSSP, the general and vital feature engineering is to use multiple sequence alignment (MSA) for Position-Specific Scoring Matrix (PSSM) extraction. However, when only low-quality PSSM can be obtained due to poor sequence homology, previous PSSP accuracy (merely around 65%) is far from practical usage for subsequent tasks. In this paper, we propose a novel PSSM-Distil framework for PSSP on low-quality PSSM, which not only enhances the PSSM feature at a lower level but also aligns the feature distribution at a higher level. In practice, the PSSM-Distil first exploits the proteins with high-quality PSSM to achieve a teacher network for PSSP in a full-supervised way. Under the guidance of the teacher network, the low-quality PSSM and corresponding student network with low discriminating capacity are effectively resolved by feature enhancement through EnhanceNet and distribution alignment through knowledge distillation with contrastive learning. Further, our PSSM-Distil supports the input from a pre-trained protein sequence language BERT model to provide auxiliary information, which is designed to address the extremely low-quality PSSM cases, i.e., no homologous sequence. Extensive experiments demonstrate the proposed PSSM-Distil outperforms state-of-the-art models on PSSP by 6% on average and nearly 8% in extremely low-quality cases on public benchmarks, BC40 and CB513.

Listen, Understand and Translate: Triple Supervision Decouples End-to-end Speech-to-text Translation

An end-to-end speech-to-text translation (ST) takes audio in a source language and outputs the text in a target language. Existing methods are limited by the amount of parallel corpus. Can we build a system to fully utilize signals in a parallel ST corpus? We are inspired by human understanding system which is composed of auditory perception and cognitive processing. In this paper, we propose Listen-Understand-Translate, (LUT), a unified framework with triple supervision signals to decouple the end-to-end speech-to-text translation task. LUT is able to guide the acoustic encoder to extract as much information from the auditory input. In addition, LUT utilizes a pre-trained BERT model to enforce the upper encoder to produce as much semantic information as possible, without extra data. We perform experiments on a diverse set of speech translation benchmarks, including Librispeech English-French, IWSLT English-German and TED English-Chinese. Our results demonstrate LUT achieves the state-of-the-art performance, outperforming previous methods. The code is available at https://github.com/dqqcasia/st.

Span identification and technique classification of propaganda in news articles

Propaganda is a rhetorical technique designed to serve a specific topic, which is often used purposefully in news article to achieve our intended purpose because of its specific psychological effect. Therefore, it is significant to be clear where and what propaganda techniques are used in the news for people to understand its theme efficiently during our daily lives. Recently, some relevant researches are proposed for propaganda detection but unsatisfactorily. As a result, detection of propaganda techniques in news articles is badly in need of research. In this paper, we are going to introduce our systems for detection of propaganda techniques in news articles, which is split into two tasks, Span Identification and Technique Classification. For these two tasks, we design a system based on the popular pretrained BERT model, respectively. Furthermore, we adopt the over-sampling and EDA strategies, propose a sentence-level feature concatenating method in our systems. Experiments on the dataset of about 550 news articles offered by SEMEVAL show that our systems perform state-of-the-art.

Evaluation of the Coherence of Polish Texts Using Neural Network Models

Coherence evaluation of texts falls into a category of natural language processing tasks. The evaluation of texts’ coherence implies the estimation of their semantic and logical integrity; such a feature of a text can be utilized during the solving of multidisciplinary tasks (SEO analysis, medicine area, detection of fake texts, etc.). In this paper, different state-of-the-art coherence evaluation methods based on machine learning models have been analyzed. The investigation of the effectiveness of different methods for the coherence estimation of Polish texts has been performed. The impact of text’s features on the output coherence value has been analyzed using different approaches of a semantic similarity graph. Two neural networks based on LSTM layers and a pre-trained BERT model correspondingly have been designed and trained for the coherence estimation of input texts. The results obtained may indicate that both lexical and semantic components should be taken into account during the coherence evaluation of Polish documents; moreover, it is advisable to analyze corresponding documents in a sentence-by-sentence manner taking into account word order. According to the retrieved accuracy of the proposed neural networks, it can be concluded that suggested models may be used in order to solve typical coherence estimation tasks for a Polish corpus.

Forward Context-Aware Clickbait Tweet Identification System

Clickbait is an elusive challenge with the prevalence of social media such as Facebook and Twitter that misleads the readers while clicking on headlines. Limited annotated data makes it onerous to design an accurate clickbait identification system. The authors address this problem by purposing deep learning-based architecture with external knowledge which trains on social media post and descriptions. The pre-trained ELMO and BERT model obtains the sentence level contextual feature as knowledge; moreover, the LSTM layer helps to prevail the word level contextual feature. Training has done at different experiments (model with EMLO, model with BERT) with different regularization techniques such as dropout, early stopping, and finetuning. Forward context-aware clickbait tweet identification system (FCCTI) with BERT finetuning and model with ELMO using glove pre-trained embedding is the best model and achieves a clickbait identification accuracy of 0.847, improving on the previous baseline for this task.

Sentiment Classification Algorithm Based on the Cascade of BERT Model and Adaptive Sentiment Dictionary

The mobile social network contains a large amount of information in a form of commentary. Effective analysis of the sentiment in the comments would help improve the recommendations in the mobile network. With the development of well‐performing pretrained language models, the performance of sentiment classification task based on deep learning has seen new breakthroughs in the past decade. However, deep learning models suffer from poor interpretability, making it difficult to integrate sentiment knowledge into the model. This paper proposes a sentiment classification model based on the cascade of the BERT model and the adaptive sentiment dictionary. First, the pretrained BERT model is used to fine‐tune with the training corpus, and the probability of sentiment classification in different categories is obtained through the softmax layer. Next, to allow a more effective comparison between the probabilities for the two classes, a nonlinearity is introduced in a form of positive‐negative probability ratio, using the rule method based on sentiment dictionary to deal with the probability ratio below the threshold. This method of cascading the pretrained model and the semantic rules of the sentiment dictionary allows to utilize the advantages of both models. Different sized Chnsenticorp data sets are used to train the proposed model. Experimental results show that the Dict‐BERT model is better than the BERT‐only model, especially when the training set is relatively small. The improvement is obvious with the accuracy increase of 0.8%.

Adaptive Adapters: An Efficient Way to Incorporate BERT Into Neural Machine Translation

Large-scale pre-trained language models (e.g., BERT) have attracted great attention in recent years. It is straightforward to fine-tune them on natural language understanding tasks such as text classification, however, effectively and efficiently incorporating them into natural language generation tasks such as neural machine translation remains a challenging problem. In this paper, we integrate two pre-trained BERT models from the source and target language domains into a sequence-to-sequence model by introducing light-weight adapter modules. The adapters are inserted between BERT layers and tuned on downstream tasks, while the parameters of BERT models are fixed during fine-tuning. As pre-trained language models are usually very deep, inserting adapters into all layers will result in a considerable scale of new parameters. To deal with this problem, we introduce latent variables to decide whether using adapters or not in each layer, which are learned during fine-tuning. In this way, the model is able to automatically determine which adapters to use, therefore hugely promoting the parameter efficiency and decoding speed. We evaluate the proposed framework on various neural machine translation tasks. Equipped with parallel sequence decoding, our model consistently outperforms autoregressive baselines while reducing the inference latency by half. With automatic adapter selection, the proposed model further achieves 20% speedup while still outperforming autoregressive baselines. When applied to autoregressive decoding, the proposed model can also achieve comparable performance with the state-of-the-art baseline models.

Extracting relational facts based on hybrid Syntax-Guided transformer and pointer network

Joint extraction of entities and relations from unstructured text is an essential step in constructing a knowledge base. However, relational facts in these texts are often complicated, where most of them contain overlapping triplets, making the joint extraction task still challenging. This paper proposes a novel Sequence-to-Sequence (Seq2Seq) framework to handle the overlapping issue, which models the triplet extraction as a sequence generation task. Specifically, a unique cascade structure is proposed to connect transformer and pointer network to extract entities and relations jointly. By this means, sequences can be generated in triplet-level and it speeds up the decoding process. Besides, a syntax-guided encoder is applied to integrate the sentence’s syntax structure into the transformer encoder explicitly, which helps the encoder pay more accurate attention to the syntax-related words. Extensive experiments were conducted on three public datasets, named NYT24, NYT29, and WebNLG, and the results show the validity of this model by comparing with various baselines. In addition, a pre-trained BERT model is also employed as the encoder. Then it comes up to excellent performance that the F1 scores on the three datasets surpass the strongest baseline by 5.7%, 5.6%, and 4.4%.

Mol‐BERT: An Effective Molecular Representation with BERT for Molecular Property Prediction

Molecular property prediction is an essential task in drug discovery. Most computational approaches with deep learning techniques either focus on designing novel molecular representation or combining with some advanced models together. However, researchers pay fewer attention to the potential benefits in massive unlabeled molecular data (e.g., ZINC). This task becomes increasingly challenging owing to the limitation of the scale of labeled data. Motivated by the recent advancements of pretrained models in natural language processing, the drug molecule can be naturally viewed as language to some extent. In this paper, we investigate how to develop the pretrained model BERT to extract useful molecular substructure information for molecular property prediction. We present a novel end‐to‐end deep learning framework, named Mol‐BERT, that combines an effective molecular representation with pretrained BERT model tailored for molecular property prediction. Specifically, a large‐scale prediction BERT model is pretrained to generate the embedding of molecular substructures, by using four million unlabeled drug SMILES (i.e., ZINC 15 and ChEMBL 27). Then, the pretrained BERT model can be fine‐tuned on various molecular property prediction tasks. To examine the performance of our proposed Mol‐BERT, we conduct several experiments on 4 widely used molecular datasets. In comparison to the traditional and state‐of‐the‐art baselines, the results illustrate that our proposed Mol‐BERT can outperform the current sequence‐based methods and achieve at least 2% improvement on ROC‐AUC score on Tox21, SIDER, and ClinTox dataset.