Performance Of Relation Extraction Research Articles

Combined Networks with Multi-level Attention for Distantly-Supervised Relation Extraction

Distantly-supervised relation extraction is expected to extract relational facts from very large corpora. However, it is inevitably accompanied by the problem of mislabeling, which affects the performance of relation extraction. To obtain richer information from sentences and make full use of the information of entity pairs, we propose the combined networks containing piecewise convolutional neural networks (PCNN) and bidirectional gated recurrent units (BiGRU) with multi-level attention. In this model, PCNN and BiGRU are utilized to obtain local features and sequence information. Then the multi-level attention is proposed to extract the most correlated information, which can affect the assignment of sentence attention weight. And it reduces the influence of wrong labeled instances indirectly. The experimental results show that our method can enhance the representation power of the network and prove the effectiveness of our model compared with several baseline methods.

Improving Neural Relation Extraction with Positive and Unlabeled Learning

We present a novel approach to improve the performance of distant supervision relation extraction with Positive and Unlabeled (PU) Learning. This approach first applies reinforcement learning to decide whether a sentence is positive to a given relation, and then positive and unlabeled bags are constructed. In contrast to most previous studies, which mainly use selected positive instances only, we make full use of unlabeled instances and propose two new representations for positive and unlabeled bags. These two representations are then combined in an appropriate way to make bag-level prediction. Experimental results on a widely used real-world dataset demonstrate that this new approach indeed achieves significant and consistent improvements as compared to several competitive baselines.

Relation Extraction Exploiting Full Dependency Forests

Dependency syntax has long been recognized as a crucial source of features for relation extraction. Previous work considers 1-best trees produced by a parser during preprocessing. However, error propagation from the out-of-domain parser may impact the relation extraction performance. We propose to leverage full dependency forests for this task, where a full dependency forest encodes all possible trees. Such representations of full dependency forests provide a differentiable connection between a parser and a relation extraction model, and thus we are also able to study adjusting the parser parameters based on end-task loss. Experiments on three datasets show that full dependency forests and parser adjustment give significant improvements over carefully designed baselines, showing state-of-the-art or competitive performances on biomedical or newswire benchmarks.

Curriculum learning for distant supervision relation extraction

Relation extraction under distant supervision leverages the existing knowledge base to label data automatically, thus greatly reduced the consumption of human labors. Although distant supervision is an efficient method to obtain a large amount of labeled data, the training dataset labeled by distant supervision suffers from noise problem resulting in poor generalization ability of the relation extractor. To alleviate the noise problem, we propose a novel relation extraction method based on curriculum learning. Curriculum learning is utilized to guide the training process of relation extractor, specifically through the predefined curriculum-driven mentor network. Mentor network can dynamically adjust the weights of sentences during training, giving lower weights to noisy sentences and higher weights to truly labeled sentences. Relation extractor and mentor network are trained collaboratively to optimize joint objective. The experimental results show that the proposed method can improve the generalization ability of relation extractor in a noisy environment and obtains better performance for relation extraction.

Multiple features for clinical relation extraction: A machine learning approach.

Relation extraction aims to discover relational facts about entity mentions from plain texts. In this work, we focus on clinical relation extraction; namely, given a medical record with mentions of drugs and their attributes, we identify relations between these entities. We propose a machine learning model with a novel set of knowledge-based and BioSentVec embedding features. We systematically investigate the impact of these features with standard distance- and word-based features, conducting experiments on two benchmark datasets of clinical texts from MADE 2018 and n2c2 2018 shared tasks. For comparison with the feature-based model, we utilize state-of-the-art models and three BERT-based models, including BioBERT and Clinical BERT. Our results demonstrate that distance and word features provide significant benefits to the classifier. Knowledge-based features improve classification results only for particular types of relations. The sentence embedding feature provides the largest improvement in results, among other explored features on the MADE corpus. The classifier obtains state-of-the-art performance in clinical relation extraction with F-measure of 92.6%, improving F-measure by 3.5% on the MADE corpus.

Semantic relation extraction using sequential and tree-structured LSTM with attention

Semantic relation extraction is crucial to automatically constructing a knowledge graph (KG), and it supports a variety of downstream natural language processing (NLP) tasks such as query answering (QA), semantic search and textual entailment. In addition, the semantic relation extraction task is mainly responsible for identifying entity pairs from raw texts and extracting the semantic relations between the extracted entity pairs. Existing methods consider only lexical-level features and often ignore syntactic features, resulting in poor relation extraction performance. By analyzing the necessity of the syntactic dependency and the contributions of words in a sentence to relation extraction, this paper proposes an end-to-end method that uses bidirectional tree-structured long short-term memory (LSTM) to extract structural features based on the dependency tree of a sentence. To enhance the performance of the relation extraction, the bidirectional sequential LSTM with attention is used to identify word-based features including the positional information of entity pairs and the contribution of words. Then, structural features and word-based features are concatenated to optimize the relation extraction performance. Finally, the proposed method is used on the SemEval 2010 task 8 and the CoNLL04 datasets to validate its performance. The experimental results show that the proposed method achieves state-of-the-art results on the SemEval 2010 task 8 and the CoNLL04 datasets.

Improving Distantly Supervised Relation Extraction with Neural Noise Converter and Conditional Optimal Selector

Distant supervised relation extraction has been successfully applied to large corpus with thousands of relations. However, the inevitable wrong labeling problem by distant supervision will hurt the performance of relation extraction. In this paper, we propose a method with neural noise converter to alleviate the impact of noisy data, and a conditional optimal selector to make proper prediction. Our noise converter learns the structured transition matrix on logit level and captures the property of distant supervised relation extraction dataset. The conditional optimal selector on the other hand helps to make proper prediction decision of an entity pair even if the group of sentences is overwhelmed by no-relation sentences. We conduct experiments on a widely used dataset and the results show significant improvement over competitive baseline methods.

Dual CNN for Relation Extraction with Knowledge-Based Attention and Word Embeddings.

Relation extraction is the underlying critical task of textual understanding. However, the existing methods currently have defects in instance selection and lack background knowledge for entity recognition. In this paper, we propose a knowledge-based attention model, which can make full use of supervised information from a knowledge base, to select an entity. We also design a method of dual convolutional neural networks (CNNs) considering the word embedding of each word is restricted by using a single training tool. The proposed model combines a CNN with an attention mechanism. The model inserts the word embedding and supervised information from the knowledge base into the CNN, performs convolution and pooling, and combines the knowledge base and CNN in the full connection layer. Based on these processes, the model not only obtains better entity representations but also improves the performance of relation extraction with the help of rich background knowledge. The experimental results demonstrate that the proposed model achieves competitive performance.

Exploring semi-supervised variational autoencoders for biomedical relation extraction.

The biomedical literature provides a rich source of knowledge such as protein-protein interactions (PPIs), drug-drug interactions (DDIs) and chemical-protein interactions (CPIs). Biomedical relation extraction aims to automatically extract biomedical relations from biomedical text for various biomedical research. State-of-the-art methods for biomedical relation extraction are primarily based on supervised machine learning and therefore depend on (sufficient) labeled data. However, creating large sets of training data is prohibitively expensive and labor-intensive, especially so in biomedicine as domain knowledge is required. In contrast, there is a large amount of unlabeled biomedical text available in PubMed. Hence, computational methods capable of employing unlabeled data to reduce the burden of manual annotation are of particular interest in biomedical relation extraction. We present a novel semi-supervised approach based on variational autoencoder (VAE) for biomedical relation extraction. Our model consists of the following three parts, a classifier, an encoder and a decoder. The classifier is implemented using multi-layer convolutional neural networks (CNNs), and the encoder and decoder are implemented using both bidirectional long short-term memory networks (Bi-LSTMs) and CNNs, respectively. The semi-supervised mechanism allows our model to learn features from both the labeled and unlabeled data. We evaluate our method on multiple public PPI, DDI and CPI corpora. Experimental results show that our method effectively exploits the unlabeled data to improve the performance and reduce the dependence on labeled data. To our best knowledge, this is the first semi-supervised VAE-based method for (biomedical) relation extraction. Our results suggest that exploiting such unlabeled data can be greatly beneficial to improved performance in various biomedical relation extraction, especially when only limited labeled data (e.g. 2000 samples or less) is available in such tasks.

Utilizing soft constraints to enhance medical relation extraction from the history of present illness in electronic medical records

Relation extraction between medical concepts from electronic medical records has pervasive applications as well as significance. However, previous researches utilizing machine learning algorithms judge the semantic types of medical concept pair mentions independently. In fact, different concept pair mentions in the same context are of dependencies which can provide beneficial evidences for identifying their relation types. To the best of our knowledge, only one study has considered such dependencies in discharge summaries. However, its hard constraints are not applied effectively to the History of Present Illness (HPI) in electronic Medical Records. According to the writing characteristics of HPI records, we generalize two regularities of dependencies among concept pairs mentioned in an HPI record to enhance the performance of relation extraction. We incorporate the two soft constraints corresponding to the regularities and the posterior probabilities returned by a local classifier into a joint inference process which applies Integer Quadratic Programming method to carry out collective classification for all concept pair mentions in an HPI record. We implement four local classification models including support vector machine, logistics regression, random forest and piecewise convolutional neural networks to examine the performance of our approach. A series of experimental results demonstrate that our collective classification method has made a principal improvement and outperforms the other state-of-the-art methods.

Research on Pattern Representation Based on Keyword and Word Embedding in Chinese Entity Relation Extraction

With the rapid development of the Internet, it is becoming more and more important to extract the relationship between the entity from the massive network text and then to build the knowledge graph or the knowledge base. In this paper, we focus on the research of the pattern representation in relation extraction, and extract the high accuracy Chinese entity pairs from large scale web texts. Past relation patterns only consider shallow lexical and syntax, not accurately and deeply express pattern context information, and do not consider keywords information. According to the new entity relation extraction technology and the characteristics of Chinese corpora, we define pattern representation based on keywords and word embedding information, extract deep semantic feature of context information, and strengthen keywords information effect for relation extraction. In addition, we propose a method for obtaining sentence keyword based on word embedding. In the experiment, we use ChineseHudongEncyclopedia corpus to implement the character relation extraction system, and test the character relation extraction effect. The experimental results show that this method effectively improves the quality of the pattern, and obtains a favorable relation extraction performance.

SEE: Syntax-Aware Entity Embedding for Neural Relation Extraction

Distant supervised relation extraction is an efficient approach to scale relation extraction to very large corpora, and has been widely used to find novel relational facts from plain text. Recent studies on neural relation extraction have shown great progress on this task via modeling the sentences in low-dimensional spaces, but seldom considered syntax information to model the entities. In this paper, we propose to learn syntax-aware entity embedding for neural relation extraction. First, we encode the context of entities on a dependency tree as sentence-level entity embedding based on tree-GRU. Then, we utilize both intra-sentence and inter-sentence attentions to obtain sentence set-level entity embedding over all sentences containing the focus entity pair. Finally, we combine both sentence embedding and entity embedding for relation classification. We conduct experiments on a widely used real-world dataset and the experimental results show that our model can make full use of all informative instances and achieve state-of-the-art performance of relation extraction.

Using semantic similarity to reduce wrong labels in distant supervision for relation extraction

Distant supervision (DS) has the advantage of automatically generating large amounts of labelled training data and has been widely used for relation extraction. However, there are usually many wrong labels in the automatically labelled data in distant supervision (Riedel, Yao, & McCallum, 2010). This paper presents a novel method to reduce the wrong labels. The proposed method uses the semantic Jaccard with word embedding to measure the semantic similarity between the relation phrase in the knowledge base and the dependency phrases between two entities in a sentence to filter the wrong labels. In the process of reducing wrong labels, the semantic Jaccard algorithm selects a core dependency phrase to represent the candidate relation in a sentence, which can capture features for relation classification and avoid the negative impact from irrelevant term sequences that previous neural network models of relation extraction often suffer. In the process of relation classification, the core dependency phrases are also used as the input of a convolutional neural network (CNN) for relation classification. The experimental results show that compared with the methods using original DS data, the methods using filtered DS data performed much better in relation extraction. It indicates that the semantic similarity based method is effective in reducing wrong labels. The relation extraction performance of the CNN model using the core dependency phrases as input is the best of all, which indicates that using the core dependency phrases as input of CNN is enough to capture the features for relation classification and could avoid negative impact from irrelevant terms.

A hybrid model based on neural networks for biomedical relation extraction

Biomedical relation extraction can automatically extract high-quality biomedical relations from biomedical texts, which is a vital step for the mining of biomedical knowledge hidden in the literature. Recurrent neural networks (RNNs) and convolutional neural networks (CNNs) are two major neural network models for biomedical relation extraction. Neural network-based methods for biomedical relation extraction typically focus on the sentence sequence and employ RNNs or CNNs to learn the latent features from sentence sequences separately. However, RNNs and CNNs have their own advantages for biomedical relation extraction. Combining RNNs and CNNs may improve biomedical relation extraction. In this paper, we present a hybrid model for the extraction of biomedical relations that combines RNNs and CNNs. First, the shortest dependency path (SDP) is generated based on the dependency graph of the candidate sentence. To make full use of the SDP, we divide the SDP into a dependency word sequence and a relation sequence. Then, RNNs and CNNs are employed to automatically learn the features from the sentence sequence and the dependency sequences, respectively. Finally, the output features of the RNNs and CNNs are combined to detect and extract biomedical relations. We evaluate our hybrid model using five public (protein-protein interaction) PPI corpora and a (drug-drug interaction) DDI corpus. The experimental results suggest that the advantages of RNNs and CNNs in biomedical relation extraction are complementary. Combining RNNs and CNNs can effectively boost biomedical relation extraction performance.

BioCreative VI Precision Medicine Track system performance is constrained by entity recognition and variations in corpus characteristics.

Precision medicine aims to provide personalized treatments based on individual patient profiles. One critical step towards precision medicine is leveraging knowledge derived from biomedical publications—a tremendous literature resource presenting the latest scientific discoveries on genes, mutations and diseases. Biomedical natural language processing (BioNLP) plays a vital role in supporting automation of this process. BioCreative VI Track 4 brings community effort to the task of automatically identifying and extracting protein–protein interactions (PPi) affected by mutations (PPIm), important in the precision medicine context for capturing individual genotype variation related to disease.We present the READ-BioMed team’s approach to identifying PPIm-related publications and to extracting specific PPIm information from those publications in the context of the BioCreative VI PPIm track. We observe that current BioNLP tools are insufficient to recognise entities for these two tasks; the best existing mutation recognition tool achieves only 55% recall in the document triage training set, while relation extraction performance is limited by the low recall performance of gene entity recognition. We develop the models accordingly: for document triage, we develop term lists capturing interactions and mutations to complement BioNLP tools, and select effective features via a feature contribution study, whereas an ensemble of BioNLP tools is employed for relation extraction.Our best document triage model achieves an F-score of 66.77% while our best model for relation extraction achieved an F-score of 35.09% over the final (updated post-task) test set. Impacting the document triage task, the characteristics of mutations are statistically different in the training and testing sets. While a vital new direction for biomedical text mining research, this early attempt to tackle the problem of identifying genetic variation of substantial biological significance highlights the importance of representative training data and the cascading impact of tool limitations in a modular system.

Chemical-gene relation extraction using recursive neural network.

In this article, we describe our system for the CHEMPROT task of the BioCreative VI challenge. Although considerable research on the named entity recognition of genes and drugs has been conducted, there is limited research on extracting relationships between them. Extracting relations between chemical compounds and genes from the literature is an important element in pharmacological and clinical research. The CHEMPROT task of BioCreative VI aims to promote the development of text mining systems that can be used to automatically extract relationships between chemical compounds and genes. We tested three recursive neural network approaches to improve the performance of relation extraction. In the BioCreative VI challenge, we developed a tree-Long Short-Term Memory networks (tree-LSTM) model with several additional features including a position feature and a subtree containment feature, and we also applied an ensemble method. After the challenge, we applied additional pre-processing steps to the tree-LSTM model, and we tested the performance of another recursive neural network model called Stack-augmented Parser Interpreter Neural Network (SPINN). Our tree-LSTM model achieved an F-score of 58.53% in the BioCreative VI challenge. Our tree-LSTM model with additional pre-processing and the SPINN model obtained F-scores of 63.7 and 64.1%, respectively.Database URL: https://github.com/arwhirang/recursive_chemprot

SSEL-ADE: A semi-supervised ensemble learning framework for extracting adverse drug events from social media.

With the development of Web 2.0 technology, social media websites have become lucrative but under-explored data sources for extracting adverse drug events (ADEs), which is a serious health problem. Besides ADE, other semantic relation types (e.g., drug indication and beneficial effect) could hold between the drug and adverse event mentions, making ADE relation extraction - distinguishing ADE relationship from other relation types - necessary. However, conducting ADE relation extraction in social media environment is not a trivial task because of the expertise-dependent, time-consuming and costly annotation process, and the feature space's high-dimensionality attributed to intrinsic characteristics of social media data. This study aims to develop a framework for ADE relation extraction using patient-generated content in social media with better performance than that delivered by previous efforts. To achieve the objective, a general semi-supervised ensemble learning framework, SSEL-ADE, was developed. The framework exploited various lexical, semantic, and syntactic features, and integrated ensemble learning and semi-supervised learning. A series of experiments were conducted to verify the effectiveness of the proposed framework. Empirical results demonstrate the effectiveness of each component of SSEL-ADE and reveal that our proposed framework outperforms most of existing ADE relation extraction methods The SSEL-ADE can facilitate enhanced ADE relation extraction performance, thereby providing more reliable support for pharmacovigilance. Moreover, the proposed semi-supervised ensemble methods have the potential of being applied to effectively deal with other social media-based problems.

A Customized Attention-Based Long Short-Term Memory Network for Distant Supervised Relation Extraction.

Distant supervision, a widely applied approach in the field of relation extraction can automatically generate large amounts of labeled training corpus with minimal manual effort. However, the labeled training corpus may have many false-positive data, which would hurt the performance of relation extraction. Moreover, in traditional feature-based distant supervised approaches, extraction models adopt human design features with natural language processing. It may also cause poor performance. To address these two shortcomings, we propose a customized attention-based long short-term memory network. Our approach adopts word-level attention to achieve better data representation for relation extraction without manually designed features to perform distant supervision instead of fully supervised relation extraction, and it utilizes instance-level attention to tackle the problem of false-positive data. Experimental results demonstrate that our proposed approach is effective and achieves better performance than traditional methods.

Bootstrapping Distantly Supervised IE Using Joint Learning and Small Well-Structured Corpora

We propose a framework to improve the performance of distantly-supervised relation extraction, by jointly learning to solve two related tasks: concept-instance extraction and relation extraction. We further extend this framework to make a novel use of document structure: in some small, well-structured corpora, sections can be identified that correspond to relation arguments, and distantly-labeled examples from such sections tend to have good precision. Using these as seeds we extract additional relation examples by applying label propagation on a graph composed of noisy examples extracted from a large unstructured testing corpus. Combined with the soft constraint that concept examples should have the same type as the second argument of the relation, we get significant improvements over several state-of-the-art approaches to distantly-supervised relation extraction, and reasonable extraction performance even with very small set of distant labels.

Improving chemical disease relation extraction with rich features and weakly labeled data

BackgroundDue to the importance of identifying relations between chemicals and diseases for new drug discovery and improving chemical safety, there has been a growing interest in developing automatic relation extraction systems for capturing these relations from the rich and rapid-growing biomedical literature. In this work we aim to build on current advances in named entity recognition and a recent BioCreative effort to further improve the state of the art in biomedical relation extraction, in particular for the chemical-induced disease (CID) relations.ResultsWe propose a rich-feature approach with Support Vector Machine to aid in the extraction of CIDs from PubMed articles. Our feature vector includes novel statistical features, linguistic knowledge, and domain resources. We also incorporate the output of a rule-based system as features, thus combining the advantages of rule- and machine learning-based systems. Furthermore, we augment our approach with automatically generated labeled text from an existing knowledge base to improve performance without additional cost for corpus construction. To evaluate our system, we perform experiments on the human-annotated BioCreative V benchmarking dataset and compare with previous results. When trained using only BioCreative V training and development sets, our system achieves an F-score of 57.51 %, which already compares favorably to previous methods. Our system performance was further improved to 61.01 % in F-score when augmented with additional automatically generated weakly labeled data.ConclusionsOur text-mining approach demonstrates state-of-the-art performance in disease-chemical relation extraction. More importantly, this work exemplifies the use of (freely available) curated document-level annotations in existing biomedical databases, which are largely overlooked in text-mining system development.