Shortest Dependency Path Research Articles

Leveraging shortest dependency paths in low-resource biomedical relation extraction

BackgroundBiomedical Relation Extraction (RE) is essential for uncovering complex relationships between biomedical entities within text. However, training RE classifiers is challenging in low-resource biomedical applications with few labeled examples.MethodsWe explore the potential of Shortest Dependency Paths (SDPs) to aid biomedical RE, especially in situations with limited labeled examples. In this study, we suggest various approaches to employ SDPs when creating word and sentence representations under supervised, semi-supervised, and in-context-learning settings.ResultsThrough experiments on three benchmark biomedical text datasets, we find that incorporating SDP-based representations enhances the performance of RE classifiers. The improvement is especially notable when working with small amounts of labeled data.ConclusionSDPs offer valuable insights into the complex sentence structure found in many biomedical text passages. Our study introduces several straightforward techniques that, as demonstrated experimentally, effectively enhance the accuracy of RE classifiers.

Enhancing relation extraction using multi-task learning with SDP evidence

Relation extraction (RE) is a crucial subtask of information extraction, which involves recognizing the relation between entity pairs in a sentence. Previous studies have extensively employed syntactic information, notably the shortest dependency path (SDP), to collect word evidence, termed SDP evidence, which gives clues about the given entity pair, thus improving RE. Nevertheless, prevalent transformer-based techniques lack syntactic information and cannot effectively model essential syntactic clues to support relations. This study exerts multi-task learning to address these issues by imbibing an SDP token position prediction task into the RE task. To this end, we introduce SGA, an SDP evidence guiding approach that transfers the SDP evidence into two novel supervisory signal labels: SDP tokens label and SDP matrix label. The former guides the attention modules to assign high attention weights to SDP token positions, emphasizing relational clues. In the meantime, the latter supervises SGA to predict a parameterized asymmetric product matrix among the SDP tokens for RE. Experimental outcomes demonstrate the model's enhanced ability to leverage SDP information, thereby directing attention modules and predicted matrix labels to focus on SDP evidence. Consequently, our proposed approach surpasses existing publicly available optimal baselines across four RE datasets: SemEval2010-Task8, KBP37, NYT, and WebNLG.1

Protein-Protein Interaction Network Extraction Using Text Mining Methods Adds Insight into Autism Spectrum Disorder.

Text mining methods are being developed to assimilate the volume of biomedical textual materials that are continually expanding. Understanding protein-protein interaction (PPI) deficits would assist in explaining the genesis of diseases. In this study, we designed an automated system to extract PPIs from the biomedical literature that uses a deep learning sentence classification model, a pretrained word embedding, and a BiLSTM recurrent neural network with additional layers, a conditional random field (CRF) named entity recognition (NER) model, and shortest-dependency path (SDP) model using the SpaCy library in Python. The automated system ensures that it targets sentences that contain PPIs and not just these proteins mentioned in the framework of disease discovery or other context. Our first model achieved 13% greater precision on the Aimed/BioInfr benchmark corpus than the previous state-of-the-art BiLSTM neural network models. The NER model presented in this study achieved 98% precision on the Aimed/BioInfr corpus over previous models. In order to facilitate the production of an accurate representation of the PPI network, the processes were developed to systematically map the protein interactions in the texts. Overall, evaluating our system through the use of 6027 abstracts pertaining to seven proteins associated with Autism Spectrum Disorder completed the manually curated PPI network for these proteins. When it comes to complicated diseases, these networks would assist in understanding how PPI deficits contribute to disease development while also emphasizing the influence of interactions on protein function and biological processes.

A Concise Relation Extraction Method Based on the Fusion of Sequential and Structural Features Using ERNIE

Relation extraction, a fundamental task in natural language processing, aims to extract entity triples from unstructured data. These triples can then be used to build a knowledge graph. Recently, pre-training models that have learned prior semantic and syntactic knowledge, such as BERT and ERNIE, have enhanced the performance of relation extraction tasks. However, previous research has mainly focused on sequential or structural data alone, such as the shortest dependency path, ignoring the fact that fusing sequential and structural features may improve the classification performance. This study proposes a concise approach using the fused features for the relation extraction task. Firstly, for the sequential data, we verify in detail which of the generated representations can effectively improve the performance. Secondly, inspired by the pre-training task of next-sentence prediction, we propose a concise relation extraction approach based on the fusion of sequential and structural features using the pre-training model ERNIE. The experiments were conducted on the SemEval 2010 Task 8 dataset and the results show that the proposed method can improve the F1 value to 0.902.

Enhancing Semantic Relation Classification With Shortest Dependency Path Reasoning

Relation Classification (RC) is a basic and essential task of Natural Language Processing. Existing RC methods can be classified into two categories: sequence-based methods and dependency-based methods. Sequence-based methods identify the target relation based on the overall semantics of the whole sentence, which will inevitably introduce noisy features. Dependency-based methods extract indicative word-level features from the Shortest Dependency Path (SDP) between given entities and attempt to establish a statistical association between the words and the target relations. This pattern relatively eliminates the influence of noisy features and achieves a robust performance on long sentences. Nevertheless, we observe that majority of relation classification processes involve complex semantic reasoning which is hard to be achieved based on the word-level statistical association. To solve this problem, we categorize all relations into atomic relations and composed-relations. The atomic relations are the basic relations that can be identified based on the word-level features, while the composed-relation requires to be deducted from multiple atomic relations. Correspondingly, we propose the <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">At</b> omic Relation <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">E</b> ncoding and <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">R</b> easoning <bold xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">M</b> odel (ATERM). In the atomic relation encoding stage, ATERM groups the word-level features and encodes multiple atomic relations in parallel. In the atomic relation reasoning stage, ATERM establishes the atomic relation chain where relation-level features are extracted to identify composed-relations. Experiments show that our method achieves state-of-the-art results on the three most popular relation classification datasets – TACRED, TACRED-Revisit, and SemEval 2010 task 8 with significant improvements.

DocR-BERT: Document-Level R-BERT for Chemical-Induced Disease Relation Extraction via Gaussian Probability Distribution.

Chemical-induced disease (CID) relation extraction from biomedical articles plays an important role in disease treatment and drug development. Existing methods are insufficient for capturing complete document level semantic information due to ignoring semantic information of entities in different sentences. In this work, we proposed an effective document-level relation extraction model to automatically extract intra-/inter-sentential CID relations from articles. Firstly, our model employed BERT to generate contextual semantic representations of the title, abstract and shortest dependency paths (SDPs). Secondly, to enhance the semantic representation of the whole document, cross attention with self-attention (named cross2self-attention) between abstract, title and SDPs was proposed to learn the mutual semantic information. Thirdly, to distinguish the importance of the target entity in different sentences, the Gaussian probability distribution was utilized to compute the weights of the co-occurrence sentence and its adjacent entity sentences. More complete semantic information of the target entity is collected from all entities occurring in the document via our presented document-level R-BERT (DocR-BERT). Finally, the related representations were concatenated and fed into the softmax function to extract CIDs. We evaluated the model on the CDR corpus provided by BioCreative V. The proposed model without external resources is superior in performance as compared with other state-of-the-art models (our model achieves 53.5%, 70%, and 63.7% of the F1-score on inter-/intra-sentential and overall CDR dataset). The experimental results indicate that cross2self-attention, the Gaussian probability distribution and DocR-BERT can effectively improve the CID extraction performance. Furthermore, the mutual semantic information learned by the cross self-attention from abstract towards title can significantly influence the extraction performance of document-level biomedical relation extraction tasks.

Enhancing Biomedical Relation Extraction with Transformer Models using Shortest Dependency Path Features and Triplet Information

Entity relation extraction plays an important role in the biomedical, healthcare, and clinical research areas. Recently, pre-trained models based on transformer architectures and their variants have shown remarkable performances in various natural language processing tasks. Most of these variants were based on slight modifications in the architectural components, representation schemes and augmenting data using distant supervision methods. In distantly supervised methods, one of the main challenges is pruning out noisy samples. A similar situation can arise when the training samples are not directly available but need to be constructed from the given dataset. The BioCreative V Chemical Disease Relation (CDR) task provides a dataset that does not explicitly offer mention-level gold annotations and hence replicates the above scenario. Selecting the representative sentences from the given abstract or document text that could convey a potential entity relationship becomes essential. Most of the existing methods in literature propose to either consider the entire text or all the sentences which contain the entity mentions. This could be a computationally expensive and time consuming approach. This paper presents a novel approach to handle such scenarios, specifically in biomedical relation extraction. We propose utilizing the Shortest Dependency Path (SDP) features for constructing data samples by pruning out noisy information and selecting the most representative samples for model learning. We also utilize triplet information in model learning using the biomedical variant of BERT, viz., BioBERT. The problem is represented as a sentence pair classification task using the sentence and the entity-relation pair as input. We analyze the approach on both intra-sentential and inter-sentential relations in the CDR dataset. The proposed approach that utilizes the SDP and triplet features presents promising results, specifically on the inter-sentential relation extraction task. We make the code used for this work publicly available on Github.11https://github.com/vanikanjirangat/BioRelation-Extraction_BERT_SDP_Triplets

Improving Distant Supervised Relation Extraction using shortest dependency path

Distant supervised relation extraction is the problem of classifying the semantic relation between a pair of entities in a given sentence, where data is created by aligning unlabelled text with knowledge base automatically. Prior researches indicated that most of the sentences in the distant supervised relation extraction setting are very long and benefit from the attention over words. We note that the shortest dependency path between the entity pair in the syntax analysis tree of a sentence can help word based attention. In this paper, we propose a noval distant supervised neural relation extraction method, which makes use of the shortest dependency path to supervise the learning of word attention. Through extensive experiments on benchmark datasets, we demonstrate effectiveness of our model.

Structurally Comparative Hinge Loss for Dependency-Based Neural Text Representation

Dependency-based graph convolutional networks (DepGCNs) are proven helpful for text representation to handle many natural language tasks. Almost all previous models are trained with cross-entropy (CE) loss, which maximizes the posterior likelihood directly. However, the contribution of dependency structures is not well considered by CE loss. As a result, the performance improvement gained by using the structure information can be narrow due to the failure in learning to rely on this structure information. To face the challenge, we propose the novel structurally comparative hinge (SCH) loss function for DepGCNs. SCH loss aims at enlarging the margin gained by structural representations over non-structural ones. From the perspective of information theory, this is equivalent to improving the conditional mutual information of model decision and structure information given text. Our experimental results on both English and Chinese datasets show that by substituting SCH loss for CE loss on various tasks, for both induced structures and structures from an external parser, performance is improved without additional learnable parameters. Furthermore, the extent to which certain types of examples rely on the dependency structure can be measured directly by the learned margin, which results in better interpretability. In addition, through detailed analysis, we show that this structure margin has a positive correlation with task performance and structure induction of DepGCNs, and SCH loss can help model focus more on the shortest dependency path between entities. We achieve the new state-of-the-art results on TACRED, IMDB, and Zh. Literature datasets, even compared with ensemble and BERT baselines.

Direction-sensitive relation extraction using Bi-SDP attention model

Relation extraction is a crucial task of natural language processing (NLP). It plays a key role in question answering, web search, and information retrieval and so on. Previous research on this task has verified the effectiveness of using attention mechanisms, shortest dependency paths (SDP) and LSTM. However, most of these methods focus on learning a semantic representation of the whole sentence, highlighting the importance of partial words, or pruning the sentence with SDP. They ignore the lose of information in these methods, such as the dependency relation of each word and preposition words to indicate the relation direction. Besides, the SDP-based approach is prone to over-pruning. Based on the above observations, this paper presents a framework with a Bi-directional SDP (Bi-SDP) attention mechanism to tackle these challenges. The Bi-SDP is a novel representation of SDP, including original SDP and its reverse. The attention mechanism, based on Bi-SDP, builds a parallel word-level attention to capture relational semantic words and directional words. Furthermore, we explored a novel pruning strategy to minimize the length of input instance and the number of RNN cells simultaneously. Moreover, experiments are conducted on two datasets: SemEval-2010 Task 8 dataset and KBP37 dataset. Compared with the previous public models, our method can achieve better competitive performance on the SemEval-2010 Task 8 dataset and outperform existing models on the KBP37 dataset. Additionally, our experimental results also evidence that the directional prepositions in sentences are useful for relation extraction and can improve the performance of relationship with apparent physical direction.

A Hybrid Deep Learning Model for Protein–Protein Interactions Extraction from Biomedical Literature

The exponentially increasing size of biomedical literature and the limited ability of manual curators to discover protein–protein interactions (PPIs) in text has led to delays in keeping PPI databases updated with the current findings. The state-of-the-art text mining methods for PPI extraction are primarily based on deep learning (DL) models, and the performance of a DL-based method is mainly affected by the architecture of DL models and the feature embedding methods. In this study, we compared different architectures of DL models, including convolutional neural networks (CNN), long short-term memory (LSTM), and hybrid models, and proposed a hybrid architecture of a bidirectional LSTM+CNN model for PPI extraction. Pretrained word embedding and shortest dependency path (SDP) embedding are fed into a two-embedding channel model, such that the model is able to model long-distance contextual information and can capture the local features and structure information effectively. The experimental results showed that the proposed model is superior to the non-hybrid DL models, and the hybrid CNN+Bidirectional LSTM model works well for PPI extraction. The visualization and comparison of the hidden features learned by different DL models further confirmed the effectiveness of the proposed model.

Research on relation extraction of named entity on social media in smart cities

Social media make significant contribution to the evolution of smart cities. The key issue of smart cities is to develop a series of automatic methods to support smart applications. As one of the basic techniques of smart cities, the task of relation extraction of named entities on social media provides an indispensable means to construct and expand knowledge map and contributes to the utilization of information resources in smart cities. What’s more, it is conducive to improve the efficiency of network supervision. This paper proposes an automatic method to extract entity relations via deep learning techniques on a two-level neural network named Bi-CLSTM. The research conquers some challenges of relation extraction on social media. To extract entity relations on conversation scenarios, Bi-CLSTM represents texts with the strategy of “word embedding + position embedding + shortest dependency path” and extracts relations via a hybrid model of LSTM and PCNN. The nodes and networks of Bi-CLSTM are designed to adapt to the scenarios of conversation and over-sentence. To reduce the dependency on training data, distant supervised strategy is employed and a two-level attention mechanism is used to prevent noise signals. Experiments are carried out on Sina Microblog corpus, and the results show that Bi-CLSTM model makes outstanding performance.

Using Bidirectional LSTM and Shortest Dependency Path for Classifying Arabic Temporal Relations

Abstract The classification of temporal relations between two entities is a crucial task for the comprehension of natural languages. Several methods have been proposed, but most of them have focused on English and several other languages. Contrarily, Arabic has received less attention for the classification of temporal relations. Recently, methods based on deep learning have obtained more relevant results than traditional methods. These latter have used several features, but they have lacked the modeling of syntactic structures. In this context, we present a first model for classifying temporal relations between pairs of events (E-E) or an event and a temporal expression (E-T) in Arabic intra-sentence. Our idea is to combine two branches of bidirectional LSTM networks to extract syntactic information by modeling the Shortest Dependency Path (SDP) between two target entities as well as the whole sentence sequence. Each word in the sentence is concatenated with its Parts Of Speech (POS). For the SDP branch, each word is concatenated with its POS as well as its generated dependency relations. To evaluate our model, and due to the lack of publicly available Arabic resources, we create an annotated Arabic corpus for temporal relations between E-E or E-T based on TimeML specifications. In addition, we propose a data augmentation method using transitivity and symmetry rules for temporality. The experimental results based on our annotated corpus give 84.6% of F1 for the classification of temporal relations between E-E, and 73.32% of F1 for the same task between E-T.

Relation extraction between bacteria and biotopes from biomedical texts with attention mechanisms and domain-specific contextual representations

BackgroundThe Bacteria Biotope (BB) task is a biomedical relation extraction (RE) that aims to study the interaction between bacteria and their locations. This task is considered to pertain to fundamental knowledge in applied microbiology. Some previous investigations conducted the study by applying feature-based models; others have presented deep-learning-based models such as convolutional and recurrent neural networks used with the shortest dependency paths (SDPs). Although SDPs contain valuable and concise information, some parts of crucial information that is required to define bacterial location relationships are often neglected. Moreover, the traditional word-embedding used in previous studies may suffer from word ambiguation across linguistic contexts.ResultsHere, we present a deep learning model for biomedical RE. The model incorporates feature combinations of SDPs and full sentences with various attention mechanisms. We also used pre-trained contextual representations based on domain-specific vocabularies. To assess the model’s robustness, we introduced a mean F1 score on many models using different random seeds. The experiments were conducted on the standard BB corpus in BioNLP-ST’16. Our experimental results revealed that the model performed better (in terms of both maximum and average F1 scores; 60.77% and 57.63%, respectively) compared with other existing models.ConclusionsWe demonstrated that our proposed contributions to this task can be used to extract rich lexical, syntactic, and semantic features that effectively boost the model’s performance. Moreover, we analyzed the trade-off between precision and recall to choose the proper cut-off to use in real-world applications.

Relation Classification via Keyword-Attentive Sentence Mechanism and Synthetic Stimulation Loss

Previous studies have shown that attention mechanisms and shortest dependency paths have a positive effect on relation classification. In this paper, a keyword-attentive sentence mechanism is proposed to effectively combine the two methods. Furthermore, to effectively handle the imbalanced classification problem, this paper proposes a new loss function called the synthetic stimulation loss, which uses a modulating factor to allow the model to focus on hard-to-classify samples. The proposed two methods are integrated into a bidirectional gated recurrent unit (BiGRU). As a single model is not strong in noise immunity, this paper applies the mutual learning method to our model and forces the networks to teach each other. Therefore, we call the final model SSL-KAS-MuBiGRU. Experiments on the SemEval-2010 Task 8 data set and the TAC40 data set demonstrate that the keyword-attentive sentence mechanism and synthetic stimulation loss are useful for relation classification, and our model achieves state-of-the-art results.

A corpus of plant-disease relations in the biomedical domain.

BackgroundMany new medicines have been derived from natural sources such as plants, which have a long history of being used for disease treatment. Thus, their benefits and side effects have been studied, and plant-related information including plant and disease relations have been accumulated in Medline articles. Because numerous articles are available in Medline and are written in natural language, text-mining is important. However, a corpus of plant and disease relations is not available yet. Thus, we aimed to construct such a corpus.Methods and resultsIn this study, we designed and annotated a plant–disease relations corpus, and proposed a computational model to predict plant–disease relations using the corpus. We categorized plant and disease relations into four types: treatments of diseases, causes of diseases, associations, and negative relations. To construct a corpus of plant–disease relations, we first created its annotation guidelines and randomly selected 200 Medline abstracts. From these abstracts, we identified 1,405 and 1,755 plant and disease mentions, annotated to 105 and 237 unique plant and disease identifiers, respectively. When we selected sentences containing at least one plant and one disease mention, we extracted 878 plant and 1,077 disease entities, which finally generated a corpus of plant-disease relations including 1,309 relations from 199 abstracts. To verify the effectiveness of the corpus, we proposed a convolutional neural network model with the shortest dependency path (SDP-CNN) and applied it to the constructed corpus. The micro F-score with ten-fold cross-validation was found to be 0.764. We also applied the proposed SDP-CNN model to all Medline abstracts. When we measured its performance for 483 randomly selected plant-disease co-occurring sentences, the model showed a precision of 0.707.ConclusionThe plant–disease relations corpus is unique and represents an important resource for biomedical text-mining. The corpus of plant and disease relations is available at http://gcancer.org/pdr/.

Biomolecular-Level Event Detection: A New Representation of Generating Short Sentence and Sample Selection Strategy

Background: Biomolecular-level event extraction is one of the most important branches of information extraction. With the rapid growth of biomedical literature, it is difficult for researchers to manually obtain information of interest, e.g. unknown information of threatening human disease or some biological processes. Therefore, researchers are interested in automatically acquiring information of biomolecular-level events. However, the annotated biomolecular-level event corpus is limited and highly imbalanced, which affects the performance of the classification algorithms and can even lead to over-fitting. associations while known disease-lncRNA associations are required only. Method: In this paper, a new approach using the Pairwise model and convolutional neural network for biomolecular-level event extraction is introduced. The method can identify more accurate positive instances from unlabeled data to enlarge the labeled data. First, unlabeled samples are categorized using the Pairwise model. Then, the shortest dependency path with additional information is generated. Furthermore, two input forms with a new representation of the convolutional neural network model, which are dependency word sequence and dependency relation sequence are presented. Finally, with the sample selection strategy, the expanded labeled samples from unlabeled domain corpus incrementally enlarge the training data to improve the performance of the classifier. &lt;/P&gt;&lt;P&gt; Result &amp; Conclusion: Our proposed method achieved better performance than other excellent systems. This is due to our new representation of generated short sentence and proposed sample selection strategy, which greatly improved the accuracy of classification. The extensive experimental results indicate that the new method can effectively inculcate unlabeled data to improve the performance of classifier for biomolecular-level events extraction.&lt;/P&gt;

Integrating shortest dependency path and sentence sequence into a deep learning framework for relation extraction in clinical text

BackgroundExtracting relations between important clinical entities is critical but very challenging for natural language processing (NLP) in the medical domain. Researchers have applied deep learning-based approaches to clinical relation extraction; but most of them consider sentence sequence only, without modeling syntactic structures. The aim of this study was to utilize a deep neural network to capture the syntactic features and further improve the performances of relation extraction in clinical notes.MethodsWe propose a novel neural approach to model shortest dependency path (SDP) between target entities together with the sentence sequence for clinical relation extraction. Our neural network architecture consists of three modules: (1) sentence sequence representation module using bidirectional long short-term memory network (Bi-LSTM) to capture the features in the sentence sequence; (2) SDP representation module implementing the convolutional neural network (CNN) and Bi-LSTM network to capture the syntactic context for target entities using SDP information; and (3) classification module utilizing a fully-connected layer with Softmax function to classify the relation type between target entities.ResultsUsing the 2010 i2b2/VA relation extraction dataset, we compared our approach with other baseline methods. Our experimental results show that the proposed approach achieved significant improvements over comparable existing methods, demonstrating the effectiveness of utilizing syntactic structures in deep learning-based relation extraction. The F-measure of our method reaches 74.34% which is 2.5% higher than the method without using syntactic features.ConclusionsWe propose a new neural network architecture by modeling SDP along with sentence sequence to extract multi-relations from clinical text. Our experimental results show that the proposed approach significantly improve the performances on clinical notes, demonstrating the effectiveness of syntactic structures in deep learning-based relation extraction.

Extraction of chemical-protein interactions from the literature using neural networks and narrow instance representation.

The scientific literature contains large amounts of information on genes, proteins, chemicals and their interactions. Extraction and integration of this information in curated knowledge bases help researchers support their experimental results, leading to new hypotheses and discoveries. This is especially relevant for precision medicine, which aims to understand the individual variability across patient groups in order to select the most appropriate treatments. Methods for improved retrieval and automatic relation extraction from biomedical literature are therefore required for collecting structured information from the growing number of published works. In this paper, we follow a deep learning approach for extracting mentions of chemical–protein interactions from biomedical articles, based on various enhancements over our participation in the BioCreative VI CHEMPROT task. A significant aspect of our best method is the use of a simple deep learning model together with a very narrow representation of the relation instances, using only up to 10 words from the shortest dependency path and the respective dependency edges. Bidirectional long short-term memory recurrent networks or convolutional neural networks are used to build the deep learning models. We report the results of several experiments and show that our best model is competitive with more complex sentence representations or network structures, achieving an F1-score of 0.6306 on the test set. The source code of our work, along with detailed statistics, is publicly available.

Multi-Gram CNN-Based Self-Attention Model for Relation Classification

Relation classification is a crucial ingredient in numerous information-extraction systems and has attracted a great deal of attention in recent years. Traditional approaches largely rely on feature engineering and suffer from the limitations of the domain adaption and the error propagation. To overcome the above-mentioned problems, many deep neural network-based methods have been proposed; however, these methods cannot effectively locate and utilize the relation trigger features. To locate the relation trigger features and make full use of them, we propose a novel multi-gram convolution neural network-based self-attention model with a recurrent neural network framework. The multi-gram conventional neural network attention model can learn the adaptive relational semantics of inputs based on the fact that a relation can be totally defined by the shortest dependency path between its two entities. With the learned relational semantics, we can obtain the corresponding importance distribution over input sentences and locate the relation trigger features. For effective information propagation and integration, we utilize a bidirectional gated recurrent unit to encode the high-level features during recurrent propagation. The experimental results on two benchmark datasets demonstrate that the proposed model outperforms most of the state-of-the-art models.