Search

Over the past decade, information for precision disease medicine has accumulated in the form of textual data. To effectively utilize this expanding medical text, we proposed a multi-task learning-based framework based on hard parameter sharing for knowledge graph construction (MKG), and then used it to automatically extract gastric cancer (GC)-related biomedical knowledge from the literature and identify GC drug candidates. In MKG, we designed three separate modules, MT-BGIPN, MT-SGTF and MT-ScBERT, for entity recognition, entity normalization, and relation classification, respectively. To address the challenges posed by the long and irregular naming of medical entities, the MT-BGIPN utilized bidirectional gated recurrent unit and interactive pointer network techniques, significantly improving entity recognition accuracy to an average F1 value of 84.5% across datasets. In MT-SGTF, we employed the term frequency-inverse document frequency and the gated attention unit. These combine both semantic and characteristic features of entities, resulting in an average Hits@ 1 score of 94.5% across five datasets. The MT-ScBERT integrated cross-text, entity, and context features, yielding an average F1 value of 86.9% across 11 relation classification datasets. Based on the MKG, we then developed a specific knowledge graph for GC (MKG-GC), which encompasses a total of 9129 entities and 88,482 triplets. Lastly, the MKG-GC was used to predict potential GC drugs using a pre-trained language model called BioKGE-BERT and a drug-disease discriminant model based on CNN-BiLSTM. Remarkably, nine out of the top ten predicted drugs have been previously reported as effective for gastric cancer treatment. Finally, an online platform was created for exploration and visualization of MKG-GC at https://www.yanglab-mi.org.cn/MKG-GC/.

This paper focuses on the multi-choice tasks, which aim to select the correct choice for a given query by reasoning over texts, such as sentences and passages. Benefiting from the provided knowledge in these tasks, the reasoning of multi-choice models becomes well-founded. However, besides the evidence of the correct choice, the text usually contains some content related to other choices as well, which causes multi-choice models vulnerable to being confused by candidate choices, and we denote it as choice confusion problem. To alleviate this challenge, we propose two auxiliary mechanisms to distinguish the confusing choices. Specifically, a query-guided attention (QGA) mechanism is designed to automatically filter the text contents causing confusion by measuring the syntactic relevance between different contents and the query. Meanwhile, a confusion-aware training (CAT) mechanism is designed to learn the correct and easily confused choices asynchronously and perform a pushing-away operation between their selection processes. We conduct experiments on multi-choice tasks based on sentences and passages. The results show that our framework improves the choice selection accuracy compared to strong baselines. Furthermore, the ablation test and case study verify the effectiveness of our proposed QGA and CAT, especially in addressing the choice confusion problem. [ABSTRACT FROM AUTHOR]

With the development of neural networks, more and more complex and excellent relation classification models are constantly proposed. Although they can be compressed by some model compression methods at the cost of effectiveness, they are still insufficient to deploy on resource-constrained devices. Knowledge distillation can transfer the excellent predictive abilities of superior models to lightweight models, but the gap between models limits its effects. Due to the huge gaps between relation classification models, it is painstakingly difficult to select and train a superior teacher model to guide student models when we use knowledge distillation to get a lightweight model. Therefore, how to obtain a lightweight relation classification model with high effectiveness is still a hot research topic. In this paper, we construct an alternate distillation framework with three modules. The weight adaptive external distillation module is built based on an adaptive weighting module based on cosine similarity. The progressive internal distillation module allows the model to be its own teacher to guide its own training. Finally, a combination module based on the attention mechanism combines the above two modules. On SemEval-2010 Task 8 and WiKi80 datasets, we demonstrate the great effect of our approach on improving the relation classification effectiveness of lightweight models. The complex relation classification models compressed at the cost of effectiveness are still insufficient to deploy on resource-constrained devices. Besides, due to the significant differences between relation classification models, it is challenging to find a suitable teacher model for knowledge distillation. In this paper, we propose an alternate distillation framework (including external distillation and internal distillation) to obtain lightweight relation classification models with high effectiveness. Our approach effectively transfers the excellent predictive capability of complex models to lightweight models even when there is a significant gap between them [ABSTRACT FROM AUTHOR]

Relation classification (RC) is an essential task in information extraction. The distance supervision (DS) method can use many unlabeled data and solve the lack of training data on the RC task. However, the DS method has the problems of long tails and noise. Intuitively, people can solve these problems using few-shot learning (FSL). Our work aims to improve the accuracy and rapidity of convergence on the few-shot RC task. We believe that entity pairs have an essential role in the few-shot RC task. We propose a new context encoder, which is improved based on the bidirectional encoder representations from transformers (BERT) model to fuse entity pairs and their dependence information in instances. At the same time, we design hybrid attention, which includes support instance-level and query instance-level attention. The support instance level dynamically assigns the weight of each instance in the support set. It makes up for the insufficiency of prototypical networks, which distribute weights to sentences equally. Query instance-level attention is dynamically assigned weights to query instances by similarity with the prototype. The ablation study shows the effectiveness of our proposed method. In addition, a fusion network is designed to replace the Euclidean distance method of previous works when class matching is performed, improving the convergence's rapidity. This makes our model more suitable for industrial applications. The experimental results show that the proposed model's accuracy is better than that of several other models. [ABSTRACT FROM AUTHOR]

Relation classification is an important task in information extraction that involves identifying potential semantic connections between two entities in a sentence. Most relation extraction models either use semantic or structure information as the relation representation for classification. Although some researchers have tried to fuse both types of information, they simply combine different information as relation representation, which ignores semantic and structure information contribute unequally to the relation representation at the instance level, leading to degradation of performance and generalization capability. To address this issue, we propose a fusing multi-information model with a self-attentive strategy (Att-FMI) for relation classification. Our approach utilizes the pre-trained BERT model and dependency syntactic parser to obtain semantic and structure representations from the sequence, respectively and efficiently combines the two with support the selfattention mechanism. Experimental results on four commonly used datasets, including TACRED, TACREV, KBP-37 and SemEval-2010 Task8, demonstrate that our approach yields significant improvements over baseline methods. Additionally, we observe that the Att-FMI model exhibits greater robustness against information interference in the extraction of entity relations from longer sentences than previous methods. [ABSTRACT FROM AUTHOR]

Few-shot relation classification is to recognize the semantic relation between an entity pair with very few samples. Prototypical network has proven to be a simple yet effective few-shot learning method for relation extraction. However, under the condition of data scarcity, the relation prototypes we achieve are usually biased compared to the real ones computed from all samples within a relation class. To alleviate this issue, we propose hybrid enhancement-based prototypical networks. In particular, our model contains three main enhancement modules: 1) a query-guided prototype enhancement module using rich interactive information between the support instances and the query instance as guidance to obtain more accurate prototype representations; 2) a query enhancement module to diminish the distribution gap between the query set and the support set; 3) a support enhancement module adopting a pseudo-label strategy to expand the scale of available data. On basis of these modules, we further design a novel prototype attention fusion mechanism to fuse information and compute discriminative relation prototypes for classification. In this way, we hope to obtain unbiased representations closer to our expected prototypes by improving the available data scale and data utilization efficiency. Extensive experimental results on the widely-used FewRel dataset demonstrate the superiority of our proposed model. [ABSTRACT FROM AUTHOR]

The airfreight industry of shipping goods with special handling needs, also known as special cargo, often deals with non-transparent data and outdated technology, resulting in significant inefficiency. A special cargo ontology is a means of extracting, structuring, and storing domain knowledge and representing the concepts and relationships that can be processed by computers. This ontology can be used as the base of semantic data retrieval in many artificial intelligence applications, such as planning for special cargo shipments. Domain information extraction is an essential task in implementing and maintaining special cargo ontology. However, the absence of domain information makes instantiating the cargo ontology challenging. We propose a relation representation learning approach based on a hierarchical attention-based multi-task model and leverage it in the special cargo domain. The proposed relation representation learning architecture is applied for identifying and categorizing samples of various relation types in the special cargo ontology. The model is trained with domain-specific documents on a number of semantic tasks that vary from lightweight tasks in the bottom layers to the heavyweight tasks in the top layers of the model in a hierarchical setting. Therefore, it conveys complementary input features and learns a rich representation. We also train a domain-specific relation representation model that relies only on an entity-linked corpus of cargo shipment domain. These two relation representation models are then employed in a supervised multi-class classifier called Special Cargo Relation Extractor (SCRE). The results of the experiments show that the proposed relation representation models can represent the complex semantic information of the special cargo domain efficiently. [ABSTRACT FROM AUTHOR]

Relation classification is a significant task within the field of natural language processing. Its objective is to extract and identify relations between two entities in a given text. Within the scope of this paper, we construct an artificial dataset (CS13K) for relation classification in the realm of cybersecurity and propose two models for processing such tasks. For any sentence containing two target entities, we first locate the entities and fine-tune the pre-trained BERT model. Next, we utilize graph attention networks to iteratively update word nodes and relation nodes. A new relation classification model is constructed by concatenating the updated vectors of word nodes and relation nodes. Our proposed model achieved exceptional performance on the SemEval-2010 task 8 dataset, surpassing previous approaches with a remarkable F1 value of 92.3%. Additionally, we propose the integration of a ranking-based voting mechanism into the existing model. Our best results are an F1 value of 92.5% on the SemEval-2010 task 8 dataset and a value 94.6% on the CS13K dataset. These findings highlight the effectiveness of our proposed models in tackling relation classification tasks. [ABSTRACT FROM AUTHOR]

Relation classification (RC) identifies the semantic relation between entity pairs and plays a critical role in knowledge graph construction and knowledge graph completion. However, insufficient labeled instances of long-tail relations make the training of supervised and distant supervised (DS) relation classification models difficult. Few-shot RC is an effective solution to this problem. At present, metric-based few-shot RC models focus on the representation of relation prototypes and the interaction between instances, ignoring meaningful entity representation and the association of entities and other words in the instance. We propose a prototypical attention network with an entity-aware embedding module (PAN-EAEM) to solve this problem. Firstly, the entity-aware embedding module (EAEM) draws more attention to entity-related words to capture key features. This plug-and-play module can improve the performance of other metric-based models as well. Secondly, the prototypical attention network (PAN) decreases the influence of noise on relation prototype representation by reducing intra-class differences and inter-class ambiguities. Extensive experiments prove that our proposed model obtains state-of-the-art performance on the FewRel dataset. [ABSTRACT FROM AUTHOR]

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Relation classification as a core technique for building knowledge graphs becomes a critical task in natural language processing. The fact that humans can learn by summarizing and generalizing limited knowledge motivates scholars to explore few-shot learning. Graph neural networks provide a method to measure the distance between nodes, which improves the model effect in the problem of few-shot relation classification. However, graph neural network methods focus only on node information and ignore edge information which implies inter-class and intra-class relations. This paper proposes edge-labeled and node-aggregated graph neural networks (ENGNNs) for few-shot relation classification: edge labels are encoded and used for node information aggregation. In addition, a process of semi-supervised learning is designed to discover a better solution for one-shot learning. Compared with previous methods, experimental results show that the proposed ENGNN model improves the performance of the graph neural network on the FewRel dataset. [ABSTRACT FROM AUTHOR]

In the digital age, many sources of textual content are devoted to studying and expressing many sorts of relationships, including employer–employee, if–then, part–whole, product–producer, and cause–effect relations/causality. Mining cause–effect relations are a key topic in many NLP (natural language processing) applications, such as future event prediction, information retrieval, healthcare, scenario generation, decision making, commerce risk management, question answering, and adverse drug reaction. Many statistical and non-statistical methods have been developed in the past to address this topic. Most of them frequently used feature-driven supervised approaches and hand-crafted linguistic patterns. However, the implicit and ambiguous statement of causation prevented these methods from achieving great recall and precision. They cover a limited set of implicit causality and are difficult to extend. In this work, a novel MCKN (multi-column knowledge-oriented network) is introduced. This model includes various knowledge-oriented channels/columns (KCs), where each channel integrates prior human knowledge to capture language cues of causation. MCKN uses unique convolutional word filters (wf) generated automatically using WordNet and FrameNet. To reduce MCKN's dimensionality, we use filter selection and clustering approaches. Our model delivers superior performance on the Alternative Lexicalization (AltLexes) dataset, proving that MCKN is a simpler and distinctive approach for informal datasets. [ABSTRACT FROM AUTHOR]

With more and more news articles appearing on the Internet, discovering causal relations between news articles is very important for people to understand the development of news. Extracting the causal relations between news articles is an inter-document relation extraction task. Existing works on relation extraction cannot solve it well because of the following two reasons: (1) most relation extraction models are intra-document models, which focus on relation extraction between entities. However, news articles are many times longer and more complex than entities, which makes the inter-document relation extraction task harder than intra-document. (2) Existing inter-document relation extraction models rely on similarity information between news articles, which could limit the performance of extraction methods. In this paper, we propose an inter-document model based on storytree information to extract causal relations between news articles. We adopt storytree information to integer linear programming (ILP) and design the storytree constraints for the ILP objective function. Experimental results show that all the constraints are effective and the proposed method outperforms widely used machine learning models and a state-of-the-art deep learning model, with F1 improved by more than 5% on three different datasets. Further analysis shows that five constraints in our model improve the results to varying degrees and the effects on the three datasets are different. The experiment about link features also suggests the positive influence of link information. [ABSTRACT FROM AUTHOR]

To improve the accuracy of the few-shot relation classification task, the research, which weakens the influence of confounder on the performance of the model and enhances the semantic representation and feature extraction ability of the model, is carried out based on the prototypical network. And then, the weaken-confounders method based on causal intervention(WCCI) is proposed, and the RBERTI-Proto model is constructed. In WCCI, the pre-trained knowledge is stratified by the backdoor adjustment based on causal intervention, the optimal stratified number is determined by a stratified method, and the BN layer is introduced for the gradient disappearance problem. In the RBERTI-Proto model, the abilities of semantic representation and feature extraction of the model are enhanced by which the RoBERTa is used as the feature extractor of the model. Experimental results demonstrate the effectiveness of our proposed methods and the RoBERTa model as feature extractor of our model, and the ACC value of the RBERTI-Proto model achieve 93.38% on the 5-way 5-shot scenario of the FewRel dataset.

Copyright of Journal of Computer Engineering & Applications is the property of Beijing Journal of Computer Engineering & Applications Journal Co Ltd. and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Relation classification is an important fundamental task in information extraction, and convolutional neural networks have been commonly applied to relation classification with good results. In recent years, due to the proposed pre-training model BERT, the use of which as a feature extraction architecture has become more and more popular, convolutional neural networks have gradually withdrawn from the stage of NLP, and the relation classification/extraction model based on pre-training BERT has achieved state-of-the-art results. However, none of these methods consider how to accurately capture the semantic features of the relationships between entities to reduce the number of noisy words in a sentence that are not helpful for relation classification. Moreover, these methods do not have a systematic prediction structure to fully utilize the extracted features for the relational classification task. To address these problems, a SpanBert-based relation classification model is proposed in this paper. Compared with existing Bert-based architectures, the model is able to understand the semantic information of the relationships between entities more accurately, and it can fully utilize the extracted features to represent the degree of dependency of a pair of entities with each type of relationship. In this paper, we design a feature fusion method called "SRS" (Strengthen Relational Semantics) and an attention-based prediction structure. Compared with existing methods, the feature fusion method proposed in this paper can reduce the noise interference of irrelevant words when extracting relational semantics, and the prediction structure proposed in this paper can make full use of semantic features for relational classification. We achieved advanced results on the SemEval-2010 Task 8 and the KBP37 relational dataset. [ABSTRACT FROM AUTHOR]

Featured Application: With this contribution, we aim to aid the drug development process as well as the identification of possible adverse drug events due to simultaneous drug use. The task of extracting drug entities and possible interactions between drug pairings is known as Drug–Drug Interaction (DDI) extraction. Computer-assisted DDI extraction with Machine Learning techniques can help streamline this expensive and time-consuming process during the drug development cycle. Over the years, a variety of both traditional and Neural Network-based techniques for the extraction of DDIs have been proposed. Despite the introduction of several successful strategies, obtaining high classification accuracy is still an area where further progress can be made. In this work, we present a novel Knowledge Graph (KG) based approach that utilizes a unique graph structure in combination with a Transformer-based Language Model and Graph Neural Networks to classify DDIs from biomedical literature. The KG is constructed to model the knowledge of the DDI Extraction 2013 benchmark dataset, without the inclusion of additional external information sources. Each drug pair is classified based on the context of the sentence it was found in, by utilizing transfer knowledge in the form of semantic representations from domain-adapted BioBERT weights that serve as the initial KG states. The proposed approach was evaluated on the DDI classification task of the same dataset and achieved a F1-score of 79.14% on the four positive classes, outperforming the current state-of-the-art approach. [ABSTRACT FROM AUTHOR]

Causality mining in NLP is a significant area of interest, which benefits in many daily life applications, including decision making, business risk management, question answering, future event prediction, scenario generation, and information retrieval. Mining those causalities was a challenging and open problem for the prior non-statistical and statistical techniques using web sources that required hand-crafted linguistics patterns for feature engineering, which were subject to domain knowledge and required much human effort. Those studies overlooked implicit, ambiguous, and heterogeneous causality and focused on explicit causality mining. In contrast to statistical and non-statistical approaches, we present Bidirectional Encoder Representations from Transformers (BERT) integrated with Multilevel Feature Networks (MFN) for causality recognition, called BERT+MFN for causality recognition in noisy and informal web datasets without human-designed features. In our model, MFN consists of a three-column knowledge-oriented network (TC-KN), bi-LSTM, and Relation Network (RN) that mine causality information at the segment level. BERT captures semantic features at the word level. We perform experiments on Alternative Lexicalization (AltLexes) datasets. The experimental outcomes show that our model outperforms baseline causality and text mining techniques. [ABSTRACT FROM AUTHOR]

(1) Background: Geological surveying is undergoing a digital transformation process towards the adoption of intelligent methods in China. Cognitive intelligence methods, such as those based on knowledge graphs and machine reading, have made progress in many domains and also provide a technical basis for quality detection in unstructured lithographic description texts. (2) Methods: First, the named entities and the relations of the domain-specific knowledge graph of petrography were defined based on the petrographic theory. Second, research was carried out based on a manually annotated corpus of petrographic description. The extraction of N-ary and single-entity overlapping relations and the separation of complex entities are key steps in this process. Third, a petrographic knowledge graph was formulated based on prior knowledge. Finally, the consistency between knowledge triples extracted from the corpus and the petrographic knowledge graph was calculated. The 1:50,000 sheet of Fengxiangyi located in the Dabie orogenic belt was selected for the empirical research. (3) Results: Using machine reading and the knowledge graph, petrographic knowledge can be extracted and the knowledge consistency calculation can quickly detect description errors about textures, structures and mineral components in petrographic description. (4) Conclusions: The proposed framework can be used to realise the intelligent inspection of petrographic knowledge with complex entities and relations and to improve the quality of petrographic description texts effectively. [ABSTRACT FROM AUTHOR]

The social network is an indispensable part of our life. Text is the most common carrier in social networks. Extracting entities and relationships from a text in social media can help to mine people's views and attitudes. However, identifying the entity pairs that overlap between multiple relations in a sentence and the subject and object that overlap in a relation is a tricky question to be solved urgently. We propose a new relation extraction model named GraphJoint, which models the relation extraction task as a mapping from the relation to the entity. Firstly, we apply the pre-trained BERT encoder to encode the words and generate a text graph for each sentence. We use the graph neural network message-passing mechanism to extract the text features in a sentence, which are used to classify the relations in the sentences. Secondly, we reuse the extracted features and add the relation features to extract the entities. The self-attention mechanism and dilated gate convolution are used to extract entity features further. Finally, we use the joint annotation method to mark the head, tail, and overlapping parts of the subject and the object and transform the task into a sequence labeling task. Experiments compared with other advanced algorithms on two public data sets prove that our method increases the F1 value of the two data sets by 3.6% and 3.4% and achieves a perfect recognition effect in the recognition of overlapping entity pairs. [ABSTRACT FROM AUTHOR]

Relation classification is a significant task within the field of natural language processing. Its objective is to extract and identify relations between two entities in a given text. Within the scope of this paper, we construct an artificial dataset (CS13K) for relation classification in the realm of cybersecurity and propose two models for processing such tasks. For any sentence containing two target entities, we first locate the entities and fine-tune the pre-trained BERT model. Next, we utilize graph attention networks to iteratively update word nodes and relation nodes. A new relation classification model is constructed by concatenating the updated vectors of word nodes and relation nodes. Our proposed model achieved exceptional performance on the SemEval-2010 task 8 dataset, surpassing previous approaches with a remarkable F1 value of 92.3%. Additionally, we propose the integration of a ranking-based voting mechanism into the existing model. Our best results are an F1 value of 92.5% on the SemEval-2010 task 8 dataset and a value 94.6% on the CS13K dataset. These findings highlight the effectiveness of our proposed models in tackling relation classification tasks.

Relation classification (RC) is a fundamental task to building knowledge graphs and describing semantic formalization. It aims to classify a relation between the head and the tail entities in a sentence. The existing RC method mainly adopts the distant supervision (DS) scheme. However, DS still has the problem of long-tail and suffers from data sparsity. Recently, few-shot learning (FSL) has attracted people's attention. It solves the long-tail problem by learning from few-shot samples. The prototypical networks have a better effect on FSL, which classifies a relation by distance. However, the prototypical networks and their related variants did not consider the critical role of entity words. In addition, not all sentences in support set equally contributed to classifying relations. Furthermore, an entity pair in a sentence may have true and confusing relations, which is difficult for the RC model to distinguish them. A new context encoder BERT_FE is proposed to address those problems, which uses the BERT model as pre-training and fuses the information of head and tail entities by entity word-level attention (WLA). At the same time, the sentence-level attention (SLA) is proposed to give more weight to sentences of the support set similar to the query instance and improve the classification accuracy. A confusing loss function (CLF) is designed to enhance the model's ability to distinguish between true and confusing relations. The experiment results demonstrate that our proposed model (HACLF) is better than several baseline models. [ABSTRACT FROM AUTHOR]

Relation extraction is an important information extraction task that must be solved in order to transform data into Knowledge Graph (KG), as semantic relations between entities form KG edges of the graph. Although much effort has been devoted to solve this task during the last three decades, but the results achieved are not as good yet. For instance, winner at Text Analysis Conference's (TAC) Knowledge Base Population (KBP) 2015 slot filling task, the Stanford's system, achieves F1 score of 60.5% on standard Relation Extraction (RE) dataset (Zhang et al., in: Position-aware attention and supervised data improve slot_lling. In: EMNLP 2017-Conference on Empirical Methods in Natural Language Processing, Proceedings, (2017). https://doi.org/10.18653/v1/d17-1004). The RE task therefore needs better solutions. This paper presents our system, CustRE, for better identification and classification of family relations from English text. CustRE is a rule based system, that uses regular expressions for pattern matching to extract family relations explicitly mentioned in text, and uses co-reference and propagation rules to extract family relations implicitly implied in the text. The proposed system, its implementation and the results obtained are presented in this paper. The results show that our approach makes a great improvement over existing methods by achieving F1 scores of 79.7% and 76.6% on TACRED family relations and CustFRE datasets respectively, which are 6.3 and 18.5 points higher than LUKE, the best score reporter on TACRED. [ABSTRACT FROM AUTHOR]

The objective of the relation classification task is to extract relations between entities. Recent studies have found that R-BERT (Wu and He, 2019) based on pre-trained BERT (Devlin et al., 2019) acquires extremely good results in the relation classification task. However, this method does not take into account the semantic differences between different kinds of entities and global semantic information either. In this paper, we set two different fully connected layers to take into account the semantic difference between subject and object entities. Besides, we build a new module named Concat Module to fully fuse the semantic information among the subject entity vector, object entity vector, and the whole sample sentence representation vector. In addition, we apply the average pooling to acquire a better entity representation of each entity and add the activation operation with a new fully connected layer after our Concat Module. Modifying R-BERT, we propose a new model named BERT with Global Semantic Information (GSR-BERT) for relation classification tasks. We use our approach on two datasets: the SemEval-2010 Task 8 dataset and the Chinese character relationship classification dataset. Our approach achieves a significant improvement over the two datasets. It means that our approach enjoys transferability across different datasets. Furthermore, we prove that these policies we used in our approach also enjoy applicability to named entity recognition task. • We take into account the semantic difference between subject and object entities. • We fully fuse the semantic information between entities and the whole sentence. • These policies we used enjoy high applicability to different NLP tasks. [ABSTRACT FROM AUTHOR]

Relation extraction is an essential task in the field of biomedical literature mining and offers significant benefits for various downstream applications, including database curation, drug repurposing, and literature-based discovery. The broad-coverage natural language processing (NLP) tool SemRep has established a solid baseline for extracting subject–predicate–object triples from biomedical text and has served as the backbone of the Semantic MEDLINE Database (SemMedDB), a PubMed-scale repository of semantic triples. While SemRep achieves reasonable precision (0.69), its recall is relatively low (0.42). In this study, we aimed to enhance SemRep using a relation classification approach, in order to eventually increase the size and the utility of SemMedDB. We combined and extended existing SemRep evaluation datasets to generate training data. We leveraged the pre-trained PubMedBERT model, enhancing it through additional contrastive pre-training and fine-tuning. We experimented with three entity representations: mentions, semantic types, and semantic groups. We evaluated the model performance on a portion of the SemRep Gold Standard dataset and compared it to SemRep performance. We also assessed the effect of the model on a larger set of 12K randomly selected PubMed abstracts. Our results show that the best model yields a precision of 0.62, recall of 0.81, and F 1 score of 0.70. Assessment on 12K abstracts shows that the model could double the size of SemMedDB, when applied to entire PubMed. We also manually assessed the quality of 506 triples predicted by the model that SemRep had not previously identified, and found that 67% of these triples were correct. These findings underscore the promise of our model in achieving a more comprehensive coverage of relationships mentioned in biomedical literature, thereby showing its potential in enhancing various downstream applications of biomedical literature mining. Data and code related to this study are available at https://github.com/Michelle-Mings/SemRep_RelationClassification. [Display omitted] [ABSTRACT FROM AUTHOR]

Relation classification tends to struggle when training data are limited or when it needs to adapt to unseen categories. In such challenging scenarios, recent approaches employ the metric-learning framework to measure similarities between query and support examples and to determine relation labels of the query sentences based on the similarities. However, these approaches treat each support class independently from one another, never looking at the entire task as a whole. Because of this, they are constrained to using a shared set of features for all meta-tasks, which hinders the ability to compose discriminative features for the task at hand. For example, if two similar relation types occur in a meta-task, the model needs to construct more detailed, task-related features instead of common features shared by all tasks. In this paper, we propose a novel task-aware relation classification model to tackle this issue. We first build a task embedding component to capture task-specific information, after which two mechanisms, such as task-specific gate and gated feature combination methods, are proposed to utilize the task-specific information to guide feature composition dynamically for each meta-task. Experiment results show that our model improves performance considerably over high performing baseline systems on both FewRel 1.0 and FewRel 2.0 benchmarks. Moreover, our proposed methods can be incorporated into metric-learning-based methods and significantly improve their performance. [ABSTRACT FROM AUTHOR]

Meaningful Information extraction is an extremely important and challenging task due to the ever growing size of data. Training and evaluating automated systems for the task requires annotated datasets which are rarely available because of the great amount of human effort and time required for annotating data. The dataset described in this manuscript, CustFRE, is meant for systems that learn extracting family relations from text. Sentences having at least two persons have been collected from the internet. The texts are first processed using Stanford's NLP pipeline for basic NLP tagging. Next, a team of natural language processing experts annotated the dataset. All family relations among persons in the texts have been annotated, or a no_relation is annotated if no family relation between two persons can be inferred from the text. After annotation, the dataset was verified by an NLP expert for completeness and correctness. CustFRE contains in total 2,716 annotations. The dataset can be used by information extraction researchers as a benchmark for evaluating their systems, and can also be used for training and evaluating family relation extraction systems.