Event Extraction Method Research Articles

Tactical mission event logic graph construction for Network-Centric Warfare

Constructing a Tactical Mission Event Logic Graph (TMELG) can assist military staff to analyze the composition and evolution process of tactical missions. We can construct Military Internet of Things (MIoT) applications based on the TMELG to assist military staff to make planning and study tactical missions in the Network-Centric Warfare situation. In this paper, we proposed a representation framework for tactical mission events, based on the Simple Event Model (SEM), to present and construct the TMELG. First of all, sub-mission events and event elements are extracted from military texts with an invented Two-Stage Event Extraction (TSEE) method. Secondly, the Hierarchical Task Network (HTN) was applied to decompose a tactical mission into sub-missions, which were further decomposed into atomic missions, then basic relations were established for sub-missions. Finally, we exploited the explicit causality among sub-mission events, and the implicit causality between atomic mission events and military resources with the Granger Causality Model (GCM). With respect to automatic extraction of sub-mission events and event elements from military texts, the experimental results showed that our TSEE method outperforms the listed methods. The successful application of TMELG to a joint landing operation case verifies the feasibility of our construction method proposed in this paper.

Reinforcement Learning-Based Dialogue Guided Event Extraction to Exploit Argument Relations

Event extraction is a ftask for natural language processing. Finding the roles of event arguments like event participants is essential for event extraction. However, doing so for real-life event descriptions is challenging because an argument’s role often varies in different contexts. While the relationship and interactions between multiple arguments are useful for settling the argument roles, such information is largely ignored by existing approaches. This paper presents a better approach for event extraction by explicitly utilizing the relationships of event arguments. We achieve this through a carefully designed task-oriented dialogue system. To model the argument relation, we employ reinforcement learning and incremental learning to extract multiple arguments via a multi-turned, iterative process. Our approach leverages knowledge of the already extracted arguments of the same sentence to determine the role of arguments that would be difficult to decide individually. It then uses the newly obtained information to improve the decisions of previously extracted arguments. This two-way feedback process allows us to exploit the argument relations to effectively settle argument roles, leading to better sentence understanding and event extraction. Experimental results show that our approach consistently outperforms seven state-of-the-art event extraction methods for the classification of events and argument role and argument identification.

Cross-Correlation Wavelet-Domain-Based Particle Swarm Optimization for Lightning Mapping

Several processing methods have been proposed for estimating the real pattern of the temporal location and spatial map of the lightning strikes. However, due to the complexity of lightning signals, providing accurate lightning maps estimation remains a challenging task. This paper presents a cross-correlation wavelet-domain-based particle swarm optimization (CCWD-PSO) technique for an accurate and robust representation of lightning mapping. The CCWD method provides an initial estimate of the lightning map, while the PSO attempts to optimize the trajectory of the lightning map by finding the optimal sliding window of the cross-correlation. The technique was further enhanced through the introduction of a novel lightning event extraction method that enables faster processing of the lightning mapping. The CCWD-PSO method was validated and verified using three narrow bipolar events (NBEs) flashes. The observed results demonstrate that this technique offers high accuracy in representing the real lightning mapping with low estimation errors.

LSLSD: Fusion Long Short-Level Semantic Dependency of Chinese EMRs for Event Extraction

Most existing medical event extraction methods have primarily adopted a simplex model based on either pattern matching or deep learning, which ignores the distribution characteristics of entities and events in the medical corpus. They have not categorized the granularity of event elements, leading to the poor generalization ability of the model. This paper proposes a diagnosis and treatment event extraction method in the Chinese language, fusing long short-level semantic dependency of the corpus, LSLSD, for solving these problems. LSLSD can effectively capture different levels of semantic information within and between event sentences in the electronic medical record (EMR) corpus. Moreover, the event arguments are divided into short word-level and long sentence-level, with the sequence annotation and pattern matching combined to realize multi-granularity argument recognition, as well as to improve the generalization ability of the model. Finally, this paper constructs a diagnosis and treatment event data set of Chinese EMRs by proposing a semi-automatic corpus labeling method, and an enormous number of experiment results show that LSLSD can improve the F1-value of event extraction task by 7.1% compared with the several strong baselines.

Research on Chinese Event Extraction Method Based on HMM and Multi-stage Method

The existing Chinese event extraction methods are mostly one-stage methods. Even if there are two-stage methods, the correlation between the stages is not high, the extracted event trigger words and event elements have low matching degree, and there is a class imbalance problem in the training process. This paper proposes a Chinese event extraction method based on HMM and multi-stage method. The first stage of the method recognises the event trigger word positive examples in the text; the second stage classifies the trigger word positive examples recognised in the first stage, determines the event type, and forms the sequence of event trigger words; the third stage matches the event elements, and forms the sequence of event elements, according to the first two stages of event trigger word extraction results. This method effectively alleviates the class imbalance problem in training, improves the matching degree between event trigger words and event elements extraction, and obtains better extraction performance. The accuracy, recall rate and F value of Chinese event extraction are all More than 78%.

基于BERT-DGCNN的中文事件抽取方法研究

基于BERT-DGCNN的中文事件抽取方法研究

A Knowledge Graph Based Framework in Relationship Modelling and Real-time Monitoring of Market Participants

Various types of relationships among participants exist in the foreign exchange market, such as transaction, credit relations inside the market and subcompany, investing and shareholding relations outside the market. In the market with complex relationships, an incident happened in one institution may have “butterfly effect”. To monitor systemic financial risks, we integrate various types of data and establish a Knowledge Graph describing the relationships between market participants. Based on the Knowledge Graph, we design a framework to measure the impact of various events on the entire market by connecting the internal market relationships to external market related incidents. By using the BiLSTM-CRF model, we extract the relevant entities in the market news with high accuracy and measure the impact on the market with graph related indicators such as closeness and betweenness centralities. In addition, to use external unstructured data, we use a document-level event extraction method to analyse these unstructured datasets and extract new events and relationships from them to dynamically update the Knowledge Graph.

Dynamic Spatio-Temporal Tweet Mining for Event Detection: A Case Study of Hurricane Florence

Extracting information about emerging events in large study areas through spatiotemporal and textual analysis of geotagged tweets provides the possibility of monitoring the current state of a disaster. This study proposes dynamic spatio-temporal tweet mining as a method for dynamic event extraction from geotagged tweets in large study areas. It introduces the use of a modified version of ordering points to identify the clustering structure to address the intrinsic heterogeneity of Twitter data. To precisely calculate the textual similarity, three state-of-the-art text embedding methods of Word2vec, GloVe, and FastText were used to capture both syntactic and semantic similarities. The impact of selected embedding algorithms on the quality of the outputs was studied. Different combinations of spatial and temporal distances with the textual similarity measure were investigated to improve the event detection outcomes. The proposed method was applied to a case study related to 2018 Hurricane Florence. The method was able to precisely identify events of varied sizes and densities before, during, and after the hurricane. The feasibility of the proposed method was qualitatively evaluated using the Silhouette coefficient and qualitatively discussed. The proposed method was also compared to an implementation based on the standard density-based spatial clustering of applications with noise algorithm, where it showed more promising results.

Event-based summarization of news articles

In recent years, with the increase of available digital information on the Web, the time needed to find relevant information is also increased. Therefore, to reduce the time spent on searching, research on automatic text summarization has gained importance. The proposed summarization process is based on event extraction methods and is called an event-based extractive single-document summarization. In this method, the important features of event extraction and summarization methods are analyzed and combined together to extract the summaries from single-source news documents. Among the tested features, six features are found to be the most effective in constructing good summaries. The constructed summaries are tested on benchmark Document Understanding Conferences 2001 and 2002 datasets, and the results outperformed most of the other well-known summarization methods.

Joint Event Extraction Model based on Multi-feature Fusion

Abstract Event extraction is a challenging problem in information extraction, designed to extract structured information from unstructured text. The existing event extraction methods are mostly based on the pipeline model and use the traditional language representation model for word embedding. It is impossible to model the polysemy and the features is very simple, besides, the pipeline model is also easy to generate cascading errors. Aiming at the above problems, we propose a joint event extraction model, which combines part-of-speech features, entity features and word representation of BERT model. Because of the BERT can’t capture local features, we use multi-CNN to capture the local features of different scales. and further use BiLSTM to capture context features to assist joint event extraction. Experimental results show that the precision, recall and F1 of the method on the CEC dataset are 88.83%, 90.91% and 89.86%, respectively, which are 1.61%, 3.6% and 2.6% higher than the PLMEE model.

A novel method for drug-adverse event extraction using machine learning

Abstract Background An extensive amount of data derived from medical case reports regarding potential adverse events is subjected to manual review. Devising efficient strategies for identification and information extraction concerning potential adverse events are needed to support timely monitoring of the reports and decision making. Methods This paper aims at providing a machine learning (ML) and natural language processing (NLP) based solution for extracting suspect drugs and adverse events. The solution is based upon two approaches: Causal Sentence Classification classifies the relationship between drug and medical condition as causal or non-causal, and Suspect Drug Identification classifies each drug present in the report as a suspect drug or non-suspect drug. Results Causal Sentence Classification yielded a precision of 0.85 and recall of 0.84 in establishing causality between drugs and medical conditions on the testing dataset consisting of 6252 records. After evaluation on a reliable testing dataset of 3522 records, the Suspect Drug Identification successfully identified suspect drugs with a precision of 0.72 and recall of 0.77. Conclusions The developed solution relies on semantic and syntactic based features to capture the writing style of incoming reports, and showcases the potential of ML and NLP for Pharmacovigilance.

Stock Market Prediction via Multi-Source Multiple Instance Learning

Forecasting the stock market movements is an important and challenging task. As the Web information grows, researchers begin to extract effective indicators (e.g., the events and sentiments) from the Web to facilitate the prediction. However, the indicators obtained in previous studies are usually based on only one data source and thus may not fully cover the factors that can affect the stock market movements. In this paper, to improve the prediction for stock market composite index movements, we exploit the consistencies among different data sources, and develop a multi-source multiple instance model that can effectively combine events, sentiments, as well as the quantitative data into a comprehensive framework. To effectively capture the news events, we successfully apply a novel event extraction and representation method. Evaluations on the data from the year 2015 and 2016 demonstrate the effectiveness of our model. In addition, our approach is able to automatically determine the importance of each data source and identify the crucial input information that is considered to drive the movements, making the predictions interpretable.

A multiple distributed representation method based on neural network for biomedical event extraction

BackgroundBiomedical event extraction is one of the most frontier domains in biomedical research. The two main subtasks of biomedical event extraction are trigger identification and arguments detection which can both be considered as classification problems. However, traditional state-of-the-art methods are based on support vector machine (SVM) with massive manually designed one-hot represented features, which require enormous work but lack semantic relation among words.MethodsIn this paper, we propose a multiple distributed representation method for biomedical event extraction. The method combines context consisting of dependency-based word embedding, and task-based features represented in a distributed way as the input of deep learning models to train deep learning models. Finally, we used softmax classifier to label the example candidates.ResultsThe experimental results on Multi-Level Event Extraction (MLEE) corpus show higher F-scores of 77.97% in trigger identification and 58.31% in overall compared to the state-of-the-art SVM method.ConclusionsOur distributed representation method for biomedical event extraction avoids the problems of semantic gap and dimension disaster from traditional one-hot representation methods. The promising results demonstrate that our proposed method is effective for biomedical event extraction.

Prior Knowledge-Based Event Network for Chinese Text

Text representation is a basic issue of text information processing and event plays an important role in text understanding; both attract the attention of scholars. The event network conceals lexical relations in events, and its edges express logical relations between events in document. However, the events and relations are extracted from event-annotated text, which makes it hard for large-scale text automatic processing. In the paper, with expanded CEC (Chinese Event Corpus) as data source, prior knowledge of manifestation rules of event and relation as the guide, we propose an event extraction method based on knowledge-based rule of event manifestation, to achieve automatic building and improve text processing performance of event network.

Global Social Event Extraction and Analysis by Processing Online News

The understanding of global social pattern can benefit the society operation a lot, including the domestic social governance, international situation awareness, risk assessment and forecasting, conflict resolution, crisis response and future policy planning. As the development of Internet, hundreds of millions of news could be found online every day, reporting the social events around the world, including political events, diplomatic events, cultural events, natural disasters, etc. However, by manual reading and analyzing, it is too difficult to deal with the vast amount of data and obtain valuable information quickly. Thus, in this paper we investigate the global social event extraction and analysis method based on the automatic processing of online news, including 1) event model building, 2) event information extraction and automatic classification based on English news text, 3) global social dynamic analysis and visualization based on event data. Finally, we constructed the method on the real global news data collected from more than 200 sitesto evaluate their performance and interpret some underlying insights of the results.

A Survey of event extraction methods from text for decision support systems

Event extraction, a specialized stream of information extraction rooted back into the 1980s, has greatly gained in popularity due to the advent of big data and the developments in the related fields of text mining and natural language processing. However, up to this date, an overview of this particular field remains elusive. Therefore, we give a summarization of event extraction techniques for textual data, distinguishing between data-driven, knowledge-driven, and hybrid methods, and present a qualitative evaluation of these. Moreover, we discuss common decision support applications of event extraction from text corpora. Last, we elaborate on the evaluation of event extraction systems and identify current research issues.

Analyzing depression tendency of web posts using an event-driven depression tendency warning model.

The Internet has become a platform to express individual moods/feelings of daily life, where authors share their thoughts in web blogs, micro-blogs, forums, bulletin board systems or other media. In this work, we investigate text-mining technology to analyze and predict the depression tendency of web posts. In this paper, we defined depression factors, which include negative events, negative emotions, symptoms, and negative thoughts from web posts. We proposed an enhanced event extraction (E3) method to automatically extract negative event terms. In addition, we also proposed an event-driven depression tendency warning (EDDTW) model to predict the depression tendency of web bloggers or post authors by analyzing their posted articles. We compare the performance among the proposed EDDTW model, negative emotion evaluation (NEE) model, and the diagnostic and statistical manual of mental disorders-based depression tendency evaluation method. The EDDTW model obtains the best recall rate and F-measure at 0.668 and 0.624, respectively, while the diagnostic and statistical manual of mental disorders-based method achieves the best precision rate of 0.666. The main reason is that our enhanced event extraction method can increase recall rate by enlarging the negative event lexicon at the expense of precision. Our EDDTW model can also be used to track the change or trend of depression tendency for each post author. The depression tendency trend can help doctors to diagnose and even track depression of web post authors more efficiently. This paper presents an E3 method to automatically extract negative event terms in web posts. We also proposed a new EDDTW model to predict the depression tendency of web posts and possibly help bloggers or post authors to early detect major depressive disorder.

A data-dependent Fourier filter based on image segmentation for random seismic noise attenuation

In this paper, we propose a data-dependent Fourier filter (DDFF) based on image segmentation for random seismic noise attenuation. In the proposed method, the original seismic data is divided into some overlapped small blocks. For each block, a local Fourier filter is designed automatically in two steps. At first, a binary mask is obtained by segmenting the Fourier amplitude spectra (FAS) of this block. The histogram of the FAS is used to get the threshold for the FAS segmentation. Secondly, an average filter is applied on the binary mask to get the tapered Fourier filter. In the proposed method, the DDFFs for all blocks are compact and time–space variant. After all blocks are processed, they are merged together to form the filtered result. We illustrate our method by a 2D synthetic seismic data, and give a comparison with the coherent event extraction method in Fourier domain. At last, a real 3D seismic data example demonstrates that the proposed method obtains some promising results.

Semi-supervised method for biomedical event extraction

BackgroundBiomedical extraction based on supervised machine learning still faces the problem that a limited labeled dataset does not saturate the learning method. Many supervised learning algorithms for bio-event extraction have been affected by the data sparseness.MethodsIn this study, a semi-supervised method for combining labeled data with large scale of unlabeled data is presented to improve the performance of biomedical event extraction. We propose a set of rich feature vector, including a variety of syntactic features and semantic features, such as N-gram features, walk subsequence features, predicate argument structure (PAS) features, especially some new features derived from a strategy named Event Feature Coupling Generalization (EFCG). The EFCG algorithm can create useful event recognition features by making use of the correlation between two sorts of original features explored from the labeled data, while the correlation is computed with the help of massive amounts of unlabeled data. This introduced EFCG approach aims to solve the data sparse problem caused by limited tagging corpus, and enables the new features to cover much more event related information with better generalization properties.ResultsThe effectiveness of our event extraction system is evaluated on the datasets from the BioNLP Shared Task 2011 and PubMed. Experimental results demonstrate the state-of-the-art performance in the fine-grained biomedical information extraction task.ConclusionsLimited labeled data could be combined with unlabeled data to tackle the data sparseness problem by means of our EFCG approach, and the classified capability of the model was enhanced through establishing a rich feature set by both labeled and unlabeled datasets. So this semi-supervised learning approach could go far towards improving the performance of the event extraction system. To the best of our knowledge, it was the first attempt at combining labeled and unlabeled data for tasks related biomedical event extraction.

Extracting Events from Web Documents for Social Media Monitoring Using Structured SVM

Event extraction is vital to social media monitoring and social event prediction. In this paper, we propose a method for social event extraction from web documents by identifying binary relations between named entities. There have been many studies on relation extraction, but their aims were mostly academic. For practical application, we try to identify 130 relation types that comprise 31 predefined event types, which address business and public issues. We use structured Support Vector Machine, the state of the art classifier to capture relations. We apply our method on news, blogs and tweets collected from the Internet and discuss the results. key words: Relation Extraction, Structured SVM, Natural Language Processing, Information Extraction