基于混合模型的新闻事件要素提取方法

为了帮助读者从大量新闻报道信息中迅速地把握其主要内容,本文分析了事件要素对新闻主要内容的影响,结合新闻报道的基本原则和要求,提出了一种基于混合模型的事件要素提取方法.该方法首先对新闻数据中识别的实体进行加权,然后使用依存句法树分析实体在新闻事件中扮演的角色,并对关于要素的指代现象进行消解,最终融合频率及角色关系对实体加权的方法进行改进,有效地提取出新闻事件关联性较为重要的要素.实验结果表明,本文所述方法能够准确地提取出与新闻事件关联性较强的事件要素,提高了读者快速筛选新闻事件要素的效率.     

基于二值检测器的交通异常事件传感器设计

针对高速公路交通异常事件检测难度大、成本高等问题,提出了一种基于二值检测器的高速公路异常事件传感器的设计方法;利用二值检测器的特点,通过对SCM设计并结合状态识别算法获得车流量、时间占有率等交通状态参数;为了降低管控中心异常事件检测算法的设计难度,提高整体系统的运行效率,传感器选择性地提供必要的交通状态参数;最后管控中心通过对交通状态参数分析处理,实现异常事件的检测;仿真实验结果表明,该设计具备可行性。     

基于感知器的生物医学命名实体边界识别算法

在生物信息学领域内生物医学命名实体识别（Bio-NER）是生物医学文献挖掘、利用的基础工作,由于实体边界识别的困难导致目前Bio-NER效率较低,因此提出了基于感知器的实体边界识别算法,该算法采用随机梯度下降算法训练权重,利用token过滤器、n-gram模型及实体过滤器实现生物医学命名实体边界识别。在GENIA corpus 3.02语料库上进行的实验表明,该算法可以达到71.5%的准确率和79.2%的召回率,与相关工作相比均有一定提高。另外算法相对简单,识别算法速度较快,易在生产中应用。     

面向大规模数据的在线新事件检测

通过分析基于新闻要素的在线新事件检测算法的时间消耗,提出一种面向大规模数据环境的在线新事件检测算法。该算法利用基于倒排索引的高效相似报道搜索机制,有效减少单路径聚类算法中的相似度比较次数。通过对报道预处理、报道与事件比较以及索引搜索这3个过程的并行化,提高算法在多机环境下的运行效率和可伸缩性。实验结果表明,该算法在不影响漏检率和误检率的基础上,提高了新事件检测的速度,并且在千万到亿级别的报道规模下,其吞吐量达到150条/s~200条/s。     

基于灰色关联分析的中文新闻事件关联性识别

针对中文新闻事件关联性识别准确率较低的问题,提出一种基于灰色关联分析(GRA)的中文新闻事件关联性识别算法,该算法是一种多因素分析法。首先,通过分析中文新闻事件的特性,提出三个影响事件关联性的因素,分别为触发词的共现性、事件的共享名词以及事件句的相似度;其次,对多个影响因素进行量化处理,计算每个影响因素的影响权值;最后,运用GRA将多个影响因素结合在一起,建立事件之间的灰色关联性分析模型,实现事件关联性识别。通过实验验证了三个影响因素对事件关联性识别的有效性,而且相对于只考虑单一影响因素的关联性识别算法,所提算法提高了事件关联性识别的准确率。     

机场不正常事件实体检测与识别方法研究

民航安全自愿报告系统收集的海量故障报告以非结构化文本形式存储,不便于相关人员针对大量不正常事件加以分析并采取控制措施；命名实体识别技术可以将海量非结构化文本中的关键要素进行检测和识别,抽取成类别分明的结构化信息,作为进一步分析不正常事件并加以控制的基础工作；将机场不正常事件报告作为研究对象,提出了一种基于神经网络的中文命名实体识别模型,对文本进行了结构化处理；针对随机选用的训练样本一些实体类别分布比较稀疏和人工标注费时费力的问题,提出了基于模型预测分数的样本选择策略,实现了预标注样本的高效筛选；经过实验验证,该模型与BiLSTM_CRF模型、BiLSTM_self-attention_CRF模型相比F1值均提高了约6个百分点,该样本选择策略明显提高了人工标注效率,筛选出足够多的含有稀疏实体的样本。     

基于复杂事件处理机制的RFID数据流处理方法

射频识别（RFID）数据流处理技术中基于复杂事件处理机制（CEP）的方法受到业界广泛关注。针对射频识别数据的特点以及目前射频识别复杂事件处理方法的不足,通过扩展和改进高效模式匹配算法Rete,提出了复合事件处理方法ERD。该方法实现了事件检测机制与中间结果共享机制的有机统一,从而达到了提高系统整体检测效率的目的。仿真实验证明该方法具有较好的效果。     

基于数字图像处理的玻璃缺陷在线检测系统

研究了一种基于数字图像处理的玻璃缺陷在线检测系统,实现了对玻璃缺陷的实时检测、识别和分类,在缺陷提取上运用的是一种改进的归一化互相关算法,该算法可以显著提高图像的匹配速率和缺陷检测效率,在检测效率上比传统的算法提高了25.7%;在缺陷的识别分类上研究采用的是一种改进的支持向量机算法,该算法在检测精度和效率上都有较大的提高,在缺陷种类的识别上,其准确率可达到95%。     

一种中文地址类相似重复信息的检测方法

数据仓库中相似重复记录的识别与消除是数据清洗的热点问题,其中地址类信息对相同实体识别起着非常重要的作用.针对中文地址类信息的处理,建立了包含分词规则的元数据库,提出一种相似重复检测模型.在此基础上,描述了基于特征字符的分词算法和利用可变权值策略计算记录相似度的算法.实验结果表明该方法能有效解决中文地址类重复信息的检测,提高了算法的执行效率及检测精度.     

数据空间中一种灵活的集合式实体识别框架

数据空间是一个异构、动态的环境,具有随时间演化的特性.针对数据空间中随时间演化的、彼此间具有相互关联关系的大量异构实体,提出一个灵活的、基于演化的集合式实体识别框架TE-CER.TE-CER在实体识别过程中考虑时间信息和演化的影响,提出基于时间的blocking算法T-blocking和基于时间演化的聚类算法TE-Clustering.TE-Clustering聚类算法考虑了演化对实体相似度的影响,在相似度度量中引入属性演化系数和关联关系演化系数;在聚类过程中引入簇的时间约束.并且TE-CER提出基于识别顺序依赖图来确定集合式实体识别的识别顺序的策略,这样一种实体的识别结果可以用来后续另一种相关联实体的识别证据,以提高识别的准确率.在真实数据集上的大量实验结果表明所提出方法的有效性.     

增量采样聚类驱动的新闻事件发现

为获得更好的事件发现和代表性新闻抽取性能,引入数据集代表点采样聚类的视角,研究实现了一种事件发现及表示的集成分析方法。对于给定的新闻流数据,首先引入信息支撑度定义新闻间关系权重和事件关系权重,并通过引入双层近邻传播算法的迭代构建整体时间流上的单向事件内容支撑度网络,实现代表性新闻的分层增量采样,进一步考虑以最大相似度划分策略实现代表性新闻上的整体新闻流数据聚类。实验结果表明,相比于现有相关方法,新方法在大规模新闻流数据上具有显著的计算效率,可提取出新闻流中极有代表性的新闻,以及获得更好的新闻文档聚类质量,其热点事件发现结果与权威机构评选的重大新闻有极高吻合度。     

一种改进的回溯事件检测算法*

重点研究事件检测模型中层次聚类算法的改进,提出利用在关键词抽取基础上利用新闻的各种要素信息计算新闻之间相似度的方式,搭建了一个在线新闻检索系统,在其上利用新华社的新闻语料进行实验。实验结果表明改进方法的效果明显,性能较之未使用前有显著的提升。     

Event phase oriented news summarization

Event summarization is a task to generate a single, concise textual representation of an event. This task does not consider multiple development phases in an event. However, news articles related to long and complicated events often involve multiple phases. Thus, traditional approaches for event summarization generally have difficulty in capturing event phases in summarization effectively. In this paper, we define the task of Event Phase Oriented News Summarization (EPONS). In this approach, we assume that a summary contains multiple timelines, each corresponding to an event phase. We model the semantic relations of news articles via a graph model called Temporal Content Coherence Graph. A structural clustering algorithm EPCluster is designed to separate news articles into several groups corresponding to event phases. We apply a vertex-reinforced random walk to rank news articles. The ranking results are further used to create timelines. Extensive experiments conducted on multiple datasets show the effectiveness of our approach.     

融合多特征的新闻故事自动分割

新闻视频是由一系列的新闻故事构成的,准确地对新闻故事进行探测与分割将对新闻视频的自动检索与语义的理解产生重要作用。通过对新闻视频的结构特征进行分析,提出了融合静音、镜头切变、主持人特征和文本信息等多种特征的新闻故事探测与分割的方法。通过对不同的新闻视频进行实验,获得了平均95.2%的探测准确率。实验证明,提出的方法能够较好地解决新闻故事分割的任务。     

Event-based summarization using a centrality-as-relevance model

Event detection is a fundamental information extraction task, which has been explored largely in the context of question answering, topic detection and tracking, knowledge base population, news recommendation, and automatic summarization. In this article, we explore an event detection framework to improve a key phrase-guided centrality-based summarization model. Event detection is based on the fuzzy fingerprint method, which is able to detect all types of events in the ACE 2005 Multilingual Corpus. Our base summarization approach is a two-stage method that starts by extracting a collection of key phrases that will be used to help the centrality-as-relevance retrieval model. We explored three different ways to integrate event information, achieving state-of-the-art results in text and speech corpora: (1) filtering of nonevents, (2) event fingerprints as features, and (3) combination of filtering of nonevents and event fingerprints as features.     

Integrated approach of cognitive maps and neural networks using qualitative information on the World Wide Web: the KBNMiner

Abstract: In this paper, we propose a method for integrating cognitive maps and neural networks to gain competitive advantage using qualitative information acquired from news information on the World Wide Web. We have developed the KBNMiner, which is designed to represent the knowledge of domain experts with cognitive maps, to search and retrieve news information on the Internet according to the knowledge and to apply the information to a neural network model. In addition, we investigate ways to train neural networks more effectively by separating the learning data into two groups on the basis of event information acquired from news information. To validate our proposed method, we applied 180,000 news articles to the KBNMiner. The experimental results are found to support our proposed method through tenfold cross‐validation.     

Semantics-based information extraction for detecting economic events

As today’s financial markets are sensitive to breaking news on economic events, accurate and timely automatic identification of events in news items is crucial. Unstructured news items originating from many heterogeneous sources have to be mined in order to extract knowledge useful for guiding decision making processes. Hence, we propose the Semantics-Based Pipeline for Economic Event Detection (SPEED), focusing on extracting financial events from news articles and annotating these with meta-data at a speed that enables real-time use. In our implementation, we use some components of an existing framework as well as new components, e.g., a high-performance Ontology Gazetteer, a Word Group Look-Up component, a Word Sense Disambiguator, and components for detecting economic events. Through their interaction with a domain-specific ontology, our novel, semantically enabled components constitute a feedback loop which fosters future reuse of acquired knowledge in the event detection process.     

Event detection from online news documents for supporting environmental scanning

Environmental scanning, the acquisition and use of the information about events, trends, and relationships in an organization's external environment, permits an organization to adapt to its environment and to develop effective responses to secure or improve the organization's position in the future. Event detection technique that identifies the onset of new events from streams of news stories would facilitate the process of organization's environmental scanning. However, traditional event detection techniques generally adopted the feature co-occurrence approach that identifies whether a news story contains an unseen event by comparing the similarity of features between the new story and past news stories. Such feature-based event detection techniques greatly suffer from the word mismatch and inconsistent orientation problems and do not directly support event categorization and news stories filtering. In this study, we developed an information extraction-based event detection (NEED) technique that combines information extraction and text categorization techniques to address the problems inherent to traditional feature-based event detection techniques. Using a traditional feature-based event detection technique (i.e., INCR) as benchmarks, the empirical evaluation results showed that the proposed NEED technique improved the effectiveness of event detection measured by the tradeoff between miss and false alarm rates.     

Precluding Useless Events for On-Line Global Predicate Detections

Detecting global predicates is an important task in testing and debugging distributed programs. In this paper, we propose an approach that effectively precludes useless events for global predicate detections, facilitating the process of an independent on-line checking routine. To identify more useless events than a simple causality-check method can do, our method tracks and maintains the precedence information of event intervals as a graph. To reduce the potentially expensive space and time cost as the graph expands, we propose an effective scheme to prune the graph. The performance of our method is analyzed and evaluated by simulations. The result shows that our approach outperforms conventional approaches in terms of the number of useless events found.     

A text summary-based method to detect new events from streams of online news articles

New event detection (NED), which is crucial to firms’ environmental surveillance, requires timely access to and effective analysis of live streams of news articles from various online sources. These news articles, available in unprecedent frequency and quantity, are difficult to sift through manually. Most of existing techniques for NED are full-text-based; typically, they perform full-text analysis to measure the similarity between a new article and previous articles. This full-text-based approach is potentially ineffective, because a news article often contains sentences that are less relevant to define the focal event being reported and the inclusion of these less relevant sentences into the similarity estimation can impair the effectiveness of NED. To address the limitation of the full-text-based approach and support NED more effectively and efficiently, this study proposes and develops a summary-based event detection method that first selects relevant sentences of each article as a summary, then uses the resulting summaries to detect new events. We empirically evaluate our proposed method in comparison with some prevalent full-text-based techniques, including a vector space model and two deep-learning-based models. Our evaluation results confirm that the proposed method provides greater utilities for detecting new events from online news articles. This study demonstrates the value and feasibility of the text summarization approach for generating news article summaries for detecting new events from live streams of online news articles, proposes a new method more effective and efficient than the benchmark techniques, and contributes to NED research in several important ways.