利用D-S证据理论进行特征融合的同义实体识别*

针对现实中同一实体存在不同表象的问题,提出一种基于D-S证据理论特征融合的同义实体识别方法。以搜索引擎为外部知识库获取实体特征信息,利用相似函数计算特征值,由D-S证据理论融合n个特征值,经阈值判断完成同义实体的识别。特征融合识别算法在医疗机构数据集上的识别精度、召回率和F值分别达到了85.80%、81.18%、83.43%,比单纯利用实体名的算法分别提高了4.09%、4.30%和4.21%。实验表明D-S证据理论将多特征融合,对同义实体识别具有更好的识别效果。     

一种基于SVM和AdaBoost的Web实体信息抽取方法

提出一种基于SVM和AdaBoost的Web实体信息抽取方法。首先提出一种基于SVM的Web页面主数据区域识别方法,基于Web实体实例在页面中的展示特征,有效地将Web页面进行数据区域分割,识别出Web实体实例所在的主数据区域;然后基于Web实体属性标签的特征,提出一种基于AdaBoost的集成学习方法,从页面的主数据区域自动地抽取Web实体信息。在两个真实数据集上进行实验,并与相关研究工作进行比较,实验结果说明该方法能够取得良好的抽取效果。     

一种基于数据挖掘的Deep Web模式匹配方法

模式匹配是Deep Web异构信息集成中的关键问题.介绍了一种整体性匹配方法,即同时发现大量模式,并一次性进行匹配.主要通过分析和比较两种已经存在的大规模模式匹配原型系统:MGS和DCM,结合它们核心算法的优点,提出一种新的基于数据挖掘技术的算法(Correlated-clustering).该算法先利用积极相关发现组匹配,再通过概念相似度的计算聚类同义属性,最后进行匹配选择.实验结果表明,本算法全面、效率高,充分体现了整体性方法的思想.     

一种应用于Deep Web环境下的重复记录识别模型

随着Web数据库数量和其蕴含数据量飞速的增长,对Deep Web数据的集成越来越成为研究领域关注的问题.然而由于Web上的信息以半结构化及无结构化的数据信息居多,导致了抽取的结果中包含诸多的不确定性.如有噪音数据、重复字符、简写与全称混合等问题.这给识别重复记录带来了很大不便,导致传统的去重算法无法达到很好的效果.为此,提出了一种面向deep Web结果整合的重复记录识别模型.在该模型中,提出了一种基于编辑距离的改进算法,基于该算法实现字符串匹配;通过构建属性匹配图,并采用二次确认机制实现识别重复记录.应用该模型,既提高了识别效率又保证了识别精确度,并通过实验证明了提出的算法和模型的可行性.     

元搜索引擎结果生成技术研究

元Web搜索引擎是通过将搜索请求传送给它所引用的搜索引擎,然后将这些搜索引擎返回的结果按照一定的结果集成算法合并,并将合并后的结果返回给用户.所用结果集成算法的好坏将直接影响该元搜索引擎的查询精度、查询完全度和响应速度.本文在分析常用的几种结果集成方法的基础上,提出几个改进的算法来改进搜索结果的一致性.     

一种基于语义及统计分析的Deep Web实体识别机制

分析了常见的实体识别方法,提出了一种基于语义及统计分析的实体识别机制(deep Web entity identification mechanism based on semantics and statistical analysis,简称SS-EIM),能够有效解决Deep Web数据集成中数据纠错、消重及整合等问题.SS-EIM主要由文本匹配模型、语义分析模型和分组统计模型组成,采用文本粗略匹配、表象关联关系获取以及分组统计分析的三段式逐步求精策略,基于文本特征、语义信息及约束规则来不断精化识别结果;根据可获取的有限的实例信息,采用静态分析、动态协调相结合的自适应知识维护策略,构建和完善表象关联知识库,以适应Web数据的动态性并保证表象关联知识的完备性.通过实验验证了SS-EIM中所采用的关键技术的可行性和有效性.     

基于用户兴趣的查询扩展语义模型

自然语言中词的同义现象和歧义现象一直是降低信息检索查全率和查准率的关键,在Web搜索引擎上显得更加突出。提出了一种基于用户兴趣的查询扩展语义模型,通过构建基于Yahoo的语义ontology知识库消除同义现象,设计客户端的用户兴趣挖掘模型消除歧义现象。实验结果显示该方法能有效提高Web信息检索的查全率与查准率。     

基于实例的强分类器快速集成方法

针对集成分类器由于基分类器过弱,需要牺牲大量训练时间才能取得高精度的问题,提出一种基于实例的强分类器快速集成方法——FSE。首先通过基分类器评价方法剔除不合格分类器,再对分类器进行精确度和差异性排序,从而得到一组精度最高、差异性最大的分类器;然后通过FSE集成算法打破已有的样本分布,重新采样使分类器更多地关注难学习的样本,并以此决定各分类器的权重并集成。实验通过与集成分类器Boosting在UCI数据库和真实数据集上进行比对,Boosting构造的集成分类器的识别精度最高分别能达到90.2%和90.4%,而使用FSE方法的集成分类器精度分别能达到95.6%和93.9%;而且两者在达到相同精度时,使用FSE方法的集成分类器分别缩短了75%和80%的训练时间。实验结果表明,FSE集成模型能有效提高识别精度、缩短训练时间。     

基于本体的Web信息抽取系统

为了解决已有信息抽取系统中方法不具有重用性及不能抽取语义信息的问题,提出了一个基于领域本体的面向主题的Web信息抽取框架.对Web中文页面,借助外部资料,利用本体解析信息,对文件采集及预处理中的源文档及信息采集、文档预处理、文档存储等技术进行了分析设计,提出了文本转换中的分词及词表查询和命名实体识别算法,并给出了一种知识抽取方案.实验结果表明,该方法可以得到性能较高的抽取结果.     

Deep Web数据源聚类与分类

随着Internet信息的迅速增长,许多Web信息已经被各种各样的可搜索在线数据库所深化,并被隐藏在Web查询接口下面.传统的搜索引擎由于技术原因不能索引这些信息--Deep Web信息.本文分析了Deep Web查询接口的各种类型,研究了基于查询接口特征的数据源聚类方法和基于聚类结果的数据源分类方法,讨论了从基于规则与线性文档分类器中抽取查询探测集的规则抽取算法和Web文档数据库分类的查询探测算法.     

Exploiting semantic annotations for open information extraction: an experience in the biomedical domain

The increasing amount of unstructured text published on the Web is demanding new tools and methods to automatically process and extract relevant information. Traditional information extraction has focused on harvesting domain-specific, pre-specified relations, which usually requires manual labor and heavy machinery; especially in the biomedical domain, the main efforts have been directed toward the recognition of well-defined entities such as genes or proteins, which constitutes the basis for extracting the relationships between the recognized entities. The intrinsic features and scale of the Web demand new approaches able to cope with the diversity of documents, where the number of relations is unbounded and not known in advance. This paper presents a scalable method for the extraction of domain-independent relations from text that exploits the knowledge in the semantic annotations. The method is not geared to any specific domain (e.g., protein–protein interactions and drug–drug interactions) and does not require any manual input or deep processing. Moreover, the method uses the extracted relations to compute groups of abstract semantic relations characterized by their signature types and synonymous relation strings. This constitutes a valuable source of knowledge when constructing formal knowledge bases, as we enable seamless integration of the extracted relations with the available knowledge resources through the process of semantic annotation. The proposed approach has successfully been applied to a large text collection in the biomedical domain and the results are very encouraging.     

命名实体识别、排歧和跨语言关联

命名实体是文本中承载信息的重要语言单位,命名实体的识别和分析在网络信息抽取、网络内容管理和知识工程等领域都占有非常重要的地位。有关命名实体的研究任务包括实体识别、实体排歧、实体跨语言关联、实体属性抽取、实体关系检测等,该文重点介绍命名实体识别、排歧和跨语言关联等任务的研究现状,包括难点、评测、现有方法和技术水平,并对下一步需要重点解决的问题进行分析和讨论。该文认为,命名实体识别、排歧和跨语言关联目前的技术水平还远远不能满足大规模真实应用的需求,需要更加深入的研究。在研究方法上,要突破自然语言文本的限制,直接面向海量、冗余、异构、不规范、含有大量噪声的网页信息处理。     

基于结构分析和实体识别的信息集成

针对海量的web数据，提出了一种基于文档结构分析和实体识别的web信息提取和集成方法，利用XML强大的数据描述能力，灵活组织集成的web文档信息内容．方法首先将半结构化的HTML文档转化成具有模式结构的XML文档，然后使用实体识别的技术对不同主题区域进一步抽取出格式良好的数据，最后将得到的多数据类型的信息集成到数据库中，以支持进一步的分析和查询．实验结果证明了该方法的实用和有效性．     

A novel digital information service for federating distributed digital entities

We investigate the performance and the scalability metrics of a Digital Information Service framework that is used for unifying and federating online digital entities by retrieving and managing information located on the web. The Digital Information Service consists of tools and web services for supporting Cyberinfrastructure based scientific research. This system supports a number of existing online Web 2.0 research tools (social bookmarking, academic search, scientific databases, journal and conference content management systems) and aims to develop added-value community building tools that leverage the management and federation of digital entities and their metadata obtained from multiple services. We introduce a prototype implementation and present its evaluation. As the results indicate, the proposed system achieves federation and unification of digital entities coming from different sources with negligible processing overheads.     

Quality-aware similarity assessment for entity matching in Web data

One of the key challenges to realize automated processing of the information on the Web, which is the central goal of the Semantic Web, is related to the entity matching problem. There are a number of tools that reliably recognize named entities, such as persons, companies, geographic locations, in Web documents. The names of these extracted entities are, however, non-unique; the same name on different Web pages might or might not refer to the same entity. The entity matching problem concerns of identifying the entities, which are referring to the same real-world entity. This problem is very similar to the entity resolution problem studied in relational databases, however, there are also several differences. Most importantly Web pages often only contain partial or incomplete information about the entities.Similarity functions try to capture the degree of belief about the equivalence of two entities, thus they play a crucial role in entity matching. The accuracy of the similarity functions highly depends on the applied assessment techniques, but also on some specific features of the entities. We propose systematic design strategies for combined similarity functions in this context. Our method relies on the combination of multiple evidences, with the help of estimated quality of the individual similarity values and with particular attention to missing information that is common in Web context. We study the effectiveness of our method in two specific instances of the general entity matching problem, namely the person name disambiguation and the Twitter message classification problem. In both cases, using our techniques in a very simple algorithmic framework we obtained better results than the state-of-the-art methods.     

Web信息集成的研究

在信息和服务迅猛增长的Web上构建信息集成系统是当前的一个研究热点，本文针对web信息集成的目标，结合XML和RDF等主要的Web技术及本体的概念，构建了一个基于本体的查询-应答框架，它通过有效地集成Web上的各种异构数据源来为终端用户提供更为方便的服务和应用。     

Web信息集成的研究

在信息和服务迅猛增长的Web上构建信息集成系统是当前的一个研究热点,本文针对Web信息集成的目标,结合XML和RDF等主要的Web技术及本体的概念,构建了一个基于本体的查询-应答框架,它通过有效地集成Web上的各种异构数据源来为终端用户提供更为方便的服务和应用。     

Identification and resolution of conflicts during ontological integration using rules

Abstract: Integration of ontologies of information sources and consumers is an important phase in achieving web‐based interoperability. The present work describes an approach for identifying certain semantic conflicts while integrating ontologies of heterogeneous information sources. This paper is focused on the identification of homonymy and synonymy between elements in ontologies. In the present work the concepts of homonymy and synonymy are synonymous to naming conflicts and entity identifier conflicts, respectively, and partial synonymy is synonymous to schema isomorphism conflicts. The concept of the mask of interoperability is introduced for the identification of synonymy. The mask of interoperability is expressed in a declarative way as a set of rules, which can then be used for resolution of conflicts during integration of ontologies. As proof of concept, ontologies are implemented using the XML‐based ontology language Ontology Web Language (OWL), and the rules are implemented using the emerging rule language Semantic Web Rule Language (SWRL). This representation in OWL and SWRL allows the ontology to be executable, flexibly extendable and platform‐independent. The OWL facts and SWRL rules are used by the Jess and Bossam reasoning engine to identify semantic homonymy and synonymy.     

健康领域Web信息抽取

针对Web信息抽取(WIE)技术在健康领域应用的问题,提出了一种基于WebHarvest的健康领域Web信息抽取方法。通过对不同健康网站的结构分析设计健康实体的抽取规则,实现了基于WebHarvest的自动抽取健康实体及其属性的算法;再把抽取的实体及其属性进行一致性检查后存入关系数据库中,然后对关系数据库中隐含健康实体的属性值利用Ansj自然语言处理方法进行实体识别, 进而抽取健康实体之间的联系。该技术在健康实体抽取实验中,平均F值达到99.9%,在实体联系抽取实验中,平均F值达到80.51%。实验结果表明提出的Web信息抽取技术在健康领域抽取的健康信息具有较高的质量和可信性。