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.     

基于指针网络的实体与关系联合抽取方法

针对现有实体和关系联合抽取方法中存在的实体与关系依赖建模不足、实体发生重叠难以抽取其所涉及的多个关系的问题,设计了基于深度学习的联合抽取框架。首先针对依赖建模不足问题,从预训练语料中提取实体共现特征,建模了实体间的潜在语义关系和实体与关系之间的依赖关系。其次提出了新颖的指针标注方法,该标注方法可以通过指针表示关系类别,由于任一实体可以被多个指针指向,所以可以在一段文本中标注重叠的实体并抽取多个实体—关系三元组结果。最后,为了有效利用单词的丰富语义和指针之间依赖的信息,设计了一个标签感知注意力机制,融合了包括来自编码层的字词信息、相关的共现语义信息。与研究中前沿的联合提取方法相比,该方法在百度DuIE测试集上实现了F₁值的增加。通过实验结果表明指针标注方法在一定程度上可以解决实体重叠问题。     

图像语义分析与理解综述

语义分析是图像理解中高层认知的重点和难点,存在图像文本之间的语义鸿沟和文本描述多义性两大关键问题。以图像本体的语义化为核心,在归纳图像语义特征及上下文表示的基础上,全面阐述生成法、判别法和句法描述法3种图像语义处理策略。总结语义词汇的客观基准和评价方法。最后指出图像语义理解的发展方向。     

一种基于span的实体和关系联合抽取方法

基于span的联合抽取模型在命名实体识别和关系抽取上取得了优异的效果。这些模型将文本span作为候选实体,并将span元组视为候选关系元组。span的语义表示在实体识别和关系分类中共享。然而现有基于span的模型无法很好地捕获这些候选实体和关系的语义,为了解决这些问题,提出了一种融合attention机制的span的联合抽取模型。特别地,attention用于计算相关语义表示,包括span特定特征语义表示和句子上下文的语义表示。实验结果表明,所提出的模型优于以前的模型,并在ACE2005、CoNLL2004和ADE 3个基准数据集上达到了当前最优的结果。     

基于语义网关的信息集成服务

基于语义网关的信息集成服务能够解决企业中不同部门间因不同数据资源中语义差异、结构差异等因素而引起的信息孤岛问题。这里实现了一个语义网关,通过它来实现将基于语义的查询转换成多个近义词、同义词相关的数据的查询,并对查询结果进行融合。     

TAPON-MT: A versatile framework for semantic labelling

Semantic labelling refers to the problem of assigning known labels to the elements of structured information from a source such as an HTML table or an RDF dump with unknown semantics. In the recent years it has become progressively more relevant due to the growth of available structured information in the Web of data that need to be labelled in order to integrate it in data systems. The existing approaches for semantic labelling have several drawbacks that make them unappealing if not impossible to use in certain scenarios: not accepting nested structures as input, being unable to label structural elements, not being customisable, requiring groups of instances when labelling, requiring matching instances to named entities in a knowledge base, not detecting numeric data, or not supporting complex features. In this article, we propose TAPON-MT, a framework for machine learning semantic labelling. Our framework does not have the former limitations, which makes it domain-independent and customisable. We have implemented it with a graphical interface that eases the creation and analysis of models, and we offer a web service API for their application. We have also validated it with a subset of the National Science Foundation awards dataset, and our conclusion is that TAPON-MT creates models to label information that are effective and efficient in practice.     

基于双向语义的中文实体关系联合抽取方法

现有中文实体关系抽取方法通常利用实体间的单向关系语义特征进行关系抽取，然而仅靠单向语义特征并不能完全利用实体间的语义关系，从而使得实体关系抽取的有效性受到影响。提出一种基于双向语义的中文实体关系联合抽取方法。利用RoBERTa预训练模型获取具有上下文信息的文本字向量表征，通过首尾指针标注识别句子中可能存在关系的实体。为了同时利用文本中的双向关系语义信息，将实体分别作为关系中的主体与客体来建立正负关系，并利用两组全连接神经网络构建正负关系映射器，从而对每一个输入实体同时从正关系与负关系的角度构建候选关系三元组。将候选关系三元组分别在正负关系下的概率分布序列与实体位置嵌入特征相结合，以对候选三元组进行判别，从而确定最终的关系三元组。在DuIE数据集上进行对比实验，结果表明，该方法的精确率与召回率优于MultiR、CoType等基线模型，其F1值达到0.805，相较基线模型平均提高了12.8%。     

一种基于部件CNN的网络安全命名实体识别方法

基于知识图谱的网络安全动态预警方法,能够主动感知和应对网络安全攻击,增强感知的实时性和精准性。然而,在构建网络安全知识图谱的实体抽取过程中,传统的命名实体识别工具和方法无法识别网络安全领域中的特定类别实体,文本中的未登录和中英文混合的网络安全实体也难以被准确识别。网络安全文本中的网络安全命名实体存在中英文混合、单词缩写等问题,仅基于字的命名实体识别方法难以充分表征字或词的语义信息。因此,论文考虑中英文更细粒度的部件语义捕捉字或词的语义特征,提出一种基于部件CNN的网络安全命名实体识别方法(C C-NS-NER),利用部件CNN抽取词语部件特征中的关键语义特征,丰富字词级别的语义信息,并引入BiLSTM-CRF确保抽取字向量和部件特征中的抽象信息,同时获取标签之间的关联信息,识别文本中的网络安全命名实体。在人工标注的网络安全数据集上的实验结果表明,该方法相较于传统模型,能有效获取字或词的部件语义信息,显著提高网络安全命名实体识别的效果。     

基于多维信息融合的知识库问答实体链接

知识库问答实体链接任务需要将问句内容精准链接到知识库中实体.当前方法大多难以兼顾链接实体的召回率和精确率,并且仅能根据文本信息对实体进行区分筛选.因此,文中在合并子步骤的基础上,提出融合多维度特征的知识库问答实体链接模型(MDIIEL).通过表示学习方法,将文本符号、实体和问句类型、实体在知识库中语义结构表达等信息整合并引至实体链接任务中,加强对相似实体的区分,在提高准确率的同时降低候选集的大小.实验表明,MDIIEL模型在实体链接任务性能上具有整体性提升,在大部分指标上取得较优的链接结果.     

Clustering rule bases using ontology-based similarity measures

Rules are increasingly becoming an important form of knowledge representation on the Semantic Web. There are currently few methods that can ensure that the acquisition and management of rules can scale to the size of the Web. We previously developed methods to help manage large rule bases using syntactical analyses of rules. This approach did not incorporate semantics. As a result, rule categorization based on syntactic features may not be effective. In this paper, we present a novel approach for grouping rules based on whether the rule elements share relationships within a domain ontology. We have developed our method for rules specified in the Semantic Web Rule Language (SWRL), which is based on the Web Ontology Language (OWL) and shares its formal underpinnings. Our method uses vector space modeling of rule atoms and an ontology-based semantic similarity measure. We apply a clustering method to detect rule relatedness, and we use a statistical model selection method to find the optimal number of clusters within a rule base. Using three different SWRL rule bases, we evaluated the results of our semantic clustering method against those of our syntactic approach. We have found that our new approach creates clusters that better match the rule bases’ logical structures. Semantic clustering of rule bases may help users to more rapidly comprehend, acquire, and manage the growing numbers of rules on the Semantic Web.     

Semantic interoperability with heterogeneous information systems on the internet through automatic tabular document exchange

Internet is a common information space populated with many entities (e.g., Internet of Things) with different information system types. Each of them has its own context of how to build and process documents (e.g., form documents). This leads to heterogeneous documents in terms of syntax and semantics, which are difficult to make information fusion from one context to another. To resolve this problem, this paper uses semantic interoperability technique which consists of two automatic stages including consistent data understanding and reasonable data usage. To implement semantic interoperability, this paper proposes a novel automatic tabular document exchange (DocEx) framework comprised of a new tabular document model (TabDoc) and a semantic inference scheme to fit the two stages above respectively. In this TabDoc model, a new Tabular Document Language (DocLang) as a communication medium between users and devices is provided, which is not only an information representation language but also a rule language for semantic inference as well. Abstract sub-tree-based semantic relations constructing the logical structure of a tabular document are separated from their presentational structures, clarifying the relationship between semantic groups (e.g., a cell or a block) with the help of a common dictionary CONEX. Besides, this paper proposes a semantic inference algorithm (SIA) executing the inference procedure on received tabular documents created by a Table Designer system which integrates with SIA. Finally, the proposed framework is applied to the processing of flight ticket booking in a realistic e-business scenario. The results show that the proposed method in this paper improves the performance of information fusion among different information systems on the Internet.     

A flexible semantic inference methodology to reason about user preferences in knowledge-based recommender systems

Recommender systems arose with the goal of helping users search in overloaded information domains (like e-commerce, e-learning or Digital TV). These tools automatically select items (commercial products, educational courses, TV programs, etc.) that may be appealing to each user taking into account his/her personal preferences. The personalization strategies used to compare these preferences with the available items suffer from well-known deficiencies that reduce the quality of the recommendations. Most of the limitations arise from using syntactic matching techniques because they miss a lot of useful knowledge during the recommendation process. In this paper, we propose a personalization strategy that overcomes these drawbacks by applying inference techniques borrowed from the Semantic Web. Our approach reasons about the semantics of items and user preferences to discover complex associations between them. These semantic associations provide additional knowledge about the user preferences, and permit the recommender system to compare them with the available items in a more effective way. The proposed strategy is flexible enough to be applied in many recommender systems, regardless of their application domain. Here, we illustrate its use in AVATAR, a tool that selects appealing audiovisual programs from among the myriad available in Digital TV.     

Magpie: Experiences in supporting Semantic Web browsing

Magpie has been one of the first truly effective approaches to bringing semantics into the web browsing experience. The key innovation brought by Magpie was the replacement of a manual annotation process by an automatically associated ontology-based semantic layer over web resources, which ensured added value at no cost for the user. Magpie also differs from older open hypermedia systems: its associations between entities in a web page and semantic concepts from an ontology enable link typing and subsequent interpretation of the resource. The semantic layer in Magpie also facilitates locating semantic services and making them available to the user, so that they can be manually activated by a user or opportunistically triggered when appropriate patterns are encountered during browsing. In this paper we track the evolution of Magpie as a technology for developing open and flexible Semantic Web applications. Magpie emerged from our research into user-accessible Semantic Web, and we use this viewpoint to assess the role of tools like Magpie in making semantic content useful for ordinary users. We see such tools as crucial in bootstrapping the Semantic Web through the automation of the knowledge generation process.     

Repeatable and reliable semantic search evaluation

An increasing amount of structured data on the Web has attracted industry attention and renewed research interest in what is collectively referred to as semantic search. These solutions exploit the explicit semantics captured in structured data such as RDF for enhancing document representation and retrieval, or for finding answers by directly searching over the data. These data have been used for different tasks and a wide range of corresponding semantic search solutions have been proposed in the past. However, it has been widely recognized that a standardized setting to evaluate and analyze the current state-of-the-art in semantic search is needed to monitor and stimulate further progress in the field. In this paper, we present an evaluation framework for semantic search, analyze the framework with regard to repeatability and reliability, and report on our experiences on applying it in the Semantic Search Challenge 2010 and 2011.     

A mobile query service for integrated access to large numbers of online semantic web data sources

From the Semantic Web’s inception, a number of concurrent initiatives have given rise to multiple segments: large semantic datasets, exposed by query endpoints; online Semantic Web documents, in the form of RDF files; and semantically annotated web content (e.g., using RDFa), semantic sources in their own right. In various mobile application scenarios, online semantic data has proven to be useful. While query endpoints are most commonly exploited, they are mainly useful to expose large semantic datasets. Alternatively, mobile RDF stores are utilized to query local semantic data, but this requires the design-time identification and replication of relevant data. Instead, we present a mobile query service that supports on-the-fly and integrated querying of semantic data, originating from a largely unused portion of the Semantic Web, comprising online RDF files and semantics embedded in annotated webpages. To that end, our solution performs dynamic identification, retrieval and caching of query-relevant semantic data. We explore several data identification and caching alternatives, and investigate the utility of source metadata in optimizing these tasks. Further, we introduce a novel cache replacement strategy, fine-tuned to the described query dataset, and include explicit support for the Open World Assumption. An extensive experimental validation evaluates the query service and its alternative components.     

Graph-ranking collective Chinese entity linking algorithm

Entity linking(EL)systems aim to link entity mentions in the document to their corresponding entity records in a reference knowledge base.Existing EL approaches usually ignore the semantic correlation between the mentions in the text,and are limited to the scale of the local knowledge base.In this paper,we propose a novel graphranking collective Chinese entity linking(GRCCEL)algorithm,which can take advantage of both the structured relationship between entities in the local knowledge base and the additional background information offered by external knowledge sources.By improved weighted word2vec textual similarity and improved PageRank algorithm,more semantic information and structural information can be captured in the document.With an incremental evidence mining process,more powerful discrimination capability for similar entities can be obtained.We evaluate the performance of our algorithm on some open domain corpus.Experimental results show the effectiveness of our method in Chinese entity linking task and demonstrate the superiority of our method over state-of-the-art methods.     

基于领域知识图谱的短文本实体链接

实体链接任务是识别文本中潜在的实体指称,并将其链接到给定知识库中无歧义的实体上。在绝大多数情况下,实体链接可能存在中文短文本缺乏有效上下文信息,导致存在一词多义的歧义现象;同时候选链接过程中,候选实体的不确定相关性也影响候选实体链接精确性。针对上述两个问题,提出深度神经网络与关联图相结合的实体链接模型。模型添加字符特征、上下文、信息深层语义来增强指称和实体表示,并进行相似度匹配。利用Fast-newman算法将图谱知识库聚类划分不同类型实体簇,将相似度计算得分最高候选实体所属实体簇映射到关系平面,构建聚类实体关联图。利用偏向随机游走算法考查候选实体之间语义相关度,计算指称与候选实体的匹配程度,输入链接实体。该模型可以实现短文本到知识图谱目标实体的准确链接。     

基于语义文法的地理实体位置关系的获取

地理信息与数据是客观知识世界的重要组成部分。研究如何从大量非结构化的信息中自动抽取地理实体位置关系具有重要意义。提出一种基于语义文法的地理实体位置关系获取方法,该方法可准确地从网页文本中获取多个地理实体之间的复合位置关系。首先,设计一种反映地理实体位置关系的语义文法GeoRSG。GeoRSG反映了地理实体位置关系的层次分类关系,并采用基于规则的方式刻画地理实体位置关系在文本中的语言表达方式。然后,实现地理实体位置关系解析器GeoRSG Parser。该解析器利用GeoRSG对文本进行解析,获得谓词表达形式的位置关系知识。实验结果显示,该方法从1000条语句中获取了81条三元和816条二元地理实体位置关系,并且取得了88.85%的正确率。     

Semantic data integration in hierarchical domains

A major challenge in building the Semantic Web is resolving differences among heterogeneous databases. This article describes one approach for handling semantic data integration problems in hierarchical domains. It also describes a declarative approach for specifying pairwise mappings between a centrally maintained ontology and each local data repository maintained by an autonomous agency. In this context, it outlines a method for specifying the mappings' semantics and encoding them to resolve heterogeneities. It focuses on XML-based applications in which entities in the centrally maintained ontology are hierarchically related to those in the local data repositories.