SVM-based ontology matching approach

There are a lot of heterogeneous ontologies in semantic web, and the task of ontology mapping is to find their semantic relationship. There are integrated methods that only simply combine the similarity values which are used in current multi-strategy ontology mapping. The semantic information is not included in them and a lot of manual intervention is also needed, so it leads to that some factual mapping relations are missed. Addressing this issue, the work presented in this paper puts forward an ontology matching approach, which uses multi-strategy mapping technique to carry on similarity iterative computation and explores both linguistic and structural similarity. Our approach takes different similarities into one whole, as a similarity cube. By cutting operation, similarity vectors are obtained, which form the similarity space, and by this way, mapping discovery can be converted into binary classification. Support vector machine (SVM) has good generalization ability and can obtain best compromise between complexity of model and learning capability when solving small samples and the nonlinear problem. Because of the said reason, we employ SVM in our approach. For making full use of the information of ontology, our implementation and experimental results used a common dataset to demonstrate the effectiveness of the mapping approach. It ensures the recall ration while improving the quality of mapping results.     

Bridging Real World Semantics to Model World Semantics for Taxonomy Based Knowledge Representation System

As a mean to map ontology concepts, a similarity technique is employed. Especially a context dependent concept mapping is tackled, which needs contextual information from knowledge taxonomy. Context-based semantic similarity differs from the real world similarity in that it requires contextual information to calculate similarity. The notion of semantic coupling is introduced to derive similarity for a taxonomy-based system. The semantic coupling shows the degree of semantic cohesiveness for a group of concepts toward a given context. In order to calculate the semantic coupling effectively, the edge counting method is revisited for measuring basic semantic similarity by considering the weighting attributes from where they affect an edge's strength. The attributes of scaling depth effect, semantic relation type, and virtual connection for the edge counting are considered. Furthermore, how the proposed edge counting method could be well adapted for calculating context-based similarity is showed. Thorough experimental results are provided for both edge counting and context-based similarity. The results of proposed edge counting were encouraging compared with other combined approaches, and the context-based similarity also showed understandable results. The novel contributions of this paper come from two aspects. First, the similarity is increased to the viable level for edge counting. Second, a mechanism is provided to derive a context-based similarity in taxonomy-based system, which has emerged as a hot issue in the literature such as Semantic Web, MDR, and other ontology-mapping environments.     

A Recommendation System Keeping Both Precision and Recall by Extraction of Uninteresting Information

A recommendation system which recommends interesting information to the target user must guarantee high precision and recall. However, there is trade-off between precision and recall. In this paper, we propose a web page recommendation method balancing both of them by take advantage of uninteresting information. The proposed method extracts the interest and uninterest indicators from not only historical interesting web pages but also uninteresting ones in a target genre. The historical interesting and uninteresting information is derived based on the browsing time and bookmarking. The proposed method can keep precision and recall by excluding the uninteresting information from the recommended ones based on the interest and uninterest indicators. The experimental result proved that the proposed method can improve the precision and recall than an existing method.     

A Novel Web Video Event Mining Framework with the Integration of Correlation and Co-Occurrence Information

The massive web videos prompt an imperative demand on efficiently grasping the major events.However, the distinct characteristics of web videos, such as the limited number of features, the noisy text information, and the unavoidable error in near-duplicate keyframes (NDKs) detection, make web video event mining a challenging task.In this paper, we propose a novel four-stage framework to improve the performance of web video event mining.Data preprocessing is the first stage.Multiple Correspondence Analysis (MCA) is then applied to explore the correlation between terms and classes, targeting for bridging the gap between NDKs and high-level semantic concepts.Next, co-occurrence information is used to detect the similarity between NDKs and classes using the NDK-within-video information.Finally, both of them are integrated for web video event mining through negative NDK pruning and positive NDK enhancement.Moreover, both NDKs and terms with relatively low frequencies are treated as useful information in our experiments.Experimental results on large-scale web videos from YouTube demonstrate that the proposed framework outperforms several existing mining methods and obtains good results for web video event mining.     

Tensor semantic model for an audio classification system

Audio resources are a very important part of multimedia information.The classification effect of audio is directly related to the service mode of personal resource management systems.At present,vector features have been widely used in audio classification systems.However,some semantic correlations among different audio information can not be completely expressed by simple vector representation.Tensors are multidimensional matrices,and their mathematical expansion and application can express multi-semantic information.The tensor uniform content locator(TUCL) is proposed as a means of expressing the semantic information of audio,and a three-order tensor semantic space is constructed according to the semantic tensor.Tensor semantic dispersion(TSD) can aggregate some audio resources with the same semantics and,at the same time,its automatic classification can be accomplished by calculating the TSD.In order to effectively utilize TSD classification information,a radial basis function tensor neural network(RBFTNN) is constructed and used to train an intelligent learning model.Experimental results show that the tensor model can significantly improve the classification precision under multi-semantic classification requests within an information resource management system.     

Visual Ontology Construction for Digitized Art Image Retrieval

Current investigations on visual information retrieval are generally content-based methods. The significant difference between similarity in low-level features and similarity in high-level semantic meanings is still a major challenge in the area of image retrieval. In this work, a scheme for constructing visual ontology to retrieve art images is proposed. The proposed ontology describes images in various aspects, including type ＆ style, objects and global perceptual effects. Concepts in the ontology could be automatically derived. Various art image classification methods are employed based on low-level image features. Non-objective semantics are introduced, and how to express these semantics is given. The proposed ontology scheme could make users more naturally find visual information and thus narrows the “semantic gap”. Experimental implementation demonstrates its good potential for retrieving art images in a human-centered manner.     

Web-Based Semantic Similarity: An Evaluation in the Biomedical Domain

Computation of semantic similarity between concepts is a very common problem in many language related tasks and knowledge domains. In the biomedical field, several approaches have been developed to deal with this issue by exploiting the structured knowledge available in domain ontologies (such as SNOMED-CT or MeSH) and specific, closed and reliable corpora (such as clinical data). However, in recent years, the enormous growth of the Web has motivated researchers to start using it as the corpus to assist semantic analysis of language. This paper proposes and evaluates the use of the Web as background corpus for measuring the similarity of biomedical concepts. Several ontology-based similarity measures have been studied and tested, using a benchmark composed by biomedical terms, comparing the results obtained when applying them to the Web against approaches in which specific clinical data were used. Results show that the similarity values obtained from the Web for ontology-based measures are at least and even more reliable than those obtained from specific clinical data, showing the suitability of the Web as information corpus for the biomedical domain.     

GDM: A New Graph Based Data Model Using Functional Abstractionx

In this paper, a Graph-based semantic Data Model (GDM) is proposed with the primary objective of bridging the gap between the human perception of an enterprise and the needs of computing infrastructure to organize information in some particular manner for efficient storage and retrieval. The Graph. Data Model (GDM) has been proposed as an alternative data model to combine the advantages of the relational model with the positive features of semantic data models. The proposed GDM offers a structural representation for interacting to the designer, making it always easy to comprehend the complex relations amongst basic data items. GDM allows an entire database to be viewed as a Graph (V, E) in a layered organization. Here, a graph is created in a bottom up fashion where V represents the basic instances of data or a functionally abstracted module, called primary semantic group (PSG) and secondary semantic group (SSG). An edge in the model implies the relationship among the secondary semantic groups. The contents of the lowest layer are the semantically grouped data values in the form of primary semantic groups. The SSGs are nothing but the higher-level abstraction and are created by the method of encapsulation of various PSGs, SSGs and basic data elements. This encapsulation methodology to provide a higher-level abstraction continues generating various secondary semantic groups until the designer thinks that it is sufficient to declare the actual problem domain. GDM, thus, uses standard abstractions available in a semantic data model with a structural representation in terms of a graph. The operations on the data model are formalized in the proposed graph algebra. A Graph Query Language (GQL) is also developed, maintaining similarity with the widely accepted user-friendly SQL. Finally, the paper also presents the methodology to make this GDM compatible with the distributed environment, and a corresponding query processing technique for distributed environment is also suggested for the sake of completeness.     

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.     

An approach to implementing an intelligent web search

The retrieval efficiency of the presently used search tools cannot be significantly improved： A ＂bag of words＂ interpretation causes loosing semantics of texts. The functional approach to present English texts in the memory of computers makes it possible to keep semantic relations between words. These relations can be taken into account when indexing documents and when performing searching. Utilizing this approach, it is possible to use a natural language to express user queries. In many cases, this way is more usual for users to describe their information needs compared to the keyword style. The factoid question answering task is one of the possible its applications. Key components of the prototype of a system utilizing this approach are discussed.     

利用N-gram和语义分析的维吾尔语文本相似性检测方法

为了实现维吾尔语文本的相似性检测,提出一种基于N-gram和语义分析的相似性检测方法。根据维吾尔语单词特征,采用了N-gram统计模型来获得词语,并根据词语在文本中的出现频率来构建词语—文本关系矩阵,并作为文本模型。采用了潜在语义分析（LSA）来获得词语及其文本之间的隐藏关联,以此解决维吾尔语词义模糊的问题,并获得准确的相似度。在包含重组和同义词替换的剽窃文本集上进行实验,结果表明该方法能够准确有效地检测出相似性。     

科研项目同行评议专家学术专长匹配方法

现有的评审专家推荐过程通常依赖于人工匹配,在进行专家推荐时不能充分捕捉评审项目所属学科与专家研究兴趣之间的语义关联,导致专家推荐的精确性较低。为解决这个问题,提出了一种科研项目同行评议专家学术专长匹配方法。该方法构建学术网络以建立学术实体联系,并设计元路径捕捉学术网络中不同节点间的语义关联;使用随机游走策略获得项目所属学科与专家研究兴趣共现关联的节点序列,并通过网络表示学习模型训练得到具有语义关联的项目所属学科与专家研究兴趣的向量表示;在此基础上,按照项目学科树层次结构逐层计算语义相似度,以实现多粒度的同行评议学术专长匹配。在爬取的知网和万方论文数据集、某专家评审数据集、以及百度百科词向量数据集上得到的实验结果表明,所提方法能提升项目所属学科与专家研究兴趣间的语义关联,并能有效应用于项目评审专家的学术专长匹配。     

一种基于本体语义的信息检索模型

对基于向量空间模型的检索方法进行改进,提出基于本体语义的信息检索模型。将WordNet词典作为参照本体来计算概念之间的语义相似度,依据查询中标引项之间的相似度,对查询向量中的标引项进行权值调整,并参照Word-Net本体对标引项进行同义和上下位扩展,在此基础上定义查询与文档间的相似度。与传统的基于词形的信息检索方法相比,该方法可以提高语义层面上的检索精度。     

包含关联的语义覆盖网构建方法研究

现实世界中信息资源之间存在着各种各样的关联关系,而当前的搜索引擎只能提供基于关键字的搜索,不能为用户提供他们所关心的与关键字相关的各类信息。针对这个问题,提出了构建语义覆盖网,以用户需求为导向,以用户所关心的信息为中心,将所有与此资源相关的信息全部汇聚起来提供给用户。这里先根据语义相似度将节点聚类,然后在聚类的基础上,根据各类关联关系构建基于关联关系的语义覆盖网。     

基于Python智能机器人多渠道知识库推送仿真

机器人知识推送是智能化发展的必然产物,当前相关研究成果存在召回率和推送结果用户满意度较低的问题,提出基于Python的智能机器人多渠道知识库推送方法。利用离线和在线的方式对访问和浏览信息进行识别,离线信息识别中,对采集到的信息结构进行分析,检测出信息的特征,动态添加字符串,将得到的特征与关键词知识库中的数据特征进行配准,判断出是否识别关键词;在线信息识别中,基于Python语言,分别结合百度云识别和云聊天以及百度云语音三个体系,实现信息的在线识别。利用信息语义相似度给出知识库推送的详细过程,对义项之间的相似程度进行计算,获取关键字或关键词的相似程度,将相似度比设定阈值大的信息保存起来,将此类信息推送给使用者。实验结果表明,上述方法查全率和用户满意度均较高,是一种可行性很强的知识库推送方法。     

Semantic similarity measurement using historical google search patterns

Computing the semantic similarity between terms (or short text expressions) that have the same meaning but which are not lexicographically similar is an important challenge in the information integration field. The problem is that techniques for textual semantic similarity measurement often fail to deal with words not covered by synonym dictionaries. In this paper, we try to solve this problem by determining the semantic similarity for terms using the knowledge inherent in the search history logs from the Google search engine. To do this, we have designed and evaluated four algorithmic methods for measuring the semantic similarity between terms using their associated history search patterns. These algorithmic methods are: a) frequent co-occurrence of terms in search patterns, b) computation of the relationship between search patterns, c) outlier coincidence on search patterns, and d) forecasting comparisons. We have shown experimentally that some of these methods correlate well with respect to human judgment when evaluating general purpose benchmark datasets, and significantly outperform existing methods when evaluating datasets containing terms that do not usually appear in dictionaries.     

SyMSS: A syntax-based measure for short-text semantic similarity

Sentence and short-text semantic similarity measures are becoming an important part of many natural language processing tasks, such as text summarization and conversational agents. This paper presents SyMSS, a new method for computing short-text and sentence semantic similarity. The method is based on the notion that the meaning of a sentence is made up of not only the meanings of its individual words, but also the structural way the words are combined. Thus, SyMSS captures and combines syntactic and semantic information to compute the semantic similarity of two sentences. Semantic information is obtained from a lexical database. Syntactic information is obtained through a deep parsing process that finds the phrases in each sentence. With this information, the proposed method measures the semantic similarity between concepts that play the same syntactic role. Psychological plausibility is added to the method by using previous findings about how humans weight different syntactic roles when computing semantic similarity. The results show that SyMSS outperforms state-of-the-art methods in terms of rank correlation with human intuition, thus proving the importance of syntactic information in sentence semantic similarity computation.     

中文文本的信息自动抽取和相似检索机制

目前信息抽取成为提供高质量信息服务的重要手段,提出面向中文文本信息的自动抽取和相似检索机制,其基本思想是将用户兴趣表示为语义模板,对关键字进行概念扩充,通过搜索引擎获得初步的候选文本集合,在概念触发机制和部分分析技术基础上,利用语义关系到模板槽的映射机制,填充文本语义模板,形成结构化文本数据库.基于文本数据表述的模糊性,给出用户查询与文本语义模板的相似关系,实现了相似检索,可以更加全面地满足用户的信息需求.     

以本体构造中文信息过滤中的需求模型

在信息过滤系统中,用户模板是机器可理解的用户需求表示形式,是否能准确地反映出用户的真实需求将直接影响着过滤系统的性能。在向量空间模型中,用户的模板表现为一组带权重的特征词集,但由于在这样的用户模板中缺少必要的语义信息,很难准确地反映出用户的需求。本文提出了以本体构造需求模板的方法,以本体的形式定义需求中概念间的语义关联关系,将向量空间模型中的特征向量定义为本体中的实例,通过实例间的关联路径计算特征项间的语义关联,并通过特征项间的语义关联计算出文档与模板的语义关联度。     

语义模型在Overlay网络服务发现机制中的应用

提出一种新型Overlay网络服务发现机制,该机制充分利用概念之间的语义相似性,从语义概念树模型扩展概念相似度的计算,并将该语义概念树模型应用到Overlay网络服务发现机制的研究中。实验证明,该语义概念模型能够根据用户的想法和查询词条的内在含义进行相近语义短语的查找,实现匹配机制的语义化,提高了服务发现的查准率。