一种基于语义关系图的词语语义相关度计算模型

词语的语义计算是自然语言处理领域的重要问题之一,目前的研究主要集中在词语语义的相似度计算方面,对词语语义的相关度计算方法研究不够.为此,本文提出了一种基于语义词典和语料库相结合的词语语义相关度计算模型.首先,以HowNet和大规模语料库为基础,制定了相关的语义关系提取规则,抽取了大量的语义依存关系;然后,以语义关系三元组为存储形式,构建了语义关系图;最后,采用图论的相关理论,对语义关系图中的语义关系进行处理,设计了一个基于语义关系图的词语语义相关度计算模型.实验结果表明,本文提出的模型在词语语义相关度计算方面具有较好的效果,在WordSimilarity-353数据集上的斯皮尔曼等级相关系数达到了0.5358,显著地提升了中文词语语义相关度的计算效果.     

BEIRA: An Area-based User Interface for Map Services

This paper introduces BEIRA, an area-based map user interface for location-based contents. Recently, various web map services are widely used to search for location-based contents. However, browsing a large number of contents that are arranged on a map as points may be troublesome. We tackle this issue by using area-based representations instead of points. AOI (Area of Interest), which is core concept of BEIRA, is an arbitrary shaped area boundary with text summary information. With AOI, users can instantly grasp area characteristics without examining each point. AOI is deduced by performing geo-semantic co-clustering of location-based contents. Geo-semantic co-clustering takes both geographic and semantic features of contents into account. We confirm that the ratio of the geo-semantic blend is the key to deducing an appropriate boundary. We further propose and evaluate location-aware term weighting to obtain an informative summary.     

Fusing distributional and experiential information for measuring semantic relatedness

Models of semantic relatedness have usually focused on language-based distributional information without taking into account “experiential data” concerning the embodied sensorial source of the represented concepts. In this paper, we present an integrative cognitive model of semantic relatedness. The model – semantic family resemblance – uses a variation of the co-product as a mathematical structure that guides the fusion of distributional and experiential information. Our algorithm provides superior results in a set expansion task and a significant correlation with two benchmarks of human rated word-pair similarity datasets.     

DomESA: a novel approach for extending domain-oriented lexical relatedness calculations with domain-specific semantics

Being able to correctly model semantic relatedness between texts, and consequently the concepts represented by these texts, has become an important part of many intelligent information retrieval and knowledge processing systems. The need for such systems is especially evident within the biomedical domain, where the sheer amount of scientific publishing contributes to an information overflow. In this paper we present a novel method to approximate semantic relatedness in domain-focused settings. The approach is an extension to a well-known ESA (Explicit Semantic Analysis) method. Our extension successfully leverages the semantics of a domain-specific document corpus. We present the evaluation of the proposed method on a set of reference datasets, that are a de facto reference standard for the task of approximating biomedical semantic relatedness. The proposed method is evaluated in comparison with other state-of-the-art methods, as well as the baselines established with the original ESA method. The results of the experiments suggest that the proposed method combines the semantics of a general and domain-specific corpora to provide significant improvements over the original method.     

Taxonomy-based information content and wordnet-wiktionary-wikipedia glosses for semantic relatedness

Computing the semantic similarity/relatedness between terms is an important research area for several disciplines, including artificial intelligence, cognitive science, linguistics, psychology, biomedicine and information retrieval. These measures exploit knowledge bases to express the semantics of concepts. Some approaches, such as the information theoretical approaches, rely on knowledge structure, while others, such as the gloss-based approaches, use knowledge content. Firstly, based on structure, we propose a new intrinsic Information Content (IC) computing method which is based on the quantification of the subgraph formed by the ancestors of the target concept. Taxonomic measures including the IC-based ones consume the topological parameters that must be extracted from taxonomies considered as Directed Acyclic Graphs (DAGs). Accordingly, we propose a routine of graph algorithms that are able to provide some basic parameters, such as depth, ancestors, descendents, Lowest Common Subsumer (LCS). The IC-computing method is assessed using several knowledge structures which are: the noun and verb WordNet “is a” taxonomies, Wikipedia Category Graph (WCG), and MeSH taxonomy. We also propose an aggregation schema that exploits the WordNet “is a” taxonomy and WCG in a complementary way through the IC-based measures to improve coverage capacity. Secondly, taking content into consideration, we propose a gloss-based semantic similarity measure that operates based on the noun weighting mechanism using our IC-computing method, as well as on the WordNet, Wiktionary and Wikipedia resources. Further evaluation is performed on various items, including nouns, verbs, multiword expressions and biomedical datasets, using well-recognized benchmarks. The results indicate an improvement in terms of similarity and relatedness assessment accuracy.     

基于路径与深度的同义词词林词语相似度计算

该文提出了一种基于路径与深度的同义词词林词语语义相似度计算方法。该方法通过两个词语义项之间的最短路径以及它们的最近公共父结点在层次树中的深度计算出两个词语义项的相似度。在处理两个词语义项的最短路径与其最近公共父结点的深度时,为提高路径与深度计算的合理性,为分类树中不同层之间的边赋予不同的权值,同时通过两个义项在其最近公共父结点中的分支间距动态调节词语义项间的最短路径,从而平衡两个词语的相似度。该方法修正了目前相关算法只能得出几个固定的相似度值,所有最近公共父结点处于同一层次的义项对之间的相似度都相同的不合理现象,使词语语义相似度的计算结果更为合理。实验表明,该方法对MC30词对的相似度计算值与人工判定值相比,取得了0.856的皮尔逊相关系数,该结果高于目前大多数词语相似度算法与MC30的相关度。
     

Efficient region-aware large graph construction towards scalable multi-label propagation

With fast growing number of images on photo-sharing websites such as Flickr and Picasa, it is in urgent need to develop scalable multi-label propagation algorithms for image indexing, management and retrieval. It has been well acknowledged that analysis in semantic region level may greatly improve image annotation performance compared to that in the holistic image level. However, region level approach increases the data scale to several orders of magnitude and proposes new challenges to most existing algorithms. In this work, we present a novel framework to effectively compute pairwise image similarity by accumulating the information of semantic image regions. Firstly, each image is encoded as Bag-of-Regions based on multiple image segmentations. Secondly, all image regions are separated into buckets with efficient locality-sensitive hashing (LSH) method, which guarantees high collision probabilities for similar regions. The k-nearest neighbors of each image and the corresponding similarities can be efficiently approximated with these indexed patches. Lastly, the sparse and region-aware image similarity matrix is fed into the multi-label extension of the entropic graph regularized semi-supervised learning algorithm [1]. In combination they naturally yield the capability of handling large-scale dataset. Extensive experiments on NUS-WIDE (260k images) and COREL-5k datasets validate the effectiveness and efficiency of our proposed framework for region-aware and scalable multi-label propagation.     

一种基于图模型的维基概念相似度计算方法及其在实体链接系统中的应用

实体链接是指将文本中具有歧义的实体指称项链接到知识库中相应实体的过程。该文首先对实体链接系统进行了分析,指出实体链接系统中的核心问题—实体指称项文本与候选实体之间的语义相似度计算。接着提出了一种基于图模型的维基概念相似度计算方法,并将该相似度计算方法应用在实体指称项文本与候选实体语义相似度的计算中。在此基础上,设计了一个基于排序学习算法框架的实体链接系统。实验结果表明,相比于传统的计算方法,新的相似度计算方法可以更加有效地捕捉实体指称项文本与候选实体间的语义相似度。同时,融入了多种特征的实体链接系统在性能上获得了达到state-of-art的水平。     

A new semantic relatedness measurement using WordNet features

Computing semantic similarity/relatedness between concepts and words is an important issue of many research fields. Information theoretic approaches exploit the notion of Information Content (IC) that provides for a concept a better understanding of its semantics. In this paper, we present a complete IC metrics survey with a critical study. Then, we propose a new intrinsic IC computing method using taxonomical features extracted from an ontology for a particular concept. This approach quantifies the subgraph formed by the concept subsumers using the depth and the descendents count as taxonomical parameters. In a second part, we integrate this IC metric in a new parameterized multistrategy approach for measuring word semantic relatedness. This measure exploits the WordNet features such as the noun “is a” taxonomy, the nominalization relation allowing the use of verb “is a” taxonomy and the shared words (overlaps) in glosses. Our work has been evaluated and compared with related works using a wide set of benchmarks conceived for word semantic similarity/relatedness tasks. Obtained results show that our IC method and the new relatedness measure correlated better with human judgments than related works.     

Semantic and structural similarities between XML Schemas for integration of ubiquitous healthcare data

Currently, a lot of recent electronic health records are based on XML documents. In order to integrate these heterogeneous XML medical documents efficiently, studies on finding structure and semantic similarity between XML Schemas have been exploited. The main problem is how to harvest the most appropriate relatedness to combine two schemas as a global XML Schema for reusing and referring purposes. In this paper, we propose the novel resemblance measure that concurrently considers both structural and semantic information of two specific healthcare XML Schemas. Specifically, we introduce new metrics to compute the datatype and cardinality constraint similarities, which improve the quality of the semantic assessment. On the basis of the similarity between each element pair, we put forward an algorithm to calculate the similarity between XML Schema trees. Experimental results lead to the conclusion that our methodology provides better similarity values than the others with regard to the accuracy of semantic and structure similarities.     

Capturing contextual relationship for effective media search

One of the central problems regarding media search is the semantic gap between the low-level features computed automatically from media data and the human interpretation of them. This is because the notion of similarity is usually based on high-level abstraction but the low-level features do not sometimes reflect the human perception. In this paper, we assume the semantics of media is determined by the contextual relationship in a dataset, and introduce the method to capture the contextual information from a large media (especially image) dataset for effective search. Similarity search in an image database based on this contextual information shows encouraging experimental results.     

基于中文维基百科的词语语义相关度计算

语义相关度计算在信息检索、词义消歧、自动文摘、拼写校正等自然语言处理中均扮演着重要的角色。该文采用基于维基百科的显性语义分析方法计算汉语词语之间的语义相关度。基于中文维基百科,将词表示为带权重的概念向量,进而将词之间相关度的计算转化为相应的概念向量的比较。进一步,引入页面的先验概率,利用维基百科页面之间的链接信息对概念向量各分量的值进行修正。实验结果表明,使用该方法计算汉语语义相关度,与人工标注标准的斯皮尔曼等级相关系数可以达到0.52,显著改善了相关度计算的结果。     

Wiser: A semantic approach for expert finding in academia based on entity linking

We present Wiser, a new semantic search engine for expert finding in academia. Our system is unsupervised and it jointly combines classical language modeling techniques, based on text evidences, with the Wikipedia Knowledge Graph, via entity linking.Wiser indexes each academic author through a novel profiling technique which models her expertise with a small, labeled and weighted graph drawn from Wikipedia. Nodes in this graph are the Wikipedia entities mentioned in the author’s publications, whereas the weighted edges express the semantic relatedness among these entities computed via textual and graph-based relatedness functions. Every node is also labeled with a relevance score which models the pertinence of the corresponding entity to author’s expertise, and is computed by means of a proper random-walk calculation over that graph; and with a latent vector representation which is learned via entity and other kinds of structural embeddings derived from Wikipedia.At query time, experts are retrieved by combining classic document-centric approaches, which exploit the occurrences of query terms in the author’s documents, with a novel set of profile-centric scoring strategies, which compute the semantic relatedness between the author’s expertise and the query topic via the above graph-based profiles.The effectiveness of our system is established over a large-scale experimental test on a standard dataset for this task. We show that Wiser achieves better performance than all the other competitors, thus proving the effectiveness of modeling author’s profile via our “semantic” graph of entities. Finally, we comment on the use of Wiser for indexing and profiling the whole research community within the University of Pisa, and its application to technology transfer in our University.     

Geographic knowledge extraction and semantic similarity in OpenStreetMap

In recent years, a web phenomenon known as Volunteered Geographic Information (VGI) has produced large crowdsourced geographic data sets. OpenStreetMap (OSM), the leading VGI project, aims at building an open-content world map through user contributions. OSM semantics consists of a set of properties (called ‘tags’) describing geographic classes, whose usage is defined by project contributors on a dedicated Wiki website. Because of its simple and open semantic structure, the OSM approach often results in noisy and ambiguous data, limiting its usability for analysis in information retrieval, recommender systems and data mining. Devising a mechanism for computing the semantic similarity of the OSM geographic classes can help alleviate this semantic gap. The contribution of this paper is twofold. It consists of (1) the development of the OSM Semantic Network by means of a web crawler tailored to the OSM Wiki website; this semantic network can be used to compute semantic similarity through co-citation measures, providing a novel semantic tool for OSM and GIS communities; (2) a study of the cognitive plausibility (i.e. the ability to replicate human judgement) of co-citation algorithms when applied to the computation of semantic similarity of geographic concepts. Empirical evidence supports the usage of co-citation algorithms—SimRank showing the highest plausibility—to compute concept similarity in a crowdsourced semantic network.     

A parametric similarity method: Comparative experiments based on semantically annotated large datasets

We present the parametric method SemSim^p aimed at measuring semantic similarity of digital resources. SemSim^p is based on the notion of information content, and it leverages a reference ontology and taxonomic reasoning, encompassing different approaches for weighting the concepts of the ontology. In particular, weights can be computed by considering either the available digital resources or the structure of the reference ontology of a given domain. SemSim^p is assessed against six representative semantic similarity methods for comparing sets of concepts proposed in the literature, by carrying out an experimentation that includes both a statistical analysis and an expert judgment evaluation. To the purpose of achieving a reliable assessment, we used a real-world large dataset based on the Digital Library of the Association for Computing Machinery (ACM), and a reference ontology derived from the ACM Computing Classification System (ACM-CCS). For each method, we considered two indicators. The first concerns the degree of confidence to identify the similarity among the papers belonging to some special issues selected from the ACM Transactions on Information Systems journal, the second the Pearson correlation with human judgment. The results reveal that one of the configurations of SemSim^p outperforms the other assessed methods. An additional experiment performed in the domain of physics shows that, in general, SemSim^p provides better results than the other similarity methods.     

Evaluation of semantic similarity metrics applied to the automatic retrieval of medical documents: An UMLS approach

One promise of current information retrieval systems is the capability to identify risk groups for certain diseases and pathologies based on the automatic analysis of vast amounts of Electronic Medical Records repositories. However, the complexity and the degree of specialization of the language used by the experts in this context, make this task both challenging and complex. In this work, we introduce a novel experimental study to evaluate the performance of the two semantic similarity metrics (Path and Intrinsic IC-Path, both widely accepted in the literature) in a real-life information retrieval situation. In order to achieve this goal and due to the lack of methodologies for this context in the literature, we propose a straightforward information retrieval system for the biomedical field based on the UMLS Metathesaurus and on semantic similarity metrics. In contrast with previous studies which focus on testbeds with limited and controlled sets of concepts, we use a large amount of information (101,712 medical documents extracted from TREC Medical Records Track 2011). Our results show that in real-life cases, both metrics display similar performance, Path (F-Measure = 0.430) e Intrinsic IC-Path (F-Measure = 0.427). Thereby we suggest that the use of Intrinsic IC-Path is not justified in real scenarios.     

Coupling semantic and statistical techniques for dynamically enriching web ontologies

With the development of the Semantic Web technology, the use of ontologies to store and retrieve information covering several domains has increased. However, very few ontologies are able to cope with the ever-growing need of frequently updated semantic information or specific user requirements in specialized domains. As a result, a critical issue is related to the unavailability of relational information between concepts, also coined missing background knowledge. One solution to address this issue relies on the manual enrichment of ontologies by domain experts which is however a time consuming and costly process, hence the need for dynamic ontology enrichment. In this paper we present an automatic coupled statistical/semantic framework for dynamically enriching large-scale generic ontologies from the World Wide Web. Using the massive amount of information encoded in texts on the Web as a corpus, missing background knowledge can therefore be discovered through a combination of semantic relatedness measures and pattern acquisition techniques and subsequently exploited. The benefits of our approach are: (i) proposing the dynamic enrichment of large-scale generic ontologies with missing background knowledge, and thus, enabling the reuse of such knowledge, (ii) dealing with the issue of costly ontological manual enrichment by domain experts. Experimental results in a precision-based evaluation setting demonstrate the effectiveness of the proposed techniques.     

A hybrid approach for measuring semantic similarity based on IC-weighted path distance in WordNet

As a valuable tool for text understanding, semantic similarity measurement enables discriminative semantic-based applications in the fields of natural language processing, information retrieval, computational linguistics and artificial intelligence. Most of the existing studies have used structured taxonomies such as WordNet to explore the lexical semantic relationship, however, the improvement of computation accuracy is still a challenge for them. To address this problem, in this paper, we propose a hybrid WordNet-based approach CSSM-ICSP to measuring concept semantic similarity, which leverage the information content(IC) of concepts to weight the shortest path distance between concepts. To improve the performance of IC computation, we also develop a novel model of the intrinsic IC of concepts, where a variety of semantic properties involved in the structure of WordNet are taken into consideration. In addition, we summarize and classify the technical characteristics of previous WordNet-based approaches, as well as evaluate our approach against these approaches on various benchmarks. The experimental results of the proposed approaches are more correlated with human judgment of similarity in term of the correlation coefficient, which indicates that our IC model and similarity detection approach are comparable or even better for semantic similarity measurement as compared to others.