Set Of Triples Research Articles

Rank bounds for design matrices with block entries and geometric applications

Rank bounds for design matrices with block entries and geometric applications, Discrete Analysis 2018:5, 24 pp. It is not hard to prove, when the statement is suitably formulated, that if $A$ is a random matrix, then it is not possible to reduce the rank of $A$ by changing only a few entries. However, it is remarkably hard to find an explicit example of such a matrix. It turns out that explicit examples would have major applications in theoretical computer science, and therefore the matrix-rigidity problem is a central question in the area. In order to make progress on this problem, one can ask a more basic question: are there any combinatorial conditions on a matrix that guarantee that it has a high rank? An interesting condition has recently been identified that does not solve the matrix-rigidity problem but that does have several applications, in particular to combinatorial geometry. Roughly speaking, the condition is that there should not be many non-zero entries in any row or too few non-zero entries in any column, and for any two rows the number of columns for which the entries in both rows are non-zero is small. Such matrices are known as _design matrices_, since the conditions resemble those for a design, though they are significantly looser. The proof that such matrices have high rank uses a beautiful technique called _matrix scaling_. The rough idea is to multiply rows and columns by non-zero scalars until the entries are of broadly the same magnitude, and then to show that the matrix $A^*A$ is dominated by its diagonal. From this it follows fairly easily that $A^*A$ has high rank, and therefore that $A$ does as well. The argument can be found in [a paper of Barak, Dvir, Wigderson and Yehudayoff](https://arxiv.org/pdf/1009.4375.pdf) that this paper generalizes. Such rank bounds are useful, because there are several problems in combinatorial geometry that lead naturally to design matrices. For example, the Sylvester-Gallai theorem and its generalizations concern families of points and collinearity relations between them. If the points are $v_1,\dots,v_n$, then one can write them as column vectors and put them together to form a matrix $M$. Three of the points will be collinear if there is a linear dependence amongst the columns of this matrix with three non-zero coefficients, or equivalently a vector $w$ with three non-zero entries such that $Mw=0$. Under suitable hypotheses, one can use such vectors $w$ to create a design matrix and apply rank bounds on the design matrix to deduce combinatorial consequences about the set of points $v_1,\dots,v_n$. The generalization in this paper concerns design matrices whose entries are themselves matrices, or blocks. Rank bounds are proved for such matrices, which lead to new applications in combinatorial geometry. One of these applications is to the following question. Let $T$ be a set of triples of elements of $\{1,2,\dots,n\}$ and let $V(T)$ be the variety of all sequences $(v_1,\dots,v_n)$ of points in $\mathbb C^d$ such that for every triple $\{i,j,k\}\in T$ the three points $v_i, v_j$ and $v_k$ are collinear. Given a (non-singular) sequence $V\in V(T)$, how many degrees of freedom does it have? A trivial lower bound is 8, resulting from the fact that projective transformations preserve collinearity. The authors prove a general result that has as a consequence that if every pair $\{i,j\}$ belongs to exactly one triple in $T$ and no line contains more than half the points of $V$, then the number of degrees of freedom of $V$ is at most 15. A more precise statement, as well as other interesting applications, can be found in the paper.

An experimental study of the effective parameters on automatic line-break control valves action in natural gas pipelines

The automatic line-break control valve (LBV) is a robust self-contained pipeline monitoring system which is widely used to prevent gas releasing upon rupture or large leakage event in the oil and gas pipelines. The differential pressure set point (DPS) at each valve location is the main criteria for value setting in LBV action. If the DPS is not properly adjusted, the LBV may mistakenly close or it may not take any action at proper time. In this study, the effect of characteristic parameters such as orifice diameter (OD), pipeline operational pressure (POP) and pipeline pressure drop rate (ROD) due to rupture or major leak was experimentally investigated on DPS. 100 different conditions with triple set of typical mentioned characteristic parameters were chosen. For each condition, the differential pressure (DP) was measured over 180 s by statistically analyzing the experimental results. Therefore, 100 maximum DP values (DPSs) were obtained. The attainable time to DPS depends only on OD and is independent of the POP and ROD and decreases by increasing the OD. The DPS increases by increase in ROD or decreasing OD or POP parameters. Because of using nitrogen gas instead of natural gas due to safety reasons and the uncertainties, the DPS results can be practically applied by adding a safety factor of 10%–15%. Finally, the diagram of DPS with respect to ROD was provided for different POPs and ODs. The diagram can be used as a basis for designing and regulating of LBVs installed on gas transportation pipelines.

High-level ab initio predictions for the ionisation energy, bond dissociation energies and heats of formation of zirconium oxide and its cation (ZrO/ZrO+)

ABSTRACTThe ionisation energy (IE) of ZrO, the bond dissociation energies (D0s) and the heats of formation at 0 K (ΔH○ 0Kf) and 298 K (ΔH○ 298Kf) for ZrO/ZrO+ are predicted by the coupled cluster methods utilizing upto single, double, triple and quadruple excitations and complete basis set limit approximation (CCSDTQ/CBS). The CCSDTQ/CBS approach also includes the zero-point vibrational energy, high-order correlation, core-valence (CV) electronic and spin-orbit coupling corrections while the scalar relativistic contribution is handled by employing pseudopotential basis sets. The present calculations yield IE(ZrO) = 6.794 eV and D0(Zr+–O) − D0(Zr–O) = −0.193 eV which are in good agreement with the respective experimental values of 6.81272(10) eV and −0.1788(1) eV determined in a two-colour laser-pulsed field ionisation-photoelectron study. The CCSD(T) and multireference configuration interaction (MRCI) methods with CV correlations included give equally remarkable predictions of the harmonic frequencies and the bond lengths for ZrO/ZrO+. This study together with the previous investigations has demonstrated that, with pseudopotential basis sets, the CCSDTQ/CBS protocol can be readily extended to investigate 4d-transition metal-containing diatomic molecules to yield comparable accuracy (±20 meV) to that achieved in the IE and D0 predictions of 3d-transition metal-containing species.

The matroid structure of representative triple sets and triple-closure computation

The closure cl(R) of a consistent set R of triples (rooted binary trees on three leaves) provides essential information about tree-like relations that are shown by any supertree that displays all triples in R. In this contribution, we are concerned with representative triple sets, that is, subsets R′ of R with cl(R′)=cl(R). In this case, R′ still contains all information on the tree structure implied by R, although R′ might be significantly smaller. We show that representative triple sets that are minimal w.r.t. inclusion form the basis of a matroid. This in turn implies that minimal representative triple sets also have minimum cardinality. In particular, the matroid structure can be used to show that minimum representative triple sets can be computed in polynomial time with a simple greedy approach. For a given triple set R that “identifies” a tree, we provide an exact value for the cardinality of its minimum representative triple sets. In addition, we utilize the latter results to provide a novel and efficient method to compute the closure cl(R) of a consistent triple set R that improves the time complexity O(|R||LR|4) of the currently fastest known method proposed by Bryant and Steel (1995). In particular, if a minimum representative triple set for R is given, it can be shown that the time complexity to compute cl(R) can be improved by a factor up to |R||LR|. As it turns out, collections of quartets (unrooted binary trees on four leaves) do not provide a matroid structure, in general.

Neural Wikipedian: Generating Textual Summaries from Knowledge Base Triples

Most people need textual or visual interfaces in order to make sense of Semantic Web data. In this paper, we investigate the problem of generating natural language summaries for Semantic Web data using neural networks. Our end-to-end trainable architecture encodes the information from a set of triples into a vector of fixed dimensionality and generates a textual summary by conditioning the output on the encoded vector. We explore a set of different approaches that enable our models to verbalise entities from the input set of triples in the generated text. Our systems are trained and evaluated on two corpora of loosely aligned Wikipedia snippets with triples from DBpedia and Wikidata, with promising results.

An algorithm for fast mining top-rank-k frequent patterns based on Node-list data structure

Frequent pattern mining usually requires much run time and memory usage. In some applications, only the patterns with top frequency rank are needed. Because of the limited pattern numbers, quality of the results is even more important than time and memory consumption. A Frequent Pattern algorithm for mining Top-rank-K patterns, FP_TopK, is proposed. It is based on a Node-list data structure extracted from FTPP-tree. Each node is with one or more triple sets, which contain supports, preorder and post-order transversal orders for candidate pattern generation and top-rank-k frequent pattern mining. FP_TopK uses the minimal support threshold for pruning strategy to guarantee that each pattern in the top-rank-k table is really frequent and this further improves the efficiency. Experiments are conducted to compare FP_TopK with iNTK and BTK on four datasets. The results show that FP_TopK achieves better performance.

Intelligent medical record management: a diagnosis support system

The increasing life expectations and low birth rates radically changed the demographic structure of the European Union. The number of elderly people is growing up as well as the number of chronic diseases: this implies higher and higher healthcare costs, a reduction of healthcare personnel and requests for better care services. E-health has led to a growth of the health organisation, providing innovative and not-intrusive systems together with value-added services to the healthcare actors that contribute to enhance the efficiency and reduce the costs of complex informative systems. Many research efforts lead to innovative and not-intrusive e-health systems. In this work, we present a system for supporting medical decisions. It is based on semantic analysis of available medical data. The system implements an innovative methodology, which combines different semantic approaches in order to extract the representation of a given document expressed in natural language, and to associate it to a set of RDF triples.

Intelligent medical record management: a diagnosis support system

The increasing life expectations and low birth rates radically changed the demographic structure of the European Union. The number of elderly people is growing up as well as the number of chronic diseases: this implies higher and higher healthcare costs, a reduction of healthcare personnel and requests for better care services. E-health has led to a growth of the health organisation, providing innovative and not-intrusive systems together with value-added services to the healthcare actors that contribute to enhance the efficiency and reduce the costs of complex informative systems. Many research efforts lead to innovative and not-intrusive e-health systems. In this work, we present a system for supporting medical decisions. It is based on semantic analysis of available medical data. The system implements an innovative methodology, which combines different semantic approaches in order to extract the representation of a given document expressed in natural language, and to associate it to a set of RDF triples.

Versal deformation of realisable Markov parameters

ABSTRACTLet be the set of sequences (L1,… , Ln), , admitting a minimal partial realisation of order d. To each , we associate two sequences of integers with r1 ≥ r2 ≥ … ≥ rβ > 0 = rβ+1 = … = rn and with s1 ≥ s2 ≥ … ≥ sα > 0 = sα+1 = … = sn called the partial Brunovsky column and row indices of L, respectively. Let be the subset of formed by the sequences L for which α + β ≤ n. Let Σco be the set of matrix triples with (F, G) controllable and (H, F) observable. We denote by Σco ≤ the subset of Σco formed by the triples which are minimal partial realisations of the sequences . For every ξ ∈ Σco ≤, we obtain a versal deformation of ξ corresponding to the action of the group , we show a method for obtaining a minimal partial realisation ξ of , and we derive a versal deformation of L from the obtained versal deformation of ξ.

Toward Chemical Accuracy in ab Initio Thermochemistry and Spectroscopy of Lanthanide Compounds: Assessing Core-Valence Correlation, Second-Order Spin-Orbit Coupling, and Higher Order Effects in Lanthanide Diatomics.

The higher order (HO) correlation beyond the coupled-cluster single double (triple) CCSD(T) level of theory, second-order spin-orbit coupling (SOC), and core-valence (CV) correlation effects on bond length, re, vibrational frequency, ωe, and dissociation energy, De, are studied for a set of 17 lanthanide containing diatomics, including lanthanum, europium, ytterbium, and lutetium oxides and halides. Convergence in the magnitudes of the SOC, CV, and HO corrections with respect to basis set size is examined using a sequence of double, triple, and quadruple-ζ basis sets, with the complete basis set (CBS) limit estimates provided in most cases. The CV effects on De, re, and ωe are calculated to amount up to 1.3 kcal·mol-1, 0.008 Å, and 5 cm-1, respectively. A detailed analysis of the origin of the CV effect with a particular accounting for various subvalence shells reveals that, generally, the Ln 4d correlation makes a major contribution, although in some instances the lower-lying 4sp shells contribute largely and even more substantially than 4d. The second-order SOC effect evaluated via two-component and four-component relativistic techniques proves to be non-negligible, especially for heavier species, e.g., LaI, in which it is as large as 0.8 kcal·mol-1 in De, 0.002 Å in re, and 1.3 cm-1 in ωe. Higher order correlation effects assessed through the CCSDT(Q) level are mostly less than 0.8 kcal·mol-1, 0.004 Å, and 5 cm-1; however, in species with prominence of nondynamical correlation, e.g., EuO, YbF, and LuO, the HO correction can amount to 1.2-1.6 kcal·mol-1, 0.005-0.008 Å, and 8-30 cm-1 in De, re, and ωe, respectively. In general, the [CCSD(T)+CV]/CBS + SOC + HO composite results are in good agreement with the available experimental data, exhibiting a mean absolute deviation of 1.8 kcal·mol-1 in De, 0.0023 Å in re, and 3.5 cm-1 in ωe. A significant experimental outlier, the bond length in YbI, is revealed, implying the need for re-examination of the experimental data.

Cold collisions of SH- with He: Potential energy surface and rate coefficients.

Collisional energy transfer under cold conditions is of great importance from the fundamental and applicative point of view. Here, we investigate low temperature collisions of the SH- anion with He. We have generated a three-dimensional potential energy surface (PES) for the SH-(X1Σ+)-He(1S) van der Waals complex. The ab initio multi-dimensional interaction PES was computed using the explicitly correlated coupled cluster approach with simple, double, and perturbative triple excitation in conjunction with the augmented-correlation consistent-polarized valence triple zeta Gaussian basis set. The PES presents two minima located at linear geometries. Then, the PES was averaged over the ground vibrational wave function of the SH- molecule and the resulting two-dimensional PES was incorporated into exact quantum mechanical close coupling calculations to study the collisional excitation of SH- by He. We have computed inelastic cross sections among the 11 first rotational levels of SH- for energies up to 2500 cm-1. (De-)excitation rate coefficients were deduced for temperatures ranging from 1 to 300 K by thermally averaging the cross sections. We also performed calculations using the new PES for a fixed internuclear SH- distance. Both sets of results were found to be in reasonable agreement despite differences existing at low temperatures confirming that accurate predictions require the consideration of all internal degrees of freedom in the case of molecular hydrides. The rate coefficients presented here may be useful in interpreting future experimental work on the SH- negative ion colliding with He as those recently done for the OH--He collisional system as well as for possible astrophysical applications in case SH- would be detected in the interstellar medium.

Analogy-based classifiers for nominal or numerical data

Introduced a decade ago, analogy-based classification methods constitute a noticeable addition to the set of instance-based learning techniques. They provide valuable results in terms of accuracy on many classical datasets. They rely on the notion of analogical proportions which are statements of the form “A is to B as C is to D”. Analogical proportions have been in particular formalized in Boolean and numerical settings. In both cases, one of the four components of the proportion can be computed from the three others, when the proportion holds. Analogical classifiers look for all triples of examples in the sample set that are in analogical proportion with the item to be classified on a maximal number of attributes and for which the corresponding analogical proportion equation on the class has a solution. In this paper when classifying a new item, we specially emphasize an approach where the whole set of triples that can be built from the sample set is not considered. We just focus on a small part of the candidate triples. Namely, in order to restrict the scope of the search, we first look for examples that are as similar as possible to the new item to be classified. We then only consider the pairs of examples presenting the same dissimilarity as between the new item and one of its closest neighbors. In this way, we implicitly build triples that are in analogical proportion on all attributes with the new item. Then the classification is made on the basis of an additive aggregation of the truth values corresponding to the pairs that can be analogically associated with the pairs made of the target item and one of its nearest neighbors. We then only deal with pairs leading to a solvable analogical equation for the class. This new algorithm provides results as good as previous analogical classifiers with a lower average complexity, both in nominal and numerical cases.

Multiple reaction monitoring (MRM)-profiling for biomarker discovery applied to human polycystic ovarian syndrome.

We describe multiple reaction monitoring (MRM)-profiling, which provides accelerated discovery of discriminating molecular features, and its application to human polycystic ovary syndrome (PCOS) diagnosis. The discovery phase of the MRM-profiling seeks molecular features based on some prior knowledge of the chemical functional groups likely to be present in the sample. It does this through use of a limited number of pre-chosen and chemically specific neutral loss and/or precursor ion MS/MS scans. The output of the discovery phase is a set of precursor/product transitions. In the screening phase these MRM transitions are used to interrogate multiple samples (hence the name MRM-profiling). MRM-profiling was applied to follicular fluid samples of 22 controls and 29 clinically diagnosed PCOS patients. Representative samples were delivered by flow injection to a triple quadrupole mass spectrometer set to perform a number of pre-chosen and chemically specific neutral loss and/or precursor ion MS/MS scans. The output of this discovery phase was a set of 1012 precursor/product transitions. In the screening phase each individual sample was interrogated for these MRM transitions. Principal component analysis (PCA) and receiver operating characteristic (ROC) curves were used for statistical analysis. To evaluate the method's performance, half the samples were used to build a classification model (testing set) and half were blinded (validation set). Twenty transitions were used for the classification of the blind samples, most of them (N=19) showed lower abundances in the PCOS group and corresponded to phosphatidylethanolamine (PE) and phosphatidylserine (PS) lipids. Agreement of 73% with clinical diagnosis was found when classifying the 26 blind samples. MRM-profiling is a supervised method characterized by its simplicity, speed and the absence of chromatographic separation. It can be used to rapidly isolate discriminating molecules in healthy/disease conditions by tailored screening of signals associated with hundreds of molecules in complex samples.

Populating a linked data entity name system

Resource Description Framework (RDF) is a graph-based data model used to publish data as a Web of Linked Data (Bizer et al . 2009). RDF is an emergent foundation for large-scale data integration , the problem of providing a unified view over multiple data sources. The structure in RDF data can be conveniently visualized using directed labeled graphs , as illustrated in the real-world graph fragments in Figure 1. Nodes in the graph represent entities (e.g. the node with label dbpedia:Allen_, Paul represents the entity Paul Allen in the DBpedia knowledge graph) and edges represent either attributes of an entity (e.g. '01/21/1953' is the birthdate of Paul Allen) or relationships between two entities (e.g. Paul Allen is the co-founder of the company entity, Microsoft). Facts in the knowledge base are formally represented as a set of triples , with a triple comprising a labeled edge (denoted as a property ) in the RDF graph along with its incoming and outgoing nodes.

MIRABOLIC QUANTUM sl 2 $$ {\mathfrak{sl}}_2 $$

The convolution algebra of GL d -invariant functions over the space of pairs of partial n-step flags over a finite field was studied by Beilinson-Lusztig-MacPherson. They used it to give a construction of the quantum Schur algebras and, using a stabilization procedure, of the idempotented version of the quantum enveloping algebra of $$ {\mathfrak{gl}}_n $$ . In this paper we expand the construction to the mirabolic setting of triples of two partial flags and a vector, and examine the resulting convolution algebra. In the case of n = 2, we classify the finite-dimensional irreducible representations of the mirabolic quantum algebra and we prove that the category of such representations is semisimple. Finally, we describe a mirabolic version of the quantum Schur-Weyl duality, which involves the mirabolic Hecke algebra.

Thermophysical properties of krypton-helium gas mixtures from ab initio pair potentials.

A new potential energy curve for the krypton-helium atom pair was developed using supermolecular ab initio computations for 34 interatomic distances. Values for the interaction energies at the complete basis set limit were obtained from calculations with the coupled-cluster method with single, double, and perturbative triple excitations and correlation consistent basis sets up to sextuple-zeta quality augmented with mid-bond functions. Higher-order coupled-cluster excitations up to the full quadruple level were accounted for in a scheme of successive correction terms. Core-core and core-valence correlation effects were included. Relativistic corrections were considered not only at the scalar relativistic level but also using full four-component Dirac-Coulomb and Dirac-Coulomb-Gaunt calculations. The fitted analytical pair potential function is characterized by a well depth of 31.42 K with an estimated standard uncertainty of 0.08 K. Statistical thermodynamics was applied to compute the krypton-helium cross second virial coefficients. The results show a very good agreement with the best experimental data. Kinetic theory calculations based on classical and quantum-mechanical approaches for the underlying collision dynamics were utilized to compute the transport properties of krypton-helium mixtures in the dilute-gas limit for a large temperature range. The results were analyzed with respect to the orders of approximation of kinetic theory and compared with experimental data. Especially the data for the binary diffusion coefficient confirm the predictive quality of the new potential. Furthermore, inconsistencies between two empirical pair potential functions for the krypton-helium system from the literature could be resolved.

(Energy Technology Division Graduate Student Award Address) Advanced Water Management in Fuel Cells Using Engineered Gas Diffusion Layers with Patterned Wettability: From Concept to Fuel Cell Application

The water management in PEFCs is a rather complex issue. On one hand side, some water is needed in order to hydrate a proton conductive membrane and ionomer in the catalyst layers. Therefore, humidified gases are fed into the cell. On the other hand, excessive amounts of water in the porous layers lead to blockage of available pores for gas transport and consequently increased mass transfer limitations, especially at high current densities when large amounts of water are produced. The gas diffusion layers (GDLs) are highly porous carbon fiber materials which are sandwiched between electrodes and flow fields and provide several functionalities: conduction of electricity and heat, mechanical integrity of the membrane, distribution of reactant gases and removal of the liquid water produced. These functions imply a triple set of contradictory requirements: high diffusivity of the gas phase, high permeability of the water phase, and high thermal and electrical conductivity of the solid phase. The creation of artificial pathways for water removal within the porous material would permit the development of advanced water management strategies; however none of the existing approaches proposed a method to produce such materials in a way compatible with mass productions. To tackle this issue, we invented a method to produce GDLs with patterned wettability based on radiation grafting [1]. The method permits creating hydrophilic patterns throughout the complete material thickness while the remaining areas remain hydrophobic [2]. The first part of the talk will focus on the synthetic method details [3] and particularly on the effect that electron beam energy has on the achievable resolution of the hydrophilic patterns [4]. In the second part, the capillary pressure characteristic of the modified materials will be shown and the influence of various material parameters (substrate type, pattern width, coating load, etc.) will be discussed. The water distribution was imaged at various capillary pressures using neutron radiography (NR) and X-ray tomographic microscopy [5]. Finally, electrochemical characterization of operando cells combined with NR will demonstrate the potential of these materials to improve power density at various conditions [6]. [1] P. Boillat et al., 2014 European Patent 15165515.6. [2] A. Forner-Cuenca et al., Adv. Mater. 2015, 27, 6317-6322. [3] A. Forner-Cuenca et al., J. Electrochem. Soc. 2016, 163 (8), F788-F801. [4] A. Forner-Cuenca et al., Radiat. Phys. Chem., Submitted. [5] A. Forner-Cuenca et al., J. Electrochem. Soc. 2016, 163(9), F1038-F1048. [6] A. Forner-Cuenca et al., J. Electrochem. Soc. 2016, 163 (13), F1389-F1398.

Semantic summarization of web news

In this paper, we present a general framework for retrieving relevant information from news papers that exploits a novel summarization algorithm based on a deep semantic analysis of texts. In particular, we extract from each Web document a set of triples (subject, predicate, object) that are then used to build a summary through an unsupervised clustering algorithm exploiting the notion of semantic similarity. Finally, we leverage the centroids of clusters to determine the most significant summary sentences using some heuristics. Several experiments are carried out using the standard DUC methodology and ROUGE software and show how the proposed method outperforms several summarizer systems in terms of recall and readability.

Parser Extraction of Triples in Unstructured Text

The web contains vast repositories of unstructured text. We investigate the opportunity for building a knowledge graph from these text sources. We generate a set of triples which can be used in knowledge gathering and integration. We define the architecture of a language compiler for processing subject-predicate-object triples using the OpenNLP parser. We implement a depth-first search traversal on the POS tagged syntactic tree appending predicate and object information. A parser enables higher precision and higher recall extractions of syntactic relationships across conjunction boundaries. We are able to extract 2-2.5 times the correct extractions of ReVerb. The extractions are used in a variety of semantic web applications and question answering. We verify extraction of 50,000 triples on the ClueWeb dataset.

サンプリングに基づく構造推定を用いたLOD視覚的分析支援システム

This paper proposes information visualization system for exploratory LOD (Linked Open Data) Analysis. The LOD is a framework to make data open to the public. Recently, it has been widely used to publish various kinds of data such as statistical data, geographical data, and academic data. The RDF (Resource Description Framework), which describes data as a set of triples consisting of subject, predicate and object, is commonly used to publish data as LOD. When we want to use LOD, it is necessary to understand its structure, such as graph structure of RDF data, used vocabularies and resources. Therefore, we often have to conduct exploratory analysis of LOD. In order to support the analysis, the proposed system analyzes the structure of LOD written with RDF, and visualizes the result of analysis. As most of currently available LOD have table structure, the proposed system identifies whether target dataset contains table structure or not using resource sampling with SPARQL queries. Graph structure of resources obtained by the resource sampling is visualized by assigning different colors to different tables. A user cannot only examine the visualized structure, but also conduct exploratory search by selecting a resource in the visualized result. Effectiveness of the proposed system is evaluated by applying it to several LOD resources. Experiments with test participants are also conducted, of which results show even users who are not familiar with RDF can perform exploratory analysis effectively.