Towards domain-independent information extraction from web tables

The growing amount of semi-structured and unstructured data on tourism Web sites with heterogeneous designs requires information extraction (IE) mechanisms, to create, for instance, tourism portals. In order to build semantic eTourism environments, the acquisition of room prices is of particular interest. Room prices and related information often appear in tabular structures, which still challenge Web information extraction techniques. In this paper, we begin by identifying various price table patterns which are characterized by the position of a number of features that determine a room price. We then describe an extended ontology model for tourism prices. Finally, we present TAINEX, a plug-in for functional and structural analysis and data interpretation of price tables, which extends the existing prototype TourIE, a rule/ontology-based information extraction system for Web sites with heterogeneous designs.

Information extraction from semi-structured webpages provides valuable long-tailed facts for augmenting knowledge graph. Relational Web tables are a critical component containing additional entities and attributes of rich and diverse knowledge. However, extracting knowledge from relational tables is challenging because of sparse contextual information. Existing work linearize table cells and heavily rely on modifying deep language models such as BERT which only captures related cells information in the same table. In this work, we propose a novel relational table representation learning approach considering both the intra- and inter-table contextual information. On one hand, the proposed Table Convolutional Network model employs the attention mechanism to adaptively focus on the most informative intra-table cells of the same row or column; and, on the other hand, it aggregates inter-table contextual information from various types of implicit connections between cells across different tables. Specifically, we propose three novel aggregation modules for (i) cells of the same value, (ii) cells of the same schema position, and (iii) cells linked to the same page topic. We further devise a supervised multi-task training objective for jointly predicting column type and pairwise column relation, as well as a table cell recovery objective for pre-training. Experiments on real Web table datasets demonstrate our method can outperform competitive baselines by +4.8% of F1 for column type prediction and by +4.1% of F1 for pairwise column relation prediction.

As the world’s largest information sharing platform, web pages contain a large amount of table data. However, these table data emphasize the presentation of data to users in a more intuitive way, which cannot be understood by computers. How to obtain large-scale web table data and carry out in-depth applications has become the focus of data science. This paper designs an extraction framework for Chinese Web table data. This framework extracts Web table data containing Chinese from the Web and constructs an open Chinese Web table data set CWTs (Chines Web Tables). The data set is quantitatively analyzed in multiple dimensions such as scale and data distribution. The results show that the data set obtained in this paper can provide a preliminary data basis for table data query, question answering system, and knowledge base construction, at the same time provide an optional channel for building a large-scale machine-understandable Chinese knowledge base.

A method is presented that permits the extraction and visualization of topological domain structure information contained in small-angle scattering (SAS) patterns without complex pretreatment. Multi-dimensional noisy raw data can be processed. Such data are, for instance, accumulated in the field of materials research from short-exposure-timein situsmall-angle X-ray scattering (SAXS) experiments with synchrotron radiation. The result is a multi-dimensional intersect or chord distribution, which is defined as the Laplacian of the correlation function. Moreover, it is equivalent to the autocorrelation of the gradient of the electron density. The procedure is, in particular, adapted to the analysis of the nanoscale structure of samples with fibre symmetry, such as polymer fibres or strained elastomers. Multi-dimensional relations among morphological components become apparent in real space and help to elucidate the nature of the processes governing formation and change of structure on the nanometre scale. Utilizing digital signal processing tools, the algorithm is based on spatial frequency filtering of the raw data. The background to be subtracted from the small-angle scattering pattern is formed from its own low spatial frequencies. Noise may be removed by suppressing high spatial frequencies. In the frequency band between these low and high spatial frequencies, the domain structure information of the studied nanocomposite appears.

The construction of a network model is one of the key techniques in organic combination of microscopic flow experiment and simulation. The construction method of a three-dimensional network model is presented on the basis of CT scanning images in this paper. A series of CT slice images describing microscopic pore structure and fluid distribution of actual rock is obtained with the help of the industrial microfocus CT system. Based on the extraction of pore space skeleton, pore and throat information, the corresponding network model is established, and the conversion from three-dimensional CT image information to pore-throat size distribution and topological information is also achieved. The feature of this method lies in the fact that complicated pore space of rock may be characterized by pores and throats with a simple shape while keeping the geometry and flow characteristics. It is indicated that the calculated results of porosity, permeability, relative permeability curve and microscopic remaining oil distribution match very well the experimental results of water flooding and polymer flooding. This network model may fairly well characterize the rock microscopic pore-throat size and topological characteristics.

As an important function of GIS, spatial query covers not only the extraction of geometric, topologic or semantic information but also the retrieval of spatial cognition related information. This study presents a shape based spatial query way that is formally described as: Select {O i From DataBase Where O i.shape LIKE Template At_Degree <C i >. To extend the new operation LIKE in formal SQL language, the shape representation and measurement have to be built. This study aims at polygon object offering a Fourier transform based method to compute the degree of shape similarity. Through cognition experiment builds the membership function of fuzzy term LIKE. The query experiments show the shape based query retrieves the building result consistent with human cognition.KeywordsSpatial querySpatial cognitionShape measurementFourier transform

Detailed investigations of the unsaturated zone in regard to water and tritium balance and tritium input into groundwater are necessary especially in the case of thick and inhomogeneous aeration zone (different soil layers, clay horizons etc.) and inhomogeneous land utilization. The paper shows a method to quantify water and tritium movement and storage in unsaturated zone by means of a two-dimensional conceptual box model called BOWAM(T). The model was tested on 4 sandstone aquifers in the Finne region (brink of the Thuringia basin). The results show that water balance depends strongly on surface morphology and infiltration capacity, non- or semiconducting layers within aeration zone, land usage (forest, agriculture, meadow…) and atmospheric conditions. Especially runoff, evapotranspiration and groundwater recharge at Finne region have a sensitive variation. Naturally, the tritium balance is connected with water balance. There is a difference in tritium input functions into groundwater between covered and uncovered aquifers. In the case of a covered aquifer with a deep groundwater table, tritium “bomb peak” is within the aeration zone and has not left the aeration zone into groundwater up to now. In the case of uncovered aquifers with a groundwater table near the surface it was found a time different tritium input function by means of the BOWAM(T) model, depends on groundwater distance from soil surface and land usage. A comparison to measured tritium activities in groundwater (wells and springs) shows a respectable correspondence. Further investigation and groundwater modelling is planned.

Understanding multiphase seepage behavior in porous media of oil and gas reservoirs is crucial for optimizing development strategies, enhancing recovery efficiency, and improving project economics. Physical simulation models, scaled-down replicas of reservoir conditions, deliver direct insights into petrophysical parameters and dynamic flow processes. This review systematically classified seepage physical models into one-dimensional (1D), two-dimensional (2D), and three-dimensional (3D) models, further delineated by structure, function, and application structure, function, and application. 1D models include sand-packed pipes, artificial cores, and natural cores models. 2D models encompass parallel plate, etched network, core slice, and particle packing models. 3D models comprise sand-packed, large-scale core, and visual models. A comprehensive discussion and systematic evaluation of these models were conducted, focusing on their advantages, limitations, error sources, and applicability. The results demonstrate that 1D models offer operational simplicity, high temperature and pressure resistance, and well-established fabrication techniques. However, they lack spatial representativeness, scaling similarity, and the capacity to capture heterogeneity. 2D models exhibit superior structural flexibility, functional diversity, and direct visualization capabilities, but cannot simulate complex 3D flow or replicate high temperature and pressure resistance conditions. 3D models minimize scaling effects, reliably simulate macroscopic flow dynamics, and withstand extreme conditions, but suffer from complex operation, prolonged experimental durations, and high cost. Given the flexibility and versatility, this paper further comprehensively reviews various types of 2D models, compares their visualization performance, representativeness, temperature and pressure resistance, applications, and constraints, and proposes potential refinements. Typical case studies illustrate practical implementations to promote the translation of theoretical research into field practice and to strengthen the connection between theory and application. Furthermore, we establish how numerical simulation and artificial intelligence transcend the limitations of physical models. This review provides a comprehensive, practical, and forward-looking reference for the selection, design, and application of physical seepage models in oil and gas reservoir studies.

For rapid modelling of ship structure and sharing of product data for a computer-integrated manufacturing system in shipbuilding, this research proposes a data exchange method between two-dimensional drawings and three-dimensional finite element models. The report of the research is followed by appropriately planning the structure of the extracted data based on the Standard for the Exchange of Product Model Data. A set of rules for two-dimensional feature recognition is established, and then the method for the extraction of graphic and non-graphic design information from two-dimensional drawings of transverse sections is presented. An algorithm for rapid modelling of the longitudinal hull structure is proposed, based on data extracted from the transverse sections. This algorithm bypasses the geometric modelling stage required in traditional manual modelling, exclusively making use of the extracted information. The proposal also enriches data exchange, greatly reduces human error in modelling and consequently improves the accuracy of the finite element modelling for the shipbuilding industry.

Robots which interact with the physical world will benefit from a fine-grained tactile understanding of objects and surfaces. Additionally, for certain tasks, robots may need to know the haptic properties of an object before touching it. To enable better tactile understanding for robots, we propose a method of classifying surfaces with haptic adjectives (e.g., compressible or smooth) from both visual and physical interaction data. Humans typically combine visual predictions and feedback from physical interactions to accurately predict haptic properties and interact with the world. Inspired by this cognitive pattern, we propose and explore a purely visual haptic prediction model. Purely visual models enable a robot to “feel” without physical interaction. Furthermore, we demonstrate that using both visual and physical interaction signals together yields more accurate haptic classification. Our models take advantage of recent advances in deep neural networks by employing a unified approach to learning features for physical interaction and visual observations. Even though we employ little domain specific knowledge, our model still achieves better results than methods based on hand-designed features.

With the exponentially growing amount of information available on the Internet, an effective technique for users to discern the useful information from the unnecessary information is urgently required. Cleaning web pages for web data extraction becomes critical for improving performance of information retrieval and information extraction. So, we investigate to remove various noise patterns in Web pages instead of extracting relevant content from Web pages to get main content information. In this paper, we propose an approach that detect multiple noise patterns and remove these noise patterns from Web pages of any Web sites. The method first build DOM tree for any Web page. Our approach is based on the basic idea of Case-Based Reasoning (CBR) to find noise pattern in current Web page by matching similar noise pattern kept in Case-Based. We also apply a back propagation neural network algorithm to classify the stored various noise patterns by matching similar noise data in current Web page. We have implemented our method on several commercial Web sites and News Web sites to evaluate the performance and improvement of our approach. Experiments show that results leads to a more accurate and effectiveness of the approach.

This paper is mainly about the design of web information retrieval module based on ontology, using ontology technology, from the current web service mechanism based on the syntax level raised to the level of knowledge (or concept).First, the paper introduces the concept of the ontology, and then completes the architecture of the web information retrieval module based on ontology, with the technology of information extraction and semantic extension. This technology applies to web information retrieval, providing more accurate, intelligent web information retrieval service for the users, and improve the recall and precision ratio. Index Terms - ontology; ontology reasoning; information extraction; semantic expansion.

웹의 발전으로 말미암아 새롭게 부상하고 있는 차세대 웹 언어인 HTML5의 웹 기술 언어 표준화가 현재 진행 중이다. 웹의 관점에서 볼 때 HTML5에 새롭게 추가된 구조적 시맨틱 요소들로 인하여 예전보다 웹의 기술에서 웹에서의 데이터들을 더욱 의미적으로 표현할 수 있게 해주는 것이 가능하게 되었다. 이는 또한 새로운 HTML5로 기술된 웹 문서에서 유용한 정보를 쉽게 추출할 수 있음을 나타내기도 한다. 따라서 웹 문서에 새롭게 추가된 시맨틱 요소들에 대한 온톨로지화를 통하여 웹 문서 정보에 대한 직관적이고 구조적인 접근방식이 필요하다. 본 연구에서는 HTML5에 새로 추가된 시맨틱 요소들에 입각하여 웹 문서 요소들의 온톨로지 모델과 인스턴스들의 매핑 방법론을 통하여 효과적인 HTML5 문서의 온톨로지화 방법론을 제안하고자 한다. HTML5, new standard for web language, is being standardized corresponding to the development of web. Since several new semantic elements have been added into HTML5 standard, current Web Environment is becoming more and more semantic. In order to provide better user experience by using information extraction from HTML5 page, new HTML5 Elements should be mapped to a corresponding Ontology. In this research, we will focus on new semantic elements to build Ontology from HTML5 document. For this purpose, we will propose a methodology of Schema level mapping rule and instance mapping rule.

During the last decade the amount of scientific articles available online has substantially grown in parallel with the adoption of the Open Access publishing model. Nowadays researchers, as well as any other interested actor, are often overwhelmed by the enormous and continuously growing amount of publications to consider in order to perform any complete and careful assessment of scientific literature. As a consequence, new methodologies and automated tools to ease the extraction, semantic representation and browsing of information from papers are necessary. We propose a platform to automatically extract, enrich and characterize several structural and semantic aspects of scientific publications, representing them as RDF datasets. We analyze papers by relying on the scientific Text Mining Framework developed in the context of the European Project Dr. Inventor. We evaluate how the Framework supports two core scientific text analysis tasks: rhetorical sentence classification and extractive text summarization. To ease the exploration of the distinct facets of scientific knowledge extracted by our platform, we present a set of tailored Web visualizations. We provide on-line access to both the RDF datasets and the Web visualizations generated by mining the papers of the 2015 ACL-IJCNLP Conference.

BackgroundMost of the existing methods to analyze high-throughput data are based on gene ontology principles, providing information on the main functions and biological processes. However, these methods do not indicate the regulations behind the biological pathways. A critical point in this context is the extraction of information from many possible relationships between the regulated genes, and its combination with biochemical regulations. This study aimed at developing an automatic method to propose a reasonable number of upstream regulatory candidates from lists of various regulated molecules by confronting experimental data with encyclopedic information.ResultsA new formalism of regulated reactions combining biochemical transformations and regulatory effects was proposed to unify the different mechanisms contained in knowledge libraries. Based on a related causality graph, an algorithm was developed to propose a reasonable set of upstream regulators from lists of target molecules. Scores were added to candidates according to their ability to explain the greatest number of targets or only few specific ones. By testing 250 lists of target genes as inputs, each with a known solution, the success of the method to provide the expected transcription factor among 50 or 100 proposed regulatory candidates, was evaluated to 62.6% and 72.5% of the situations, respectively. An additional prioritization among candidates might be further realized by adding functional ontology information. The benefit of this strategy was proved by identifying PPAR isotypes and their partners as the upstream regulators of a list of experimentally-identified targets of PPARA, a pivotal transcriptional factor in lipid oxidation. The proposed candidates participated in various biological functions that further enriched the original information. The efficiency of the method in merging reactions and regulations was also illustrated by identifying gene candidates participating in glucose homeostasis from an input list of metabolites involved in cell glycolysis.ConclusionThis method proposes a reasonable number of regulatory candidates for lists of input molecules that may include transcripts of genes and metabolites. The proposed upstream regulators are the transcription factors themselves and protein complexes, so that a multi-level description of how cell metabolism is regulated is obtained.