Data Representation Model Research Articles

Medical knowledge infused convolutional neural networks for cohort selection in clinical trials.

In this era of digitized health records, there has been a marked interest in using de-identified patient records for conducting various health related surveys. To assist in this research effort, we developed a novel clinical data representation model entitled medical knowledge-infused convolutional neural network (MKCNN), which is used for learning the clinical trial criteria eligibility status of patients to participate in cohort studies. In this study, we propose a clinical text representation infused with medical knowledge (MK). First, we isolate the noise from the relevant data using a medically relevant description extractor; then we utilize log-likelihood ratio based weights from selected sentences to highlight "met" and "not-met" knowledge-infused representations in bichannel setting for each instance. The combined medical knowledge-infused representation (MK) from these modules helps identify significant clinical criteria semantics, which in turn renders effective learning when used with a convolutional neural network architecture. MKCNN outperforms other Medical Knowledge (MK) relevant learning architectures by approximately 3%; notably SVM and XGBoost implementations developed in this study. MKCNN scored 86.1% on F1metric, a gain of 6% above the average performance assessed from the submissions for n2c2 task. Although pattern/rule-based methods show a higher average performance for the n2c2 clinical data set, MKCNN significantly improves performance of machine learning implementations for clinical datasets. MKCNN scored 86.1% on the F1 score metric. In contrast to many of the rule-based systems introduced during the n2c2 challenge workshop, our system presents a model that heavily draws on machine-based learning. In addition, the MK representations add more value to clinical comprehension and interpretation of natural texts.

EOS-LNG: A Fundamental Equation of State for the Calculation of Thermodynamic Properties of Liquefied Natural Gases

A new mixture model (EOS-LNG) for the accurate representation of thermodynamic property data of multicomponent natural gas mixtures in the liquid state is presented. The mathematical approach of the GERG-2008 equation of state of Kunz and Wagner is adopted and new binary-specific functions for methane + n-butane, methane + isobutane, methane + n-pentane, and methane + isopentane are developed. The representation of all experimental data available in the literature for the corresponding binary systems is carefully analyzed so that these functions can also be applied at fluid states beyond the liquefied natural gas (LNG) region. The EOS-LNG represents all available binary and multicomponent data in the LNG region within their specified experimental uncertainty, which is significantly more accurate than the GERG-2008 model. The main focus was given to the representation of new density data measured between 100 K and 180 K with a maximum pressure of 10 MPa. Deviations from the EOS-LNG presented here do not exceed 0.02% for binary data and 0.05% for multicomponent systems. Deviations of calculated values of these data from experimental data in other fluid regions are similar to or better than those calculated with the GERG-2008 model.

Multiresolution 3D-DenseNet for Chemical Shift Prediction in NMR Crystallography.

We have developed a deep learning algorithm for chemical shift prediction for atoms in molecular crystals that utilizes an atom-centered Gaussian density model for the 3D data representation of a molecule. We define multiple channels that describe different spatial resolutions for each atom type that utilizes cropping, pooling, and concatenation to create a multiresolution 3D-DenseNet architecture (MR-3D-DenseNet). Because the training and testing time scale linearly with the number of samples, the MR-3D-DenseNet can exploit data augmentation that takes into account the property of rotational invariance of the chemical shifts, thereby also increasing the size of the training data set by an order of magnitude without additional cost. We obtain very good agreement for 13C, 15N, and 17O chemical shifts when compared to ab initio quantum chemistry methods, with the highest accuracy found for 1H chemical shifts that is comparable to the error between the ab initio results and experimental measurements. Principal component analysis (PCA) is used to both understand these greatly improved predictions for 1H , as well as indicating that chemical shift prediction for 13C, 15N, and 17O, which have far fewer training environments than the 1H atom type, will improve once more unique training samples are made available to exploit the deep network architecture.

Quasi-holography computational model for urban computing

Vast amounts of data are produced with the development of smart cities and urban computing technologies. The data is often captured from multiple sensors, with heterogeneous structures and highly decentralized connections. Integrated data representation and smart computational models are required for more complex tasks in urban computing. We dwell deeply on two fundamental questions — can we provide an integrated data representation for the whole cyber–physical–social system? And, can we provide an integrated framework to choose the appropriate data for understanding a specific urban event? A holography data representation and the quasi-holography computational model have been proposed to address these problems. We describe case studies using the quasi-holography computational model, and discuss further problems to solve regarding our model.

DUSKG: A fine-grained knowledge graph for effective personalized service recommendation

Abstract The last decade has witnessed the rapid development of service-oriented computing, resulting in a tremendous growth of services which further calls for novel approaches for efficient and effective service recommendations. Various types of data can be used for service recommendations, such as user feedback on services, service profile, user profile, etc. In traditional recommendation algorithms, such data is usually handled separately and in the form of matrix. Since there are underlying logical relations between the data, we conjecture that deep convergence of all the data and full considerations of such logical relations would help boost accuracy of service recommendations. By a comprehensive survey on various existing data convergence methods, we found that knowledge graph had been a typical approach to describe the different relations between data. In this paper, we propose an elegant, natural and compact data representation model incorporating such kinds of logical relations, namely D omain-oriented U ser and S ervice Interaction K nowledge G raph (DUSKG). Three types of entities ( U s e r , S e r v i c e and S e r v i c e V a l u e F e a t u r e (VF)), five types of fine-grained relations (FOCUSON, BELONGTO, USIMILAR, SSIMILAR and FSIMILAR), and weight vectors that quantitatively delineate these relations are elaborately defined. An approach for constructing DUSKG from service data is presented. Extracted from DUSKG, each user’s value preference is then represented in terms of VFs. Five personalized service recommendation methods are presented in terms of such value preferences and various relations in DUSKG. To verify the proposed methods, a set of experiments are conducted on Yelp dataset. Experimental results show that our approaches achieve better recommendation performance compared to the state-of-the-art methods, and among the five ones, HR-DUSKG (Hybrid Recommendation Depending on DUSKG) exhibits the best performance with quite good achievements but least calculation time.

A Concise Temporal Data Representation Model for Prediction in Biomedical Wearable Devices

Predictive analytics and event forecasting using deep learning techniques require processing long-term historical data which is infeasible in low-power wearable devices. Such devices are constrained in memory and computational power, and are pushed to their limits by resource hungry deep neural networks. Current techniques either ignore historical data, or convert temporal sequences to pattern sequences, eliminating valuable properties such as time and/or recency. The proposed model maps arbitrary-length multivariate discrete time series to a concise sequence, called mapped interval sequence (MIS). MIS retains original data properties such as time, recency, and scale, without being susceptible to missing values. Life model for time series (LMts) mapping, is capable of mapping billions of data elements with sampling rate of several kHz or higher into a sequence of 32 elements or fewer. Furthermore, a new loss function called as tolerance error is introduced to improve long-term forecasting events using LMts. In a smart health Internet of Things environment, LMts enables real-time health predictions depending on both recent and historical data. In addition, the LMts model can predict the approximate time of events, with granularity of seconds and up to years. Experimental results show that, compared to previous studies in fall prediction, LMts achieves the same 100% accuracy with a single long short-term memory layer, while covering $16{\boldsymbol \times }$ longer time period and using $80{\boldsymbol \times }$ less weight parameters. LMts is also used to forecast human fall up to 14 s in advance even with 50% missing values.

Performance analysis of different pixel-wise processing methods for depth imaging with single photon detection data

ABSTRACTWe establish a long-range single photon counting three-dimensional (3D) imaging system based on cage optical structure. Five different pixel-wise processing methods for time-of-flight (TOF) photon counting data are compared with data collected by our 3D imaging system for ranges 40–700 m and a suitable representation model for photon counting data is proposed for pixel-wise processing. Experimental results show that these methods exploit the instrumental response function (IRF), yielding a high-quality 3D image. When the signal photon counts are greater than 13 per pixel, the resulting mean absolute error (MAE) values of the IRF-based methods are better than results from the non-IRF-based methods. Regarding IRF-based methods, the union of subspace (UOS) model-based approach and cross correlation are more suitable than the Markov chain Monte Carlo (MCMC) method in the condition of a small number of return signal photons. These results offer valuable information to promote the implementation of photon counting 3D imaging in real applications.

Shared Multi-View Data Representation for Multi-Domain Event Detection.

Internet platforms provide new ways for people to share experiences, generating massive amounts of data related to various real-world concepts. In this paper, we present an event detection framework to discover real-world events from multiple data domains, including online news media and social media. As multi-domain data possess multiple data views that are heterogeneous, initial dictionaries consisting of labeled data samples are exploited to align the multi-view data. Furthermore, a shared multi-view data representation (SMDR) model is devised, which learns underlying and intrinsic structures shared among the data views by considering the structures underlying the data, data variations, and informativeness of dictionaries. SMDR incorpvarious constraints in the objective function, including shared representation, low-rank, local invariance, reconstruction error, and dictionary independence constraints. Given the data representations achieved by SMDR, class-wise residual models are designed to discover the events underlying the data based on the reconstruction residuals. Extensive experiments conducted on two real-world event detection datasets, i.e., Multi-domain and Multi-modality Event Detection dataset, and MediaEval Social Event Detection 2014 dataset, indicating the effectiveness of the proposed approaches.

Temporal Data Representation and Querying Based on RDF

With the explosive growth of temporal data, how to query and manage temporal data has become an important research issue. Resource description framework (RDF), as the standard data and knowledge description language of the semantic web, has been widely used to represent various domain data. Aiming at the representation and querying of temporal data, this paper proposes a temporal data representation model based on RDF and its corresponding querying method. First, a representation model called RDFt is proposed, which can represent temporal data with both the time information and the update count information, and the syntax and semantics of the RDFt model are given. Then, we propose a query language called SPARQL[t] for RDFt, and we give the query syntax and operations of SPARQL[t] in detail. In addition, a querying transformation algorithm from SPARQL[t] to SPARQL is proposed, in order to achieve compatibility with the existing RDF query engines. Finally, we implemented a prototype system that can support RDFt temporal data representation and querying, and the case studies and experimental results verify the feasibility of the proposed approach.

Visualisation and personalisation of multi-representations city models

ABSTRACTCity models have a wide variety of uses that require different kind of data representation or data models. Having a dynamic model that enables picking the right representations (meshes, volumetric data, point cloud, etc.) can prove useful to adapt an application to each user's needs. In this paper, we present an original method to create personalised visualisations of 3D city models on the fly. By organising the server data in a hierarchy of tiles, we are able to generate personalised models based on the user's preferences. These preferences take the shape of a set of rules that apply to each tile or city object and allow the user to choose which representation of the object to use depending on its position or semantic information (classification, height, etc.). Our method is designed around existing standards, guaranteeing the interoperability of the produced models.

Preliminary Monitoring of Physico-Chemical Parameters of Water Wells from the Village of Bivolari (The Moldavian Plain)

Abstract The main objective of the paper is focused on water quality research in the village of Bivolari, Iasi County. The Bivolari settlement is situated in the Prut meadow and develops along the road connecting Iasi and Stefanesti. The research done in this paper starts from the collection of data on water pH, water LDO (oxygen concentration), CDC (dissolved salt content) and water temperature. The parameters were measured for 7 months using the HACH-LANGE multiparameter of the Geoarchaeology Laboratory at Alexandru Ioan Cuza University of Iasi (Inter-disciplinary Platform ArheoInvest). The study is structured in two sections: scientific substantiation and applicative research. The scientific substantiation part analyzes the water features in the context of the integrated management approach as well as an analysis of the quality aspects of the water supply. The applied research part required monthly field measurements in 50 fountains (without the winter season). The approached subject is is complex and up-to-date because the citizen is a priority for EU Member States (EUPAN, 2009): the aspect of “customer satisfaction” represents a capital priority. Finally, the database is structured into two categories: spatial information such as maps; tabular information attached to spatial data. The paper also follows the transition from the field of research into the field of application by presenting models of data representation, which are at the same time tools adapted to the practical needs. In order to achieve this goal, the concept of quality (both in terms of product - drinking water and local consumer) is defined and addressed in an integrated manner. Of the 50 wells monitored, only one had a water shortage in the last 10 years, and in two other cases there were problems with household waste and dead animals that damaged the water supply. Some fountains can dry at intervals of 3-4 years.

RDF Data Storage and Query Processing Schemes

The Resource Description Framework (RDF) represents a main ingredient and data representation format for Linked Data and the Semantic Web. It supports a generic graph-based data model and data representation format for describing things, including their relationships with other things. As the size of RDF datasets is growing fast, RDF data management systems must be able to cope with growing amounts of data. Even though physically handling RDF data using a relational table is possible, querying a giant triple table becomes very expensive because of the multiple nested joins required for answering graph queries. In addition, the heterogeneity of RDF Data poses entirely new challenges to database systems. This article provides a comprehensive study of the state of the art in handling and querying RDF data. In particular, we focus on data storage techniques, indexing strategies, and query execution mechanisms. Moreover, we provide a classification of existing systems and approaches. We also provide an overview of the various benchmarking efforts in this context and discuss some of the open problems in this domain.

Striped Data Server for Scalable Parallel Data Analysis

A columnar data representation is known to be an efficient way for data storage, specifically in cases when the analysis is often done based only on a small fragment of the available data structures. A data representation like Apache Parquet is a step forward from a columnar representation, which splits data horizontally to allow for easy parallelization of data analysis. Based on the general idea of columnar data storage, working on the FNAL LDRD Project FNAL-LDRD-2016-032, we have developed a striped data representation, which, we believe, is better suited to the needs of High Energy Physics data analysis. A traditional columnar approach allows for efficient data analysis of complex structures. While keeping all the benefits of columnar data representations, the striped mechanism goes further by enabling easy parallelization of computations without requiring special hardware. We will present an implementation and some performance characteristics of such a data representation mechanism using a distributed no-SQL database or a local file system, unified under the same API and data representation model. The representation is efficient and at the same time simple so that it allows for a common data model and APIs for wide range of underlying storage mechanisms such as distributed no-SQL databases and local file systems. Striped storage adopts Numpy arrays as its basic data representation format, which makes it easy and efficient to use in Python applications. The Striped Data Server is a web service, which allows to hide the server implementation details from the end user, easily exposes data to WAN users, and allows to utilize well known and developed data caching solutions to further increase data access efficiency. We are considering the Striped Data Server as the core of an enterprise scale data analysis platform for High Energy Physics and similar areas of data processing. We have been testing this architecture with a 2TB dataset from a CMS dark matter search and plan to expand it to multiple 100 TB or even PB scale. We will present the striped format, Striped Data Server architecture and performance test results.

Digital Data Visualization with Interactive and Virtual Reality Tools. Review of Current State of the Art and Proposal of a Model.

Massive and open data constitute a burgeoning field of study in the current context. The evolution of technology is, in turn, increasing its degree of interactivity, configuring several scenarios of great complexity in which data is understood on the basis of our interaction with it at different levels. Technologies such as virtual reality or augmented reality present an emerging framework for visualizing, representing and understanding information. Moreover, new disciplines such as interaction design, human-computer interaction, and user experience are needed to optimally configure the representation and design of data interaction dynamics, so that they can be implemented in contexts such as education. This paper reviews the current state of interactive and immersive technology (including virtual reality and alternative reality games) and of open and massive data, to highlight potential projections and propose models of data representation based on factors such as storytelling or user experience. This paper shows the need to develop models for data use and representation in fields such as education and citizen empowerment.

ETL Semantic Model for Big Data Aggregation, Integration, and Representation

ETL Semantic Model for Big Data Aggregation, Integration, and Representation

Automated extraction of Biomarker information from pathology reports

BackgroundPathology reports are written in free-text form, which precludes efficient data gathering. We aimed to overcome this limitation and design an automated system for extracting biomarker profiles from accumulated pathology reports.MethodsWe designed a new data model for representing biomarker knowledge. The automated system parses immunohistochemistry reports based on a “slide paragraph” unit defined as a set of immunohistochemistry findings obtained for the same tissue slide. Pathology reports are parsed using context-free grammar for immunohistochemistry, and using a tree-like structure for surgical pathology. The performance of the approach was validated on manually annotated pathology reports of 100 randomly selected patients managed at Seoul National University Hospital.ResultsHigh F-scores were obtained for parsing biomarker name and corresponding test results (0.999 and 0.998, respectively) from the immunohistochemistry reports, compared to relatively poor performance for parsing surgical pathology findings. However, applying the proposed approach to our single-center dataset revealed information on 221 unique biomarkers, which represents a richer result than biomarker profiles obtained based on the published literature. Owing to the data representation model, the proposed approach can associate biomarker profiles extracted from an immunohistochemistry report with corresponding pathology findings listed in one or more surgical pathology reports. Term variations are resolved by normalization to corresponding preferred terms determined by expanded dictionary look-up and text similarity-based search.ConclusionsOur proposed approach for biomarker data extraction addresses key limitations regarding data representation and can handle reports prepared in the clinical setting, which often contain incomplete sentences, typographical errors, and inconsistent formatting.

Symbolic analysis-based reduced order Markov modeling of time series data

This paper presents reduced-order modeling of time-series data for a special class of Markov models using symbolic dynamics. These models are constructed from the time-series signal by partitioning the data and then inferring a probabilistic finite state automaton (PFSA) from the resulting symbol sequence, capturing a finite history (or memory) of symbol strings. In the proposed approach, the size of the temporal memory of a symbol sequence is estimated from spectral properties of the resulting stochastic matrix corresponding to a first-order Markov model of the symbol sequence. Then, agglomerative hierarchical clustering is used to cluster states of the corresponding full-order Markov model to construct a reduced-order Markov model based on information-theoretic criteria with a non-deterministic algebraic structure; the parameters of the reduced-order model are identified from the original model by making use of a Bayesian inference rule. The model size is inferred using an information-theoretic inspired criteria; the Markov parameters of the reduced-order model are identified from the original model by making use of a Bayesian inference rule. The paper also identifies theoretical bounds on the error induced in the reduced-size model in terms of expected Hamming distance between the sequences generated by the original and final reduced-size models. The proposed concept is elucidated and validated by two examples on different data sets. The first example analyzes a set of time series of pressure oscillations in a swirl-stabilized combustor, where controlled protocols are used to induce flame instabilities. Variations in the complexity of the derived Markov model represent how the system operating condition changes from stable to an unstable combustion regime. The second example is built upon a public data set of NASA’s repository for prognosis of rolling-element bearings. It is shown that: (i) even with a small state-space, the reduced-order models are able to achieve comparable performance, and (ii) the proposed approach provides flexibility in the selection of a reduced-order model for data representation and learning.

A Learning-Based Framework for Improving Querying on Web Interfaces of Curated Knowledge Bases

Knowledge Bases (KBs) are widely used as one of the fundamental components in Semantic Web applications as they provide facts and relationships that can be automatically understood by machines. Curated knowledge bases usually use Resource Description Framework (RDF) as the data representation model. To query the RDF-presented knowledge in curated KBs, Web interfaces are built via SPARQL Endpoints. Currently, querying SPARQL Endpoints has problems like network instability and latency, which affect the query efficiency. To address these issues, we propose a client-side caching framework, SPARQL Endpoint Caching Framework (SECF), aiming at accelerating the overall querying speed over SPARQL Endpoints. SECF identifies the potential issued queries by leveraging the querying patterns learned from clients’ historical queries and prefecthes/caches these queries. In particular, we develop a distance function based on graph edit distance to measure the similarity of SPARQL queries. We propose a feature modelling method to transform SPARQL queries to vector representation that are fed into machine-learning algorithms. A time-aware smoothing-based method, Modified Simple Exponential Smoothing (MSES), is developed for cache replacement. Extensive experiments performed on real-world queries showcase the effectiveness of our approach, which outperforms the state-of-the-art work in terms of the overall querying speed.

An Effective and Scalable Framework for Authorship Attribution Query Processing

Authorship attribution aims at identifying the original author of an anonymous text from a given set of candidate authors and has a wide range of applications. The main challenge in authorship attribution problem is that the real-world applications tend to have hundreds of authors , while each author may have a small number of text samples, e.g., 5–10 texts/author. As a result, building a predictive model that can accurately identify the author of an anonymous text is a challenging task. In fact, existing authorship attribution solutions based on long text focus on application scenarios, where the number of candidate authors is limited to 50. These solutions generally report a significant performance reduction as the number of authors increases. To overcome this challenge, we propose a novel data representation model that captures stylistic variations within each document, which transforms the problem of authorship attribution into a similarity search problem. Based on this data representation model, we also propose a similarity query processing technique that can effectively handle outliers. We assess the accuracy of our proposed method against the state-of-the-art authorship attribution methods using real-world data sets extracted from Project Gutenberg. Our data set contains 3000 novels from 500 authors. Experimental results from this paper show that our method significantly outperforms all competitors. Specifically, as for the closed-set and open-set authorship attribution problems, our method have achieved higher than 95% accuracy.

Constructing Knowledge Base of Emergency Based on Hierarchical Rough Set Theory

How to dig out valuable knowledge from a large number of unexpected events, and assist the emergency managers to make effective decisions, is a difficult problem. Decision rules mining is an important technique in emergency decision-making management. However, most existing algorithms are based on flat data tables. In this paper, based on hierarchical rough set theory, an emergency rule knowledge base was constructed, which sets of decision rules mining from different abstract levels. First of all, the analysis of emergency incidents attributes, attribute value types are divided into character, language, numerical value, according to the different types construct the emergency attribute hierarchical concept tree. Then, the event data is transformed into the form of multidimensional data cube suitable for data mining. Finally, based on the representation of multidimensional data model, the top-down method is used, then, reduces the data processing, and the knowledge rules are generated by using rough set theory.