Automatic Classification Of Data Research Articles

A Novel Automatic Map Matching Method Based on Hybrid Computing Framework of Hidden Markov Model and Conditional Random Field

In recent years, automatic map matching has received great technical progress. However, when it comes to vague matching situations such as improper vocabulary use, there still lack reliable solutions. To handle the current gap, this paper proposes a novel automatic map matching method based on the hybrid computing framework of hidden Markov model (HMM) and conditional random field. First, the data filtering is completed by performing second-order transformation towards automatic matching conditions of HMM. Then, the data classification is completed using automatic data classification based on the conditional random field. After that, a hybrid computing framework with spatial elements and layer selection is built to generate map matching results. Finally, some simulation experiments are conducted for evaluation. For one thing, the trend of matching accuracy changes under specified conditions is basically the same as that of nonspecified conditions. The maximum difference in matching calculation values is about 3 times. However, once the vocabulary continues to increase, the difference in matching results between the two narrows to 10–20%. For the other thing, the matching accuracy of a specified state is higher than that of sending a specified state. While nonspecified fuzzy matching accuracy is about 3 times higher and nonspecified precision matching accuracy is about 50% higher.

Classifying Gait Alterations Using an Instrumented Smart Sock and Deep Learning

This article presents a noninvasive method of classifying gait patterns associated with various movement disorders and/or neurological conditions, utilizing unobtrusive, instrumented socks and a deep-learning network. Seamless instrumented socks were fabricated using three accelerometer-embedded yarns, positioned at the toe (hallux), above the heel, and on the lateral malleolus. Human trials were conducted on 12 able-bodied participants, an instrumented sock was worn on each foot. Participants were asked to complete seven trials consisting of their typical gait and six different gait types that mimicked the typical movement patterns associated with various movement disorders and neurological conditions. Four neural networks and an SVM were tested to ascertain the most effective method of automatic data classification. The bi-long short-term memory (LSTM) generated the most accurate results and illustrates that the use of three accelerometers per foot increased classification accuracy compared to a single accelerometer per foot by 11.4%. When only a single accelerometer was utilized for classification, the ankle accelerometer generated the most accurate results in comparison to the other two. The network was able to correctly classify five different gait types: stomp (100%), shuffle (66.8%), diplegic (66.6%), hemiplegic (66.6%), and “normal walking” (58.0%). The network was incapable of correctly differentiating foot slap (21.2%) and steppage gait (4.8%). This work demonstrates that instrumented textile socks incorporating three accelerometer yarns were capable of generating sufficient data to allow a neural network to distinguish between specific gait patterns. This may enable clinicians and therapists to remotely classify gait alterations and observe changes in gait during rehabilitation.

An automatic classification method of library archives data based on data mining

Aiming at the problems of poor accuracy of data feature extraction and large classification error in library archives data classification methods, an automatic classification method of library archives data based on data mining is designed. Firstly, the linear relationship between the characteristic variables of library archives data is determined, and the linear coefficient of archives data characteristics is calculated; Then, the characteristic states of library archives data are divided into three states, the characteristic data are normalized, and the adaptive differential evolution algorithm is used to remove the noise in the characteristics of library archives data; Finally, the mapping relation training model in data mining is used to input the data feature training set, and the file data features are labeled according to different weights; Establish automatic data classification model. The experimental results show that the highest accuracy of this method is about 97%.

Evaluating hierarchical machine learning approaches to classify biological databases.

The rate of biological data generation has increased dramatically in recent years, which has driven the importance of databases as a resource to guide innovation and the generation of biological insights. Given the complexity and scale of these databases, automatic data classification is often required. Biological data sets are often hierarchical in nature, with varying degrees of complexity, imposing different challenges to train, test and validate accurate and generalizable classification models. While some approaches to classify hierarchical data have been proposed, no guidelines regarding their utility, applicability and limitations have been explored or implemented. These include ‘Local’ approaches considering the hierarchy, building models per level or node, and ‘Global’ hierarchical classification, using a flat classification approach. To fill this gap, here we have systematically contrasted the performance of ‘Local per Level’ and ‘Local per Node’ approaches with a ‘Global’ approach applied to two different hierarchical datasets: BioLip and CATH. The results show how different components of hierarchical data sets, such as variation coefficient and prediction by depth, can guide the choice of appropriate classification schemes. Finally, we provide guidelines to support this process when embarking on a hierarchical classification task, which will help optimize computational resources and predictive performance.

Understanding the Adoption of Dietary Interventions Within a Chinese Autism Online Community: A Diffusion of Innovations Perspective

ABSTRACT Dietary interventions are common but controversial treatments for autistic people. This study aims to understand the adoption of dietary interventions based on diffusion of innovations theory in the autism online community from four aspects: popularity, adoption process, the influence of opinion leaders, and post-adoption feedback. Our data was extracted from a Chinese autism community named Baidu Tieba autism forum. We applied a mixed-method including four analytical approaches: descriptive statistics for popularity analysis; machine learning models for automatic data classification and topic detection; social network analysis for exploring the influence of opinion leaders on the adoption phase; content analysis for revealing the family caregiver-reported feedback after adoption. Dietary interventions have become increasingly popular in the autism online community since 2018. Analysis of the adoption process revealed that family caregivers at different stages of adoption focused on different topics, and the number of interactions with opinion leaders had a significant effect on the highest level (p < .001) and stage span (p < .001) of health information adoption. According to findings from the feedback of family caregivers, the effects of dietary interventions varied from individuals with autism. Our study revealed the diffusion of unproven interventions, which is of great significance in promoting evidence-based practices.

Prediction and modelling online reviews helpfulness using 1D Convolutional Neural Networks

The latest research shows that the identification of helpful reviews from a large volume of user–generated data is a trending topic. The present study uses an approach that not only predicts if an online review is helpful, neutral or not helpful with 66% accuracy, but most importantly models online review helpfulness. To this end, we use an adaptive implementation of 1D Convolutional Neural Networks (CNNs). The neuronal encoding of CNNs has the benefit of obtaining automatic data classification using cluster analysis to detect different types of clusters of helpful and not helpful reviews, categorized by their most important contextual characteristics. Findings reveal that the clusters with the most important words and documents for helpful reviews in the product category ‘Cars & Motorcycles’ describe cars and their characteristics, whereas not helpful reviews concern details about car-related shops/companies in general. By demonstrating high performance on prediction and classification of review helpfulness with our proposed methodology, we are contributing to the research on business intelligence. In addition, we provide significant practical implications for marketers, enabling them to distinguish between helpful and not helpful reviews. Using the resulting encoding can produce automatic data classification of different clusters of specific topics.

Secure File/Data Transfer Between Airgap Network

Wireless Data communication is fastest growing technology era in which the research society has recently embarked. Today, Computer data can include financial transactions such as electronic payments, M- wallets and sensitive multimedia contents. The explosive volumes of computer devices personal data, bring-up more attention to securely data storage rather than consideration on data privacy and confidentiality levels. In this scenario Air Gap Data Communication, Machine Leaning (ML) and image processing brings an important role in the electronic data management. It is always expensive and hard to manage the data manually without adopting machine learning and image processing techniques using metadata. The contribution of this research article is to demonstrate a securing computer data storage secrecy and privacy in cloud communication framework in terms of automatic data classification using computer training datasets with help of Training dataset which classifies the data based on the confidentiality level of the record with higher accuracy and powerful timelines as compared to the traditional KNN algorithms and RSA algorithm securing such confidential data category afterwards by applying various existing cryptographic solutions to assuring data privacy, confidentiality levels and alerting the use of abusive contents and simulation results demonstrates that reducing the overall cost. Training dataset which classifies the data based on the confidentiality level of the record with higher accuracy and powerful timelines as compared to the traditional KNN algorithms and RSA algorithm securing such confidential data category.

Implementation of a Cloud Computer-based Automatic Classification and Processing System for Large Data

Data technology is an important technology applied in the process of social development, which plays a very important role in data and information operation, storage and analysis. It is directly related to the accuracy of data operation and promotes the efficiency of the operation of a data platform to some extent. This paper analyzes cloud computing technology and large data processing, and puts forward a new automatic data classification and processing system, which can ensure the operation of the system more efficiently.

Clustering as a tool to support the assessment of power quality in electrical power networks with distributed generation in the mining industry

This paper presents a case study of using cluster analysis (CA) as one of the data mining (DM) techniques applied in the analysis of long-term power quality (PQ) data that is recorded in electrical power networks of the mining industry. The aim of the clustering is to highlight the impact of distributed generation (DG) on the level of power quality parameters. The carried out investigations concern the application of the K-mean clustering algorithm with Euclidean and Chebyshev distance with a different number of clusters for standardised and non-standardised data. The obtained results show the possibility to obtain automatic classification of data into distinguishable clusters that represent the period of time when local DG is active, switched-off or when a different power consumption level is denoted. It leads to the possibility of using CA as a suitable tool for assessing the impact of local generation on the working conditions of electrical power networks that depend on a DG contribution or power consumption. Additionally, the obtained results allow CA to be indicated as a proper method for the automatic identification of the PQ data which are affected by voltage events that can be treated as alternative way for present flagging concept.

PRM75 - DEVELOPMENT OF MACHINE LEARNING BASED ABSTRACT DOCUMENT CLASSIFICATION FOR SUPPORTING SYSTEMATIC REVIEWS

Machine learning is a predictive computational method which enables rapid automatic data classification, and thus could replace manual and time-consuming work. Prior to its implementation, however, careful evaluation of the prediction accuracy is necessary. This study evaluated the possibility of using a machine learning method for abstract document classification of research articles in the process of systematic review. We used Long Short-Term Memory networks (LSTM), a type of recurrent neural network architecture, in order to understand the contexts of the abstracts. Owing to the limited number of articles in one dataset, we pre-trained the LSTM by feeding the model sequences of words, and subsequently performed virtual adversarial training to stabilize LSTM learning. We used two distinct datasets. The first dataset contained 1307 articles which had previously been selected manually. Five exclusion criteria were applied to metadata about each article, and on this basis were either qualified for use (categorized into a “use” group) or excluded (categorized into “non-use” group). From each group, 60%, 20%, and 20% of randomly selected data were used for training, validation, and testing of LSTM, respectively. In the test phase, 97% of the data was correctly classified into the “non-use” group (3% false positives), whereas only 73% of data was correctly classified into the “use” group (27% false negatives). The second dataset contained 1023 articles. In the test phase, the percentage of false positives and negatives was 5% and 23%, respectively. Our results indicate that although the false positive classification was sufficiently low, the false negative classification was not negligible. Reducing false negatives is necessary for the implementation of this approach as false negatives are more critical for supporting systematic reviews than false positives.

Catching Earworms on Twitter

In recent years, so-called big data research has become a hot topic in the social sciences. This paper explores the possibilities of big data-based research within the field of music psychology. We illustrate one methodological approach by studying involuntary musical imagery, or earworms in the social networking service Twitter. Our goal was to collect a large naturalistic and culturally diverse database of discussions and to classify the encountered expressions. We describe our method and present results from automatic data classification and sentiment analyses. Over six months, we collected over 80,000 tweets from 173 locations around the world to obtain the most diverse dataset collated to date related to involuntary musical imagery. Automated classifications categorized 51% of all tweets gathered, with over 90% accuracy in each category. The most prominent categories of discussion concerned reporting earworm experiences, hyperlinks to music, spreading general information about the phenomenon, and communicating thankfulness (sincerely or ironically) about receiving earworms. Sentiment analysis revealed a balance towards negative emotional expressions in comparison to reference data. This is the first study to show this negative appraisal tendency and to demonstrate the ‘earworm’ phenomenon on a global scale. We discuss our findings in relation to previous literature and highlight the opportunities and challenges of big data research.

An Approach to Supporting Incremental Visual Data Classification.

Automatic data classification is a computationally intensive task that presents variable precision and is considerably sensitive to the classifier configuration and to data representation, particularly for evolving data sets. Some of these issues can best be handled by methods that support users' control over the classification steps. In this paper, we propose a visual data classification methodology that supports users in tasks related to categorization such as training set selection; model creation, application and verification; and classifier tuning. The approach is then well suited for incremental classification, present in many applications with evolving data sets. Data set visualization is accomplished by means of point placement strategies, and we exemplify the method through multidimensional projections and Neighbor Joining trees. The same methodology can be employed by a user who wishes to create his or her own ground truth (or perspective) from a previously unlabeled data set. We validate the methodology through its application to categorization scenarios of image and text data sets, involving the creation, application, verification, and adjustment of classification models.

Modelling the life cycle of machines that accumulate hidden damage

A model of damage accumulation of the node structure of a complex machine during its operation is developed.In predicting the lifecycle of complex technical structure at its design stage, “passport” structure of analyzed option, data of bench tests of trial sample and also statistical information on external conditions and “behavior” during field testing and operation of products-analogues are used as the initial information.Such versatility and diversity of the model allows to consider inaccessible to observation parts of the object since the state of the latter indirectly affects the processes in accessible parts.On the example of a trailer welded frame, the problem of modeling the life cycle of complex technical system in mechanical engineering is considered. Mathematical apparatus of the automatic classification of data about the current state of the system to construct a model of damage accumulation in the frame is proposed.

Auroral Sequence Representation and Classification Using Hidden Markov Models

The naturally occurring aurora phenomenon is a dynamically evolving process. Taking temporal information into consideration, the auroral image sequence analysis is more reasonable and desirable than using static images only. However, the enormous richness of space structures and temporal variations make automatic auroral sequence analysis a particularly challenging task. In this paper, a hidden Markov model (HMM) based representation method including features of spatial texture and dynamic evolution is presented to characterize auroral image sequences captured by all-sky imagers (ASIs). The uniform local binary patterns are employed to describe the 2-D space structures of ASI images. HMM is feasible to characterize the doubly stochastic process involved in the auroral evolution-measurable polar light activities and hidden dynamic plasma processes. We present an affine log-likelihood normalization technique to manage the sequences with different lengths. The proposed method is used in the automatic recognition of four primary categories of ASI auroral observations between the years 2003 and 2009 at the Yellow River Station, Ny-Ålesund, Svalbard. The supervised classification results on manually labeled data in 2003 demonstrate the effectiveness of the proposed technique. Compared with frame-based classification, the higher accuracies and the lower rejection rates show the advantages of the sequence-based method. The occurrence distributions of the four aurora categories were obtained through automatic classification of data gathered from 2004 to 2009. Their agreement with the multiple-wavelength intensity distribution of the dayside aurora and the conclusions made from the frame-based method further illustrate the validity of our method on auroral representation and classification.

Versatile two‐state land mobile satellite channel model with first application to DVB‐SH analysis

AbstractStandardization activities on Digital Video Broadcasting–Satellite services to Handheld Devices (DVB‐SH) have driven the need for a consolidated Land Mobile Satellite (LMS) narrowband channel model. In the DVB‐SH system, the satellite broadcasts a signal carrying multimedia services aimed directly to a variety of mobile (handheld or vehicular) and fixed terminals.A three‐state LMS channel model that describes the narrowband propagation channel in three possible shadowing states—line‐of‐sight conditions, moderate shadowing and deep shadowing—had been selected as a baseline for physical layer simulation of the DVB‐SH waveform. This type of model, capable of generating complex time series, was originally selected, because it is the simplest model that allows the simulation of first‐ and second‐order effects of the LMS channel in a realistic manner. The main limitations of such model are, first of all, that a classification in three states does not necessarily correspond with reality and, secondly, that the statistical parameters for each state were fixed for a given scenario and elevation angle. Those limitations may impact the selection of Physical Layer parameters of the DVB‐SH standard.A new channel model is proposed based on the original three‐state model including two major modifications: a reduction in the number of states and the introduction of a versatile selection of statistical parameters describing each state. Furthermore, the state machine is governed either by Markov or by semi‐Markov chains. The new‐state classification does not necessarily correspond to intuitive physical definitions of the states as before (line‐of‐sight, shadowing) but instead to channel variations that share similar statistical characteristics. The two‐states are termed for convenience, Good and Bad states, representing a range of LOS‐to‐moderate shadowing and moderate‐to‐deep shadowing, respectively. For the model parameters selection, datasets at L‐ and S‐band have been analysed using an iterative algorithm that includes automatic data classification and parameter extraction. The proposed model is considered more suitable for the analysis of DVB‐SH test cases.This study starts with an overview of the main DVB‐SH system parameters and assumptions. The original three‐state model is briefly introduced; the new model is presented in detail, including simulator implementation. Finally, both models and experimental data sets are compared on a statistical basis. The performance of both models are discussed to show how effective the model is for the representation of shadowed conditions and therefore, its suitability for the analysis and optimal configuration of the physical and link layer parameters (namely physical layer interleaver size, link layer protection time, overall redundancy, etc.). Copyright © 2010 John Wiley &amp; Sons, Ltd.

Spatial texture based automatic classification of dayside aurora in all-sky images

A spatial texture based representation method including features of intensity, shape and texture, was utilized to characterize all-sky auroral images. The combination of the local binary pattern (LBP) operator and a delicately designed block partition scheme achieved both global shapes and local textures capabilities. The representation method was used in automatic recognition of four primary categories of discrete dayside aurora using observations between years 2003–2009 at the Yellow River Station, Ny-Ålesund, Svalbard. The supervised classification results on labeled data in 2003 were in accordance with the labeling by scientists considering both spectral and morphological information. The occurrence distributions of the four categories were obtained through automatic classification of data between 2004–2009, which confirm the multiple-wavelength intensity distribution of dayside aurora, and further provide morphological interpretation of auroral types.

An extensive library of 2500–10 500 Å synthetic spectra

We present a complete library of synthetic spectra based on Kurucz’s codes that covers the 2500–10 500 A wavelength range at resolving powers RP = 20 000, 11 500 (≡GAIA), 8500 (≡RAVE), 2000 (≡SLOAN) and uniform dispersions of 1 and 10 A/pix. The library maps the whole HR diagram, exploring 51 288 combinations of atmospheric parameters spanning the ranges: 3500 ≤ Teff ≤ 47 500 K, 0.0 ≤ log g ≤ 5.0, −2.5 ≤ [M/H] ≤ 0.5, [α/Fe] = 0.0,+0.4, ξ = 1, 2, 4 km s −1 ,0 ≤ Vrot ≤ 500 km s −1 . The spectra are available both as absolute fluxes as well as continuum normalized. Performance tests and spectroscopic applications of the library are discussed, including automatic classification of data from spectroscopic surveys like RAVE, SLOAN, GAIA. The entire library of synthetic spectra is accessible via the web.

A spatial density analysis technique for the automatic classification of 4/spl pi/ detector data

In nuclear physics at intermediate energies, the use of multidetector arrays consisting of thousands of detectors is customary. This paper presents a novel approach to the automatic classification of data collected by the 4/spl pi/ detector CHIMERA, constituted by 1192 telescopes of a thin silicon transparent detector coupled to a thick CsI(Tl) scintillator. The method is based on spatial density data processing and produces frequency distributions of the most significant physical parameters. These one-dimensional spectra can be more easily analyzed than the standard two-dimensional scatter plots. With respect to manual or semiautomatic procedures, presently used for data analysis, our method offers the advantages of drastic time reduction and unbiased criteria for cluster identification. The proposed method can be successfully used in both off-line data analysis and multidetector stability control.

A contextual image segmentation system using a priori information for automatic data classification in nuclear physics

This paper presents an original approach to solve an automatic data classification problem by means of image processing techniques. The classification is achieved using image segmentation techniques for extracting the meaningful classes. Two types of information are merged for this purpose: the information contained in experimental images and a priori information derived from underlying physics (and adapted to image segmentation problem). This data fusion is widely used at different stages of the segmentation process. This approach yields interesting results in terms of segmentation performances, even in very noisy cases. Satisfactory classification results are obtained in cases where more “classical” automatic data classification methods fail.