Academic Dataset Research Articles

Extended Advanced Method of Clustering Big data to achieve high dimensionality

Clustering is one of the relevant knowledge engineering methods of data analysis. The clustering method will automatically directly affect the result dataset. The proposed work aims at developing an Extended Advanced Method of Clustering (EAMC) to address numerous types of issues associated to large and high dimensional dataset. The proposed Extended Advance Method of clustering will repetitively avoid computational time between each data cluster object contained by the cluster that saves execution time in term. For each iteration EAMC needs a data structure to store data that can be utilized for the next iteration. We have gained outcomes from the proposed method, which demonstrates that there is an improvement in effectiveness and pace of clustering and precision generation, which will decrease the convolution of computing over the old algorithms like SOM, HAC, and K-means. This paper includes EAMC and the investigational outcomes done using academic datasets.

Data Transformation Techniques for Academic Datasets

Data mining is a real-world procedure of discovering useful patterns from heterogeneous datasets. All most all industry uses data mining in their day to day activities. To build an effective mining model, a series of development steps are to be followed. It starts with discovering the business problem and ends with communicating the results. In this development life cycle, the most important step is data preparation or data preprocessing. Data preprocessing is converting raw data into data understandable by the machine. Data normalization is a phase in data preprocessing where the data values are scaled to 0 and 1. Right normalization of the datasets leads to improved mining results. In this paper, academic data of students is taken. The dataset is normalization using six normalization technique. Multi Layer Perceptron classifier is applied to normalized dataset and results are obtained. Results of this study reveal the best normalization technique which can be used for normalizing academic datasets. Finally, in a line, the goal of this work is to discover the best normalization technique which produces better mining result when applied to academic datasets.

Course Sequence Recommendation with Course Difficulty Index Using Subset Sum Approximation Algorithms

Abstract Choice Based Course Selection (CBCS) allows students to select courses based on their preferred sequence. This preference in selection is normally bounded by constraints set by a university like pre-requisite(s), minimum and maximum number of credits registered per semester. Unplanned course sequence selection affects the performance of the students and may prolong the time to complete the degree. Course Difficulty Index (DI) also contributes to the decline in the performance of the students. To overcome these difficulties, we propose a new Subset Sum Approximation Problem (SSAP) aims to distribute courses to each semester with approximately equal difficulty level using Maximum Prerequisite Weightage (MPW) Algorithm, Difficulty Approximation (DA) algorithm and Adaptive Genetic Algorithm (AGA). The three algorithms have been tested using our university academic dataset and DA algorithm outperforms with 98% accuracy than the MPW and AGA algorithm during course distribution.

Virtual Reality and Computer Simulation in Social Work Education: A Systematic Review

ABSTRACTThe use of virtual reality and computer simulation are gaining ground in social work as viable teaching methods. Traditionally recognized as ancillary to classroom learning, they are being incorporated strategically into course curricula. A systematic review was undertaken to answer the questions, How are virtual reality and computer simulation technology being used to train social workers, and are they effective? Seven academic data sets and the Council on Social Work Education and the Society for Social Work and Research conference abstract databases were searched. Based on inclusion criteria, seven full articles were included revealing five themes. The technology is primarily used to teach direct practice rather than macro-level skills and is geared toward the education of students rather than practitioners.

Effective Feature Selection Strategy using Manova Test

Feature selection is the most important step to develop any latest learning model. As the complexity of the leaning models increases day by day there is an increasing demand, in selecting the right features to build the model. There are many methods for feature selection. A new feature selection based on the Manova statistical test is implemented. Using the Manova test, we select attributes from academic datasets. Using the selected attributes, we build a classification model. Accuracy of the model with feature selection is compared with a model with all attributes. Results are discussed. It is proved that the classification model build with features selected by Manova test achieves more accuracy than a model built with all features.

A novel hybrid publication recommendation system using compound information

Publication recommendation is an interesting but challenging research problem. Most existing studies only use partial information of papers’ contents, reference network or co-author relationship, which leads to an unsatisfied recommendation result. In this study, we propose a novel hybrid publication recommendation approach using compound information which retrieves top-K most relevant papers from a publication depository for a set of user input keywords. Our advantages comparing to the existing methods include: (1) Reaching a better recommendation results by taking the advantages of both content-based recommendation and citation-based recommendation and exploring much richer information of papers in one method; (2) Effectively solving the cold-start problem for new published papers by considering the vitality of papers and the impact factor of venues into the citation network; (3) Saving a large overhead in calculating the content-based similarity between papers and user input keywords by doing paper clustering based on the citation network. Extensive experiments on DBLP and Microsoft Academic datasets demonstrate that PubTeller improves the state-of-the-art methods with 4% in Precision and 4.5% in Recall.

Global and Peripheral Identities in the Production of Knowledge on Higher Education Reforms: The Latin American Case

In a globalized world of knowledge production, issues arise as to the presence of, and relationships between, different regions of the world. A particular issue is that of the presence of peripheral regions such as Latin America. The focus here is research into higher education reforms and, by using bibliometrics, a word frequency analysis and a thematic analysis, patterns of publication and themes in two different academic datasets — the global Web of Science and the more regional SciELO — are examined. Although the configuration of knowledge production on higher education reforms is seen as relatively homogeneous in certain respects, separate circuits of knowledge production are also revealed and even within a region (here, Latin America). Higher education reform is posed differently according to the publication circuits of journals and their position in global knowledge markets. A geopolitical imbalance in knowledge production is revealed and it is suggested that different regions of the world possess distinctive epistemic identities. The paper concludes by suggesting that there is room for negotiation between the different zones of epistemic influence.

S PRAR: A novel relational association rule mining classification model applied for academic performance prediction

This paper analyses the problem of predicting students’ academic performance, a subject that is increasingly investigated within the Educational Data Mining literature. For a better understanding of the educational related phenomena, there is a continuous interest in applying supervised and unsupervised learning methods for obtaining additional insights into the students’ learning process. The problem of predicting if a student will pass or fail at a certain academic discipline based on the students’ grades received during the semester is a difficult one, highly dependent on various conditions such as the course, the number of examinations during the semester, the instructors and their exigences. We propose a new classification model, S PRAR (Students Performance prediction using Relational Association Rules) for predicting the final result of a student at a certain academic discipline using relational association rules (RARs). RARs extend the classical association rules for expressing various relationships between data attributes. Experiments are performed on three real academic data sets collected from Babeş-Bolyai University from Romania. The performance of the S PRAR classifier on the considered case studies is compared against existing related work, being superior to previously proposed students’ performance predictors.

Nonparametric Estimation of Price Elasticities: A Heterogeneous Treatment Effect Approach

We revisit the classical problem of price elasticity estimation from a causal perspective. When the price is perceived as a continuous but endogenous treatment, the flexible estimation of price elasticity can be turned into the estimation of heterogeneous treatment effects. To this end, we develop a control function approach to deal with treatment endogeneity when estimating heterogeneous treatment effects. This strategy works by breaking the estimation of price elasticity into several intermediate problems of point-wise expectation estimation, where modern machine learning methods, such as deep neural networks and random forests, can be used for prediction. In addition, we prove that if we use the bagged nearest neighbors for point-wise prediction, the standard bootstrap procedure can be directly employed to derive inference for the price elasticity estimates. Finally, we apply our method to the IRI academic dataset on two national brands of yogurt. It is found that the competitor's yogurt in a similar size is more a substitute compared to the own brand's yogurt in a different size.

A Unified Link Prediction Framework for Predicting Arbitrary Relations in Heterogeneous Academic Networks

Most of the existing link prediction methods for heterogeneous academic networks can only predict one or two specific relation types rather than arbitrary relation types. Although several recently proposed methods have involved multi-relational prediction problems, they do not comprehensively consider the rich semantic or temporal information of heterogeneous academic networks. Considering that researchers may have diverse requirements for different types of academic resources, in this study, we propose a new unified link prediction framework (UniLPF) for arbitrary types of academic relations. First, a weighted and directed heterogeneous academic network containing rich academic objects and relations is constructed. Then, an automatic meta-path searching method is proposed to extract the meta-paths for arbitrary prediction tasks. Two meta-path based object similarity measures combining temporal information and content relevance are also proposed to measure the features of the meta-paths. Finally, a pervasive link prediction model is built, which can be embodied based on an arbitrarily specified prediction task and the corresponding meta-path features. Extensive experiments for predicting various relation types with practical significance are conducted on a large-scale Microsoft Academic dataset. The experimental results demonstrate that our proposed UniLPF framework can predict arbitrary specified academic relations, and outperforms the comparison methods in terms of F-measure, accuracy, AUC and ROC. In addition, the time scalability experiments prove that UniLPF also achieves good performance for predicting the academic relations over time.

Intent Classification for Dialogue Utterances

In this work, we investigate several machine learning methods to tackle the problem of intent classification for dialogue utterances. We start with bag-of-words in combination with Naive Bayes. After that, we employ continuous bag-of-words coupled with support vector machines (SVM). Then, we follow long short-term memory (LSTM) networks, which are made bidirectional. The best performing model is hierarchical, such that it can take advantage of the natural taxonomy within classes. The main experiments are a comparison between these methods on an open sourced academic dataset. In the first experiment, we consider the full dataset. We also consider the given subsets of data separately, in order to compare our results with state-of-the-art vendor solutions. In general we find that the SVM models outperform the LSTM models. The former models achieve the highest macro-F1 for the full dataset, and in most of the individual datasets. We also found out that the incorporation of the hierarchical structure in the intents improves the performance.

Academic Venue Recommendations Based on Similarity Learning of an Extended Nearby Citation Network

The rapidly increasing number of potential academic venues for research publication and commentary has made sourcing the venue that would best contribute to promoting effective scientific cooperation more challenging. In this paper, we propose a similarity learning approach to determine the most appropriate venue to publish an article. We first analyze the article metadata and cited articles to build the citation network matrices of the given article and then apply these to learn and build similarity matrices between academic objects (i.e., articles, authors, and venues) at an extended nearby article citation network. Using the formed matrices, we estimate a collaborative anticipation confidence score of a relationship between the venues in the extended network. For our empirical studies, we used an actual academic dataset to validate the efficiency of our approach and recommend an appropriate academic venue. The experimental results highlight the effectiveness of our proposed approach to optimize overall recommendation quality, compared with other baseline approaches.

Visual Analysis of Citation Context-Based Article Influence Ranking

Article influence ranking is an effective way to reduce information redundancy and improve the efficiency of article retrieval. A large number of ranking models for network items have been employed for the ranking of article influence, such as PageRank and Spamming-resistant Expertise Analysis and Ranking. However, the effectiveness of article influence ranking based on the models of PageRank and SPEAR declines with the rapid growth of academic datasets, because of the increasing complexity of citation network. In order to take a rich set of contextual structures of citation context into consideration, we propose a visualization system VAIR for the citation context-based article influence ranking. Firstly, the word2vec model, a renowned technique in the field of natural language processing, is applied to transform articles into vectorized representations according to citation context. Then, a novel citation context-based article influence ranking model is designed according to the complex relationships quantified in a semantic vectorised space. Several visual designs are implemented, allowing users to perceive and compare the ranking results visually and intuitively. A set of user-friendly interactions are provided in the visualization framework, enabling users to explore the desirable article influence and obtain deep insights into the ranking model. Moreover, a series of case studies and comparison experiments are carried out based on real-world datasets, which further demonstrate the effectiveness of our algorithm for article influence ranking.

Finalizing your reference list with machine learning

A high quality reference list is important to the overall quality of a research paper. However, it requires domain knowledge and is time consuming to generate a reference list with good coverage, representativeness, and timeliness due to the large amount and fast growing of publications. In this paper, we deal with the specific problem of reference enhancement of research manuscripts with machine learning. A predictive model is trained by a large academic dataset with paper-related and venue-related information to discover additional references for a scientific draft with related information including an initial reference list. We propose a supervised approach called RefCom under the framework of learning-to-rank to predict the probability for a given paper to cite a reference candidate. Forty features in total are defined to describe pairs of papers with respect to author influence, venue influence and paper influence, as well as content and reference similarity. Unlike heuristic rule-based approaches, RefCom is able to integrate multiple features with learned weights. Experimental study with the AMiner dataset which contains 2 million papers and 1.7 million authors show the effectiveness of RefCom in citation prediction, suggesting its potential of being applied as an assistant tool in reference finalization.

BIGOWL: Knowledge centered Big Data analytics

Knowledge extraction and incorporation is currently considered to be beneficial for efficient Big Data analytics. Knowledge can take part in workflow design, constraint definition, parameter selection and configuration, human interactive and decision-making strategies. This paper proposes BIGOWL, an ontology to support knowledge management in Big Data analytics. BIGOWL is designed to cover a wide vocabulary of terms concerning Big Data analytics workflows, including their components and how they are connected, from data sources to the analytics visualization. It also takes into consideration aspects such as parameters, restrictions and formats. This ontology defines not only the taxonomic relationships between the different concepts, but also instances representing specific individuals to guide the users in the design of Big Data analytics workflows. For testing purposes, two case studies are developed, which consists in: first, real-world streaming processing with Spark of traffic Open Data, for route optimization in urban environment of New York city; and second, data mining classification of an academic dataset on local/cloud platforms. The analytics workflows resulting from the BIGOWL semantic model are validated and successfully evaluated.

Denoising with kernel prediction and asymmetric loss functions

We present a modular convolutional architecture for denoising rendered images. We expand on the capabilities of kernel-predicting networks by combining them with a number of task-specific modules, and optimizing the assembly using an asymmetric loss. The source-aware encoder---the first module in the assembly---extracts low-level features and embeds them into a common feature space, enabling quick adaptation of a trained network to novel data. The spatial and temporal modules extract abstract, high-level features for kernel-based reconstruction, which is performed at three different spatial scales to reduce low-frequency artifacts. The complete network is trained using a class of asymmetric loss functions that are designed to preserve details and provide the user with a direct control over the variance-bias trade-off during inference. We also propose an error-predicting module for inferring reconstruction error maps that can be used to drive adaptive sampling. Finally, we present a theoretical analysis of convergence rates of kernel-predicting architectures, shedding light on why kernel prediction performs better than synthesizing the colors directly, complementing the empirical evidence presented in this and previous works. We demonstrate that our networks attain results that compare favorably to state-of-the-art methods in terms of detail preservation, low-frequency noise removal, and temporal stability on a variety of production and academic datasets.

An Accurate Method Based on Deep Learning Architectures for Real-world Object Detection

Object detection and recognition play an important role in blind navigation. However, in real-world shooting, images are degraded and deviate from the statistical distribution of academic datasets, which have a bad impact on object detection. Here, using deep neural networks and door detection as an example, we simulate problems that the blind may encounter when shooting images, such as lack of illumination in the imaging environment, relative motion with the object in exposure moment, rotation and jitter of the photographic apparatus and the like. After establishing a mathematical model of image degradation, we compare and demonstrate the impact of various image degradation on door detection. By degenerating the training set, we train a robust model that improves the average precision (AP) of the door detection in real scenes and outperforms other training methods.

The duplication issue within the Drebin dataset

The Drebin dataset (in: NDSS, 2014) is the most supplied academic dataset of Android malware. Therefore it is the most used dataset in research papers on Android malware detection. The research community is using it for evaluation and comparison of their algorithms. We discovered that 49.35% of samples in this dataset has at least one other sample that is a repackaged version containing exactly the same sequence of opcode. The only differences between the original malware and the duplicated ones, in all cases, are the resources embedded and some strings in the code. For assessing the performance of malware detectors or classifiers, a part of the dataset is used for this purpose. So a major part of the testing set end up beeing the same samples that have been used in the training set. This situation can lead us, the research community, to overrate the performance of algorithms we are designing. In the worst case, it leads us to wrong conclusions and wrong directions for future research. Then we conduct an experiment where we test several classification algorithms on the Drebin dataset with and without the duplicates. Our results show that depending on the classifier the full dataset can lead from moderately (124%) to strongly (172%) underrated inaccuracy, and the order of performance of the algorithms is modified. Finally we provide the list of unique malware samples from the Drebin dataset, available on Github.

Using Ensemble StackingC Method and Base Classifiers to Ameliorate Prediction Accuracy of Pedagogical Data

Ensemble methods and conventional base class learners have effectively been applied in the realm of educational data mining to ameliorate the accuracy and consistency in prediction. Primarily in the contemporary study, researchers conducted empirical results on pedagogical real dataset acquired from University of Kashmir, using miscellaneous base classifiers viz. j48, random forest and random tree, to predict the performance of students. However, in the later phase, the pedagogical dataset was subjected to more proficient version of stacking viz. stackingC, with the principle objective to ameliorate the performance of students. Furthermore, the dataset was deployed with filtering procedures to corroborate any improvement in results, after the application of techniques such as synthetic minority oversampling technique (SMOTE) and spread sub-sampling method. Moreover, in case of ensemble stackingC, hybridization of predicted output was carried out with three base classifier vis-a- vis j48, random forest and random tree, and the classifier achieved paramount accuracy of 95.65% in predicting the actual class of students. The findings have by and large noticeably corroborated that the stackingC classifier, attained significant prediction accuracy of 95.96% when undergone through undersampling (spread sub-sampling) and 96.11% using oversampling (SMOTE). As a subject of corollary, it calls upon the researchers to broaden the canvas of literature by employing the analogous methods to uncover the diverse patterns hidden in academic datasets.

Finding linkage between sustainability science and complex networks

Sustainability science has been widely accepted worldwide as an important research field worldwide. Since it is widespread in many research fields such as environmental science and engineering, architecture, materials science, and civil engineering, it is important to discover the hidden connections between sustainability science and other academic fields. For this purpose, supporting tools are necessary. This paper investigates the relationship between sustainability science and complex networks on the basis of citation network analysis using large datasets of academic papers. First, citation networks are generated on the basis of academic paper datasets and clustered to understand the research fields of sustainability science and complex networks. Next, we measure the textual similarities among the clusters in these fields and analyse the shared terms to find the overlapping research areas that connect the two fields. These results offer overviews of how complex networks contribute to sustainability science and assist the formation of policies that promote key results in artificial intelligence toward future sustainable societies.