A Novel Method to Associate Sensor Data with Domain Ontology
Jin Liu,
Yihe Yang,
Shengjie Shang
Issue:
Volume 5, Issue 4, August 2019
Pages:
52-60
Received:
6 July 2019
Accepted:
26 July 2019
Published:
16 August 2019
Abstract: With the development of the Internet of Things, sensor ontologies have been applied to a variety of fields. Most sensor ontologies are currently built for applications in specific domains, and these ontologies are usually heterogeneous, making it difficult to share or reuse knowledge and concepts. The ontology association methods can be used to construct the semantic mapping between heterogeneous ontologies, so as to effectively determine the similarity between concepts in the ontologies. However, most of the contemporary methods do not make full use of the information that is stored in ontologies and are insufficient for the effective association. This paper proposes a novel association method based on comprehensive similarity. In our proposed method, we first use How-Net to obtain concept representation and calculate the semantic similarity of ontology concepts through sememe Tree and sememe Hierarchy. Then we calculate the structural similarity by the internal structure and the hierarchical relationship between the ontologies and remove the conceptual pairs with low relevance. Finally, we combine the semantic similarity and structural similarity to calculate the similarity matrix between ontology concepts to achieve association. The experimental results on real data show that our method can effectively associate sensor data with domain ontology by combining two different similarity calculation methods.
Abstract: With the development of the Internet of Things, sensor ontologies have been applied to a variety of fields. Most sensor ontologies are currently built for applications in specific domains, and these ontologies are usually heterogeneous, making it difficult to share or reuse knowledge and concepts. The ontology association methods can be used to con...
Show More
Student’s Academic Performance Prediction Using Factor Analysis Based Neural Network
Shamsuddeen Suleiman,
Ahmad Lawal,
Umar Usman,
Shehu Usman Gulumbe,
Aminu Bui Muhammad
Issue:
Volume 5, Issue 4, August 2019
Pages:
61-66
Received:
6 July 2019
Accepted:
26 July 2019
Published:
26 August 2019
Abstract: This study focused on the statistical technique using the neural network, hybrid models and factor analysis on constructing the new factors affecting students learning styles of the survey done among university students in predicting academic performance. The data were collected using survey questionnaires and students’ academic records. The methodologies used were descriptive statistics, factor analysis, neural network and hybrid models technique using the following Learning algorithms; Levenberg-Marquardt (LM), Bayesian Regularization (BR), BFGS Quasi-Newton (BFG), Scaled Conjugate Gradient (SCG), Gradient Descent (GD) in artificial neural network model while for the second Hybrid model only the best two algorithms where use; Levenberg-Marquardt (LM), Bayesian Regularization (BR). The results showed ten new factors were successfully constructed using factor analysis and the proposed hybrid models show that though it took longer time and number of epochs to train the hybrid models by Bayesian Regularization Algorithms, and it gives more accurate predictions than both the Levenberg-Marquadrt, Scaled Conjugate Gradient, Gradient Descent and BFGS Quasi-Newton (BFG) Algorithms. In a nutshell, the finding indicates that Bayesian Regularization is the best learning algorithms in both Neural Network and Hybrid models for predicting students’ academic performance.
Abstract: This study focused on the statistical technique using the neural network, hybrid models and factor analysis on constructing the new factors affecting students learning styles of the survey done among university students in predicting academic performance. The data were collected using survey questionnaires and students’ academic records. The method...
Show More
Predictive Modeling of the Brand Equity: Analysis Based on Multiple Logistic Regression and Backward Stepwise Model Selection Methods
Gehan Shanmuganathan Dhameeth
Issue:
Volume 5, Issue 4, August 2019
Pages:
67-72
Received:
31 July 2019
Accepted:
26 August 2019
Published:
10 September 2019
Abstract: Brands play a significant role at the point of consumer purchases decisions. Brand managers make all the efforts to induce consumers to purchase their brands and increase eventual brand associations for long-term profits. This paper focuses on how different generations, especially the Millennial and the Baby boomers, behave towards brands based on organizations’ brand building efforts to create Brand Equity (BE) using a predictive model. Prior research has not been successful to provide a detailed understanding of generations and their potential brand behavior in a predictive perspective. In this article, author used a predictive model of the brand behavior of different generations using a Multiple Logistic Regression (MLR) method. In addition, it is determined how the predictor variables (awareness, recall, relate, purchase, knowledge, trials, association, recommendations, salience, imagery, performance, feelings, judgement, and resonance) influence the response variable, brand equity, to predict brand equity in these two audiences. In this study, the author administered an online survey using Survey Monkey to reach local (US) and international college/university respondents (n=267) age 18 years and above. The survey was administered using a questionnaire (46 data points). In the analysis process, the author developed a Multiple Logistic Regression (MLR) model, tested the model error, predicted the brand equity of generations, and determined the best model with parsimonious number of predictor variables using the Backward Stepwise Method (AIC). The analysis suggested the model to be reliable model with a 100% prediction of the brand equity (BE) with a mean value of 1. Given the predictors, the model correctly predicted 63% respondents, millennial and baby boomers, to be associated with brand equity and 35% respondents to be otherwise, while the Best Model based on the Backward Stepwise Selection Method (BSSM) using Step AIC function, suggested thirteen out of fourteen predetermined predictors included in the model to predict Brand Equity (BE). In the results generated, the AIC value indicated was 106.
Abstract: Brands play a significant role at the point of consumer purchases decisions. Brand managers make all the efforts to induce consumers to purchase their brands and increase eventual brand associations for long-term profits. This paper focuses on how different generations, especially the Millennial and the Baby boomers, behave towards brands based on ...
Show More
