Implementation of Information Retrieval (IR) in an Electronic Commerce Architecture using Back propagation Network Learning Algorithm
Dr. Riktesh Srivastava
Associate Professor, Information Systems Skyline University College, Sharjah, UAE
The present study assesses the technique of back propagation neural networks to appraise the required Information Retrieval (IR) and thereby decreasing the average response time of an Electronic Commerce architecture. In order to delineate the response time, diverse array of user requests were engaged per unit time. Furthermore, engagement of Back Propagation Network Learning (BPNL) algorithm is used to train ...view middle of the document...
The study is an attempt to conduct the experiment of Information Retrieval (IR) using Artificial Neural Networks by implementing Back Propagation Neural Network. Artificial neural networks are biologically stimulated classification algorithms that entails of an input layer of nodes, one or more hidden layers and an output layer. Each node in a layer has one corresponding node in the next layer, thus spawning the stacking effect . Back propagation Network Learning Algorithm (BPNL) is one of the prevalent structures amid artificial neural networks which is extensively used to elucidate complex problems by modeling complex input-output relationships .
The preliminary BPNL algorithm was suggested by Rumelhart et al  and since then became prominent learning algorithms for ANN. BPNL uses gradient-decent search procedure to alter the connection weights. The structure of a BPNL algorithm is revealed in Figure 2. The output of each neuron is the accumulation of the numbers of neurons of the previous level multiplied by its corresponding weights. The input values are converted into output signals with the calculations of activation functions . BPNL algorithm have been extensively and efficaciously functional in varied applications, such as pattern recognition, location selection and performance evaluations.
Press , Yao  remarks on the diffusion of electronic commerce architecture with ANN for operative formation of the systems. ANN is the choice for such a diffusion as it does not necessitate any expectations about the distribution of data. Hecht-Nielsen  premeditated the mathematical analysis of such a diffusion. Research also exhibited that flexibility and generalization are two most commanding facets of ANN modeling involving BPNL. Sarle , Wieland & Leighton  directed that if ANN models are instigated appropriately in an Electronic Commerce architecture, they are proficient of modeling complex patterns in data, and they can be pooled with other models to further mend the performance.
Concerning such a diffusion of ANN and Electronic Commerce architecture, experimentation
conducted in the study trains the user requests from the web server through 7 iterations to expound the
relative stochastic exploration of the Electronic Commerce architecture. BPNL algorithm was developed
using Java programming language and employs both feed forward and back propagation approaches to
amend the weights accordingly.
Complete paper is alienated into 5 sections. Section 2 interprets the Electronic Commerce architecture used for the study. Section 3 references the mathematical valuations for the BPNL algorithm. Section 4 mentions the result investigation and depicts the actual outcomes of the experiment conducted. Section 5 explicates the supposition and impending work to be steered. 2. Electronic Commerce architecture
The proposed Electronic Commerce architecture is an extension to the system of Client Server Computing. In the architecture,...