International Journal of Intelligent Engineering and Systems

1

Authors: Vanaja Roselin Emmanvelraj Chirchi, Laxman Madhavrao Waghmare

In this paper, we propose a novel method for iris recognition system using enhanced isocentric segmentor (EISOS) and wavelet rectangular coder (WRC). At first, we locate the center of the eye within the area of the pupil on low resolution images using EISOS method. Once the iris region is successfully segmented, the next stage is to transform the iris region into the fixed dimensions. Then, we propose a novel feature vector generation method namely, wavelet rectangular coder (WRC). Finally, we recognize the iris image using fuzzy logic classifier to identify whether the iris image is present in the dataset or not. The iris recognition performance is measured using different dataset such as, CASIA, MMU and UBIRIS dataset. Experimental results indicate that the proposed method of EISOS+WRC based iris segmentation and recognition framework have outperformed by having better accuracy of 99.75% which is 94% and 93% for using existing approaches.

20

Authors: Loukal Keltoum, Benalia Leila, Bouguerra Abderrahmen

The mean of this paper is fuzzy sliding mode control of a Doubly Fed Induction Motor (DFIM); it’s the coupling of the fuzzy logic control and sliding mode control (SMC). The use of the sliding mode method provides very acceptable performance for DFIM control, and the chattering phenomenon effect is also eliminated by the fuzzy logic mode. In the first part, we carried out briefly a study of modeling on the full system. This model is intended to facilitate the procedures for setting and controlling the speed. We introduced the parameter variation to test the robustness of the control laws. The results of our simulations are conducted to validate the theory and indicate that the control performance of the DFIM is satisfactory and the proposed fuzzy sliding mode control (FSMC) can achieve favorable tracking performance.

38

Authors: Rama RaoAdimalla, Vatsavayi Valli Kumari, Ch Satyananda Reddy

Mobile ad-hoc network face a serious issue of frequent change of network topology which leads to the stability and reliability problems of routing. Many routing schemes such as multi-path routing and backup path routing are developed to increase the link reliability in the literature. Accordingly, we propose to design a backup route establishment for QoS routing protocol. In this protocol, the potential failures of network and node are identified and a backup routing is initiated. For detection of failures, a path evaluation function is determined based on the metrics energy drain rate and interference, congestion status are measured. The primary path satisfying the QoS parameters node’s static resource capacity, dynamic resource availability, neighborhood quality, link quality is established. Once failure is detected, back up routes are established and transmission is redirected on these back up routes. Simulation results show that the proposed protocol has reduced recovery delay and improved throughput as compared with the existing protocol. The improvement of 68% in terms of delay, 30% in terms of delivery ratio and 83% in terms of throughput are achieved for the proposed routing protocol.

58

Authors: Banoth Rajkumar, Gugulothu Narsimha

This paper proposes an incorporation mechanism of primitives to provide complete cryptography services for Mobile Ad-Hoc Networks (MANETs). The proposed approach is the Secure Light Weight Encryption Protocol for MANET, where an algorithm for providing availability with DoS resilience has been implemented to avoid TCP SYN flooding packets in the network and passes other packets and also an authentication code and hash function is generated. Later the intermediate nodes are allowed to recode the encoded data in the network. Through which the confidentiality and non-repudiation, key management, integration is achieved.

75

Authors: Thangavel Muthamilselvan, Balamurugan Balusamy

The growth of IT industry technology is absorbed by the cloud service technology, which leads to secure connectivity and availability of services to cloud users. This paper proposes an e-health data classification system that gives the services to e-health cloud user for their disease prediction requirements. The tree bagging method is suitable to select better weighted attributes and the careful seeding of K-means ++ algorithm improves the accuracy and speed of clustering. Most of the tree classification methods use only information gain as the strategy to select suitable attributes for classification. We used information gain in bagging technique to improve accuracy. In this research article, we proposed a method to Coalescing Decision tree classification algorithm, bagging technique, and K-means++ algorithm to build a better classifier for the e-health cloud users. It was evaluated with the standard data sets such as Diabetes, breast cancer, liver disorders and cardiotocography.

94

Authors: M. B. Jayalekshmi, S. H. Krishnaveni

Cloud computing is a paradigm which offers a range of services to the cloud users. Cloud storage is the most popular service, as the data owners are freed up from the data management and storage overhead. However, the data owners are concerned about the security of the data. In order to address this issue, this article presents a twin layered security scheme, which guarantees the security of the data and thus the data integrity is preserved. The proposed work involves two building blocks, which are hashing and encryption. The input data is separated into several blocks, followed by which a tree is constructed. This is followed by the computation of hash codes, which is done by the keccak-256 hash function. The hash codes are then encrypted by Advanced Encryption Standard (AES) (256) with a dynamic key being generated by CodeIgniter. Thus, the security of the system is tightened and misuse of the data is strictly denied. The performance of the proposed work is tested with respect to throughput, memory and time consumption. The experimental results are reasonable, when compared to the analogous algorithms.

117

Authors: Vyacheslav Shulunov, Irina Esheeva

The article considering the problem of quick transformation of solids of revolution into raster layer for Roll Powder Sintering (RPS) additive manufacturing technology. The author describes the algorithm for solving this problem. This work is devoted to the development and the validation of the impetuous algorithm for converting the 3D coordinates of a solid of revolution to spiral coordinates. It is based on the assumption that one point on the figure of revolution is predetermined multitude of points in the spiral coordinate system and these can be defined without calculations because of symmetry properties. Such method provides to accelerate the performance of the coordinate conversion thousand fold and real time forming objects 1.5 m3 in volume with layer thickness ~25 µm by non-powerful computers per hour. The proposed algorithm has been extensively tested and proven by numerical examples of several parts of computer aided design models. It is faster than other coordinates transformations and can be extended to the acceleration converting and processing geometry of 3D objects.

136

Authors: Pooja Choudhary, Jaisankar Natarajan

It is always an effective option for storing the data on cloud as it has advantages such as processing huge volume of data on user’s request. But policy updating is one of challenging issue when attribute based encryption method is used for constructing schemes such as access control. Many techniques have been proposed for achieving secure data access control in any cloud storage system, but policy updating is always a problem. In this paper, a novel scheme is proposed to enable to access control by using method known as dynamic policy updating for the data in cloud. The focus of this paper is to develop dynamic policy updating method for the data which is stored in the cloud. This method will help the data owners to update the policy for the data which is stored in the cloud. The features of this method are that data owner can update the policy on his data and data will be encrypted according to the new policy and user will be able to download only if he has new policy which is not possible in existing methods. The drawback of traditional method was removed such as user revocation and method is made more efficient access control scheme policy update based on attribute. The results show that policy can be updated for the data stored in the cloud and revoked user cannot download the file. Only the data owner can update the policy for his data in the cloud other than data owner cannot do this.

155

Authors: Jegan Govindasamy, Samundiswary Punniakodi

Nowadays, ZigBee is one of the dominating standards for wireless sensor networks and Internet of Things (IoT) networks. Even though, the ZigBee standard is formed with low per-unit costs, security in mind and network resilience, existing security mechanisms are not effective to provide security and protection against wormhole attacks and Distributed Denial of Service ( DDoS) attacks on networks such as WSN and IoT. They also introduce high consumption of energy, storage memory and processing. In this work, Energy Efficient Intrusion Detection System (EE-IDS) and Energy Efficient Intrusion Detection System with Energy Prediction (EE-IDSEP) are proposed for protection of ZigBee based wireless sensor networks in presence of wormhole attacks and ( DDoS) attacks. The EE-IDS is developed and its performance is evaluated by considering three different routing protocols such as Ad hoc On-Demand Distance Vector (AODV), Shortcut Tree Routing (STR) and Opportunistic Shortcut Tree Routing (OSTR) to improve the security against wormhole attack and to mitigate the energy consumption of the sensor nodes in the ZigBee based wireless sensor networks. The proposed EE-IDS and EE-IDSEP are evaluated through extensive simulations by using NS2 and then compared with the existing Energy Efficient Trust System for Wormhole detection (EE-TSW) and Energy Efficient Trust System (EE-TS) for detection of DDoS attack. It is inferred from the simulation results that proposed IDS namely EE-IDS-AODV, EE-IDS-STR and EE-IDS-OSTR for detection of wormhole attack have better performance than that of existing EE-TSW and proposed system EE-IDSEP for detection of DDoS attack have also shown better performance than that of existing system EE-TS in terms of performance metrics such as Packet Delivery Ratio (PDR), Average End-to-End Delay, energy consumption, detection rate, average detection time and False Positive Rate (FPR).

173

Authors: Malay Bhatt, Tejas Patalia

A large number of folk dance videos are uploaded on the web or added as a situational song in the Bollywood movies. The classification of folk dance videos is essential for dance education, to preserve cultural heritage, and for music companies to provide better customer oriented service. India is a country having many regional languages and each region has its own popular folk dances. Four different Indian folk dances namely, 'Garba', 'Lavani', 'Ghoomar' and 'Bhangra' are considered. A Folk Dance Classification Framework is proposed which extracts audio signal from video, takes a fragment of 125 seconds from the beginning and further separates it into a set of small segments, calculates Mel-frequency Cepstral Coefficients (MFCC) and Linear Predictive Coding (LPC) coefficients, generates high dimensional feature vector, reduces dimensionality using Principle Component Analysis (PCA) and classifies segments using Scale Conjugate Gradient Neural Network. The performances of chosen classifiers, K-Nearest Neighbor, Naïve Bayes and Neural Networks, are compared. Class labels of all segments are clubbed together and based on majority voting class label is assigned to a folk dance song. System achieves more than 90% accuracy.

192

Authors: Ghetiya Rutvi Babulal, Gayathri Prakasam

Proficient Algorithms for Mining High Utility Itemset in One Phase utilizing Direct Discovery of High Utility Patterns refers to the disclosure of object/item sets with high utility. The situation may end up being more terrible when the database contains long high utility itemsets.The information of high utility itemsets is kept up in a tree-based data structure UP-Tree to such a degree, to the point that required itemsets can be created productivity with only two yields of database. The execution of existing algorithm issue all utilize a two-stage, applicant era approach with an exemption which is however neither efficient nor versatile in case of huge databases. This two-stage approach experiences adaptability issue because of the enormous number of matching itemsets. This paper proposes a novel calculation that finds high utility examples in a solitary stage without creating candidates. Our straight information structure empowers us to process a tight technique for pruning and to specifically distinguish high utility examples in an efficient and versatile way, which focuses on the main driver with earlier technique. Test comes to conclusion that the proposed computations, especially Direct Discovery of High Utility Pattern reduce the amount of hopefuls satisfactorily and additionally beat diverse figuring extensively to the extent runtime, especially when databases contain clusters of long trades.

210

Authors: Thendral Puyalnithi, Madhuviswanatham Vankadara

Most of the existing Clinical Decision Support Systems(CDSS) for predicting Coronary Heart Disease(CHD) risks rely solely on the analysis of single type of diagnostic data or a certain type of classification algorithm, thus making the predicted component unreliable. Since the Risk value very much influences the further course of the treatment, the correctness of the risk value is eminent. So the novelty of the proposed method lies in considering many classification algorithms and many diagnostics data sets to predict the risk value. Moreover in the proposed method we have given flowchart for finding the risk value, which is a novelty. The results shows that the selection of classification algorithms cannot be universal as proposed by many recent researches and each type of data set will be having a certain type of classifier which works best for the data set. The proposed system has four classifiers namely Mixed cardio data classifier, Single Photon Emission Computed Tomography(SPECT) data classifier, Electrocardiogram(ECG) data classifier and 1-year survival Electrocardiogram data classifier, each trained with benchmark datasets of University of California, Irvine(UCI) Machine learning repository. The performance of algorithms such as Iterative Dichotomiser3(ID3), Naïve Bayes(NB), Support Vector Machine(SVM), k-Nearest Neighbors(kNN) and Ensemble classifiers such as Bagging and Boosting are evaluated for every data set and best classifier is chosen by its accuracy measure. Ensemble method is the most efficient for 1-year survival ECG data classifier with 89.5% accuracy, ID3 is the most efficient algorithm for Mixed cardiac data classifier with 96.9% accuracy and SVM is the most efficient algorithm for SPECT data classifier and ECG data Classifier with accuracy of 98.3% and 100% respectively. The proposed work also finds the optimal number of samples needed for Bagging and Boosting type of ensemble classifiers.

226

Authors: Leo Dencelin Xavier, Ramkumar Thirunavukarasu

Knowledge of a protein’s secondary structure, in turn, contributes to our understanding of the functions of the protein is vital to many aspects of living organisms such as those of enzymes, hormones, and structural material, etc. It also helps in designing new drugs for critical disease. In this paper, we have advocated a distributed approach to identify the Protein Secondary Structures using an ensemble method on protein primary sequences. The Ensemble based Random Forest algorithm has been adopted to build the three-way predictive model. Based on the amino acid features of each protein and decision tree parameters, the classification model allows us to assign protein structures as ‘α helix’, ‘β sheet’, or a coil. Also the proposed model is implemented in a distributed computing environment, SPARK. Experiments have been carried out using cross-validation tests on RS126 and CB513 benchmark datasets. Our results clearly confirm that ensemble approach in classifying protein secondary structures scores better accuracy with improved performance when it will be implemented in the distributed environment.

245

Authors: Alexis Cordovés García, Ricardo Sors Lancho, Arlys M. Lastre Aleaga, Alexis Cordovés Rodríguez, Leonardo L. Lorente Leyva

The goal of this paper is the adaptation of plans of manufacturing processes of Axisymmetric Solid Parts under the approach of Case Based Reasoning (CBR). To this end, a data model for feature-based representation has been developed, which contains the geometric, dimensional and technological information required for the preparation of the one-piece process plan. As a result, an algorithm to generate new parts process plans from a similar plan retrieved from the case base is presented, which is based on the specific information of the Primitive Form Features (PFF) and considers the possible interaction between neighboring PFFs. The computer support of the developed algorithm was implemented in AutoLISP as programming language of the CAD system used and the interface was created in OpenDCL.

257

Authors: Myungsuk Lee, Eunsook Bae

The purpose of this study is to explore a new teaching and learning method using MOOC as a way of solving the change of the educational environment and the learner 's unsatisfied academic desire. This study suggests 'Hybrid teaching method' as a way to utilize MOOC in liberal arts education. This teaching method is a method of perfect learning through formal learning and informal learning. Formal learning consisted of a teacher - centered knowledge acquisition stage and a learner - centered thinking extension stage. Informal learning consists of learner - led experience/reflection stages. It is a characteristic of 'Hybrid teaching method' that emphasizes the role of instructor, breaks the boundaries of education space through informal learning, and implements perfect learning through level learning. This study is meaningful that the MOOC has provided a detailed plan to utilize it in the curriculum of the university. We hope that the ‘Hybrid teaching method’ using MOOCs will be a way to adapt and lead the liberal education to the future education trends as a customized learning according to the development of science and technology, destruction of spatial areas such as online and offline, emphasis on connection between theory and reality through formal learning and informal learning.

272

Authors: Fatiha Bessaha, Nouria Mahrez, Souhila Bendenia, Fatima Kasmi, Kheira Marouf-Khelifa, Amine Khelifa

Algerian halloysite was heated at 600 °C and treated with HCl 5N. The materials were characterized by chemical analysis, TEM, and FT-IR spectroscopy, and used in the elimination of Congo red (CR) from aqueous solution. The effects of contact time and temperature were investigated. The thermal treatment at 600 °C results in the formation of dehydroxylated structure. Acid attack involves an increase in SiO2 content, due to the leaching of Al ions from octahedral sheet. Thermo-chemical treatment also diminishes the percentage of impurities and maintains the tubular morphology. Kinetic data follow the pseudo-second order model, whilst thermodynamic parameters lead to a not spontaneous and endothermic process. Significant changes occur in the vibrational spectrum of H600-5N (halloysite treated at 600 °C and with HCl 5N), after adsorption of Congo red, with the involvement of amino and sulfoxide groups. The mechanism highlights an outer-sphere surface complexation of SiOH....H2O species, i.e., SiOH linked to H2O via H-bonds.

290

Authors: Asha Jerlin, Jayakumar Chinnappan

Detection of malware has become more challenging today because of the advancements and technologies adapted to corrupt the network or the devices. Static, dynamic and hybrid malware detection analysis methods have failed to provide complete malware detection. Hence in this work, a bio inspired sequence alignment method used in bioinformatics to compare the similarity between amino acid sequences in protein structures has been adapted to give the best similarity score to detect malwares. The state of art sequence alignment methods like Smith Water Man Algorithm (SWMA) used in bio informatics suffers from the problem of more memory utilization and computation time which is in the order of n2 ie., (O(n2)) and hence in this work an efficient sequence alignment algorithm (ESSA) has been proposed to address the problem of memory utilization thereby making the memory utilization and computation time to the order of n ie., (O(n)) there by making the detection rate higher. It is also clear from the results that the similarity score is high when the sequence length is small. The accuracy of the prediction rate of malware and benign increases.

307

Authors: Gunasekaran Ganesan, Venkatesan MeenakshiSundram

3D visualization is the recent research need to be able to view the model from various angles so as to see data that is obscured by other data. While visualizing the data, view point selection is one of the major challenge which define the viewpoint information to select best viewpoints. This paper introduces teaching learning opacity based optimization (TLOO) and Kalman filter for selecting optimal viewpoint. The structural features of 3D volume dataset such as opacity and luminance is extracted for selecting optimal viewpoint position from that structural features. TLOO optimization algorithm identifies the voxels and standard information using view point evolution function. Then, kalman filter steers the optimization process with the help of cost function. The proposed flow of the technique is simple and very effective for selecting optimal viewpoints in 3D images. Computational complexity and the quality metrics are analyzed that provided maximum output efficiency.

327

Authors: Nagaraja Gundluru, Pradeep Reddy Ch

We can realize instant joint group communication by forming Mobile Ad Hoc Networks without demanding any pre-plan or pre-existing infrastructure setup. Conversely, the curbs of these networks such as unreliable wireless medium, unpredictable topology, no central administration, fuel the compulsion of a key based cryptographic algorithm to defend data traffic. In this perspective, substantial research work has been done in the last decade or so and ascertained that the trust based frameworks for group key management deliver superior performance than others. Since the nodes in ad hoc networks have limited computing resources, the overall performance of the system depends on how effectively and securely designed the system. This encourages us to work on a framework which consumes less computing power and also invulnerable to internal as well as external attacks. We propose a framework which reduces the network resource consumption for the trust request and collection using game theory concept. The energy of the wireless nodes are significantly saved by choosing the novel strategy called as finding optimal set of remote nodes to send the response for trust request using game theory. Choosing local optimal at each stage eventually leads to global optimal. Later, synthesize the collected trust and handle the attacks using fuzzy concept in order to get the degree of trustworthiness instead of binary classification. We prove with our simulation results that our proposed scheme reduces overhead of the network significantly while without compromising security aspects.

347

Authors: Balamanigandan Ramachandran, Kamalraj Subramaniam

During the last decade internet users, researchers and industries have focused on the cloud based technologies. Cloud technology is a distributed and scalable service model and in this model service providers offer the services depending on the cloud users. Security in cloud is a challenge for the service providers , data owners and the service requesters because the cloud providers provide various services such as storage, computing , database and network . All the services focus on data processing in a network to share the information from cloud to users, data owner to cloud and cloud to authenticator. To solve these security issues, in this paper we define a new approach for data sharing issues and service management of Trusted Third Party Authentication ( TPA ) based on multilevel security framework and bilinear pairing techniques. This framework consists of three modules and these modules are service module, transaction processing module and security module . The next thing is bilinear pairing technique, which uses the pairing techniques to achieve the secure cloud service and improve the computing utility power , store large amount of data and execute the efficient encryption process securely in cloud environment. In this work, the cloud ssecurity requirement has been enhanced using bilinear mapping and multilevel security framework . The result of multilevel security framework is comparatively better than the present day cloud security.

363

Authors: Kumaran Umapathy, Neelu Khare

Preservation of privacy is a significant aspect of data mining and as the secrecy of sensitive information must be maintained while sharing the data among different untrusted parties. There are many application is suffering from vulnerable, data leakage, data misuse, and sensitive data disclosure issues. To protect the privacy of sensitive data without losing the usability of data, various techniques have been used in privacy-preserving data mining (PPDM). Some of the approaches are available to maintain the tight privacy, but they fail to minimize the execution time and error rate. The main objective of the article is to contribute and retrieve the data with minimal classification error and execution time with enhanced privacy. To overcome the issues, the paper introduces the Secure Data Contribution Retrieval algorithm (SDCRA) to fulfill the current issues. Proposed algorithms define a privacy policy and arrange the security based on requirements. This design applies the privacy based on the compatibility of applications. This approach is capable of satisfying the accuracy constraints for multiple datasets. It also considers the efficient data extraction with a good ranking of attributes in tables. Here, proposed SDCRA is compared with existing approaches namely as Perturbation, singular value decomposition (SVD), Singular Value Decomposition data Perturbation (SVD+DP), K-anonymity with Decision Tree (KA+DT)[] for Cancer, HIV, Diabetes dataset. Based on experimental result proposed approach performs well regarding success rate, error rate and system execution time compare than existing methods. Proposed approach improves Success Rate 1.83% reduces the Error Rate 2.33% and minimizes the system execution time 2 seconds.

381

Authors: Shachi Mall, Umesh Chandra Jaiswal

This paper developed innovative algorithms such as shallow parsing and modified Lesk’s algorithm to resolve the issues in Word Sense Disambiguation and performed correct translation from Hindi language to English language. Shallow parsing method is based on Hidden Markov model. We also perform an evaluation for 1657 Hindi tokens with 990 phrases for Parts of speech tagging and Chunking for given Hindi sentence as input and able to achieve Precision, Recall, F-score, Accuracy for Parts of speech tagger: Accuracy: 92.09%; precision: 84.76%; recall: 89.29%; F-score: 86.97, system accuracy for Chunk: Accuracy: 93.96%; precision: 89.33%; recall: 91.31%; F-score: 90.315%. The evaluation is performed by developing confusion matrix in which the system result of Parts of speech tagger and Chunk is compared with Gold standard date provided by IIIT Hyderabad in the summer school 2015. In this paper we discuss the second problem Word Sense Disambiguation in which we enhance the Modified Lesk algorithms by using overlap based method which will find information between three pieces of words in a given context. The system generated result resolves the issues of Word Sense Disambiguation and shows the comparison result with the website Google Translator in which we input polysemy word in a Hindi sentence and same sentence input in our generated system and shows the comparison result of both. The output result shows that our system resolves Word Sense Disambiguation and produces correct translation and Google Translator is fails to resolves the correct Translation.

401

Authors: Gunavathi Chellamuthu, Premalatha Kandasamy, Sivasubramanian Kanagaraj

Microarray technology is commonly used in the study of disease diagnosis using gene expression levels. The classification of cancer is a foremost area of research in the field of bioinformatics. Microarray technology enables the researcher to investigate the expression levels of thousands of genes in a single experiment and gives the measurements of their differential expression. The main drawback of gene expression data is that it contains thousands of genes and a very few samples. Feature or gene selection methods are used to successfully extract the genes that directly involved in the classification and to eliminate irrelevant genes. These methods considerably improve the classification accuracy. The proposed method applies bat algorithm (BA) for feature selection in tumour classification. Initially, the top-10 genes are selected by T-Statistics, signal-to-noise ratio (SNR) and F-Test. The classifier accuracy of k-nearest neighbour (kNN) technique is used as the fitness function for BA. The simulated results are demonstrated and analyzed with 10 different cancer gene expression dataset. For Lung Cancer Michigan and Lung Harvard2 datasets the proposed method gives 100% classification accuracy with minimum number of genes. For 5 other datasets, the proposed method gives more than 90% of classification accuracy. The results show the suitability of the proposed algorithm for feature selection in cancer classification.

418

Authors: Haritha Akkineni, Venkata Swarajya Lakshmi Papineni, Vijaya Babu Burra

Ubiquitous online communication is producing massive amounts of data on an un-precedential scale. Many have come up in analyzing the data produced from Twitter. Sentiment analysis is being carried out to catch the pulse of the people. As the data is mostly in the unstructured format. Getting the summary of what has been expressed positively and what has been expressed negatively plays a major role. Summaries take a wide strand in this plait as the relevant content can't be comprehended all at once. Working on the premise that online social media conversations might represent a new source of information to monitor the status of the policies launched by the government, this investigate stands as a first paw which may ultimately be the source for producing abstractive summaries from Twitter. This paper mainly focuses on developing a hybrid method which takes the combination of extractive summaries using statistical approaches as well as abstractive summaries using Rich Semantic Graph based approaches. The set of classified positive and the negative tweets will be passed as input to the hybrid method and it generates concise as well as readable summaries from both set of tweets. These summaries can be easily viewable on the PDA's and enhances the decision making capability of the policymaker.

435

Authors: Dhivya Elavarasan, Durai Raj Vincent

Web mining is an application of data mining techniques to extract and process knowledge from sources such as web documents, hyperlinks, website usage logs etc. Due to the outbreak of extensive information through web resources, mining of semantically relevant data for a search keyword is one of the most intriguing studies to research. This work deals with automatically extracting the information from web documents using web content mining. The extracted data needs to be preprocessed in order to obtain the appropriate data format for further analysis. Generally, the content based search algorithm is used to find the items relevant to the keyword searched resulting in an indexed set of similar results. Further, the clustering of the similar data is done by adopting the quality threshold clustering algorithm assigning a similarity index to each of the result items. For the final list of items obtained, the weighted page ranking algorithm is applied to rank the most frequently searched item in the lists. The proposed work efficiency will be determined by the cluster's quality and the query blocks ranking efficiency. Various metrics like cluster purity, NMI, Rand Index, F-Measure, wPRF are used to evaluate the quality and the ranking efficiency of the search result obtained. Duplicate result sets are handled and are castigated for better unambiguous results. The results obtained is proved to be better and to overrun the existing approaches like QFI and QFJ. The quality of the result set obtained is further evaluated by repeating the process considering only the top n ranked items, shuffling the top items or by randomly selecting the items. Thus, enabling to validate and uphold the results of the proposed work surpassing the existing algorithms.