فهرست مطالب

International Journal Information and Communication Technology Research
Volume:5 Issue: 3, Summer 2103

  • تاریخ انتشار: 1392/05/08
  • تعداد عناوین: 7
|
  • Mohammad Reza Ahmadi Pages 1-10
    A dynamic resource allocation method for virtual resources in virtualized data centers has been proposed in this paper. Since resource allocation with constraint in virtual environment is NP-hard, the solution has been focused on approximate methods based on immune system mechanism using agent based model and Cooperative Coevolutionary Algorithm (CoCo-VM) that appears to perform well in finding a plausible answer [1]. The novelty of our approach lies in integrating an agent based greedy algorithm based on immune system functionality together with the cooperative co-evolutionary concept as an intelligent solution for virtual resource allocation in a large scale and distributed virtualized datacenters. Here, some mathematical analyses have been done to identify the parameters essential to assign a suggested allocation approach. Results of different evaluations in pure immune system (PIS)0F 1 and CoCo-VM methods demonstrate that, for the scenarios under consideration, the proposed resource allocation approach can significantly reduce resource consumption, increase number of successful services, and achieve higher performance.
    Keywords: Virtual resource allocation, cooperative co, evolutionary algorithm, immune system, agent, based greedy algorithm
  • Ehsan Arianyan, Ahmad Motamedi, Ahmad Motamedi, Mohammad Motamedi Pages 11-17
    Optical Character Recognition (OCR) is a technique by the help of which the optical characters are identified automatically by a computer. There are many methods for OCR, one of which is neural network that we use. Unfortunately, the long training and testing time of these networks is disturbing, but we healed this problem by mapping our network on graphics card by using Jacket which is the product of Accelereyes group. By so doing, we achieved the speedup of up to twelve factors. Graphics Processing Units (GPUs) have parallel structure containing many cores capable of running thousands of threads in parallel. We train a multi-layer perceptron network using back propagation rule which has a degree of parallelism that is suitable for implementation on new graphics card. We examine the Persian characters that are typed on the new system of Farsi license plates to make a database of characters uses in this system and apply them as train and test data for our network.
    Keywords: Neural network, OCR, GPU, Jacket, CUDA
  • Masoomeh Mashayekhi, Morteza Analoui Pages 19-24
    Bilingual corpus is one of the most important resources for Natural Language Processing applications and researches. The quality of bilingual corpora can influence the result of researches that used it as a resource. When translation machine is used to verify corpus quality, the quality of translation machine can affect the evaluation of corpus. One way for evaluating software or resources in ISO is verifying its own features. The expectation of finding translation for each word in each sentence by using a bilingual dictionary is verified in this paper as a factor for evaluating fidelity of corpus. Computing this expectation needed a pre-processing step that is designed with considering the differences between English and Persian languages. This method is a combination of a rule-based method with the information of a dictionary.
    Keywords: Evaluation, Bilingual Corpora, Aligned Corpora, Parallel Corpora, Bilingual dictionary
  • Mojtaba Salehi, Mohammad Bagher, Isa Nakhai Kamalabadi Pages 25-33
    Recommender system is a promising technology in online learning environments to present personalized offers for supporting activity of users. According to difficulty of locating appropriate learning materials to learners, this paper proposes an adaptive hybrid recommender framework that considers dynamic interests of learners and multi-attribute of materials in the unified model. Since learners express their preference based on some specific attributes of materials, learner preference matrix (LPM) is introduced that can model the interest of learners based on attributes of materials using historical rating of accessed materials by learners. Then, the approach uses collaborative filtering and content based filtering to generate hybrid recommendation. In addition, a new adaptive strategy is used to model dynamic preference of learners. The experiments show that our proposed method outperforms the previous algorithms on precision, recall and intra-list similarity measure and also can alleviate the sparsity problem.
    Keywords: Personalized Recommendation, Collaborative Filtering, Learning Material, E learning, Adaptive Recommender, Dynamic Interests
  • Ali Nazemian, Fattaneh Taghiyareh Pages 35-44
    Model of spreading behaviors, influences, new trends and innovations through social networks has been studied in a number of domains. These may include the diffusion of medical and technological innovations, the sudden and widespread adoption of strategies in game-theoretic settings, and the effects of word of mouth in the promotion of new products. One of the most important facts that is neglected in previous spread models is “considering cascading negative opinions”. This important fact shows that negative opinions may originate and propagate in populations as much as positive opinions and even they are stronger and more dominant. In this paper we propose a new model of influence cascade called Independent Cascade with Positive and Negative WOM (ICPN). ICPN models some important facts that people may encounter in a social environment. These facts include negativity bias, the asymmetric behavior of negative and positive WOM, and different types of consumer complaints behaviors. Moreover, the influence maximization problem is formulated in this model and also, we show that ICPN maintains submodularity in this problem. This fact allows a simple greedy approximation algorithm for maximizing the positive influence within a ratio of 􁈺􀫚 􀵆 􀫚 􀢋􁈻 approximation.
    Keywords: component, independent cascade model, influence maximization, customer behavior propagation, negative word of mouth, positive word of mouth, consumer complaining behavior
  • Mojgan Farhoodi, Alireza Yari, Saeed Shiry Ghidary Pages 45-54
    In traditional search engines, the most common way to show results for a query is to list documents in order of their computed relevance to the query. However, the ranking is independent of the topic of the document;so the results of different topics are not grouped together. In this situation, the user must scroll though many irrelevant results until his desired information need is found. One solution is to organize search results via classification. Many researchers have shown that classifying web pages can improve a search engine's ranking of results. Intuitively results should be more relevant when they match the class of a query. In this paper, we present a simple framework for classification-enhanced ranking that uses query class in combination with the classification of web pages to derive a class distribution for the query. In this regard, we propose a hybrid IR search strategy that begins with a 3-gram classification-based strategy and reverts to a ranked-list strategy if the user doesn’t find the target document in selected class.The experiment results on Hamshahri corpus show satisfactory results.
    Keywords: Information Retrieval, Hamshahri, Ranking, Classification, SVM, KNN, N, gram language modeling, Smoothing methods
  • Atefeh Tajpour, Suhaimi Ibrahim Pages 55-62
    SQLIA is a hacking technique by which the attacker adds Structured Query Language code (SQL statements) through a web application's input fields or hidden parameters to access the resources. By SQL injection an attacker gains access to underlying web application's database and destroys functionality and/or confidentiality. Researchers have proposed different techniques to detect and prevent this vulnerability. In this paper we present SQL injection attack types and also current security tools which detect or prevent this attack and compare them with each other. Finally, we propose a framework for evaluating SQL injection detection or prevention tools in common criteria. In fact, this paper provides information about current tools for researchers and also helps security officers to choose suitable SQL injection detection tools for their web application security.
    Keywords: web application security, web application vulnerability, SQL Injection attack, framework, tool, evaluation, comparison