Search published articles


Showing 2 results for Memarian

Parvaneh Memarian, Farhad Sasani,
Volume 10, Issue 6 (Vol. 10, No. 6 (Tome 54), (Articles in Persian) 2019)
Abstract

 
It is commonly thought that when a work goes through retranslation process, the latest ones are expected to contribute to a better understanding of the text. However, it is not always the case, at least in Iran.  The book market of Iran is replete with retranslated versions of classic works most of which are not genuine translation but plagiarized version of previous translations. One famous example is George Orwell’s Animal Farm which has been translated into Persian more than 70 times by different translators since 1348. Regarding this issue, the present study attempts to investigate four Persian translations of Animal Farm based on forensic linguistics framework. The main goal of the present research is to demonstrate the patterns of plagiarism detection between different versions of translations of the same original piece of work based on textual similarities and differences. The project primarily centers on this question: what linguistic criteria are determinant in detecting plagiarism in translated texts? For our analysis, the data of the present research has been elicited from four Persian translations of this novel chosen by a time-lapse of 20 years between translations. Data were analyzed based on plagiarism detection patterns introduced by Turell (2004). The results of the study revealed a case of plagiarism among investigated translations. Disputed text overlaps 73.5% and 42.6% with plagiarized text in terms of vocabulary and phrasal similarity, respectively. In terms of unique vocabulary, they show the proportion of 17.6% to 15.16%. The disputed and plagiarized texts also have 35 shared-once only words and 22 shared-once only phrases. The article concludes that the proposed quantitative criteria of Turell’s model perform well in plagiarism detection which replicates the results of previous studies. We believe that science society of Iran must pay more attention to plagiarism in order to find a solution to suppress publication and proliferation of the plagiarized texts.
 

 
 
 
 
 

Volume 18, Issue 5 (11-2018)
Abstract

Air pollution as a silent murderer of metropolitan areas demanded huge amounts of attractions. During the past few decades, after London 1954 black days, the world encountered a novel problem which was made by anthropologic actions. Scientific researches for scrutinizing the air pollution and its effects on humankind and the environment, started and improved after chronic influences of contaminations which in this era prognostication of pollutants and finding the relationships between parameters out, seems to be undeniable. Ozone as a tropospheric gas, has severe impacts on the all creatures while the human beings are more delicate in conjunction with this gas where it can destroy ability lungs and cause asthma and other pulmonary diseases. In the present article, the two most prevailing approaches for prediction, applied to the forecast tropospheric ozone value considering eight other photochemical precursors and meteorological parameters. Sulfur dioxide (SO2), nitrogen dioxide (NO2), carbon monoxide (CO) and particulate matters (PM2.5, PM10) as photochemical precursors, and also humidity, air temperature and wind speed as meteorological parameters, after data preparation, used for ground level ozone prognostication in Tehran, Iran, with a condensed population where suffers from severe air contaminations and high rate of daily death, related to the air pollution. Used data series, have been collected from 22 regions of the cited city during 2 years (2014 and 2015). Two evaluation criteria, root mean square error (RMSE) and correlation coefficient (R), selected for comparison of applications. Support vector machine (SVM) and artificial neural networks (ANN) as capable soft computing approaches which have been used in numerous areas of science, opted in this research. Support vector machine with classification of other eight parameters and by 286 vectors as a classifier and 97 border vectors, sorted the 70 percent of data sets as training and the residual amount of parameters used as testing data sets. Radial basis function (RBF) selected as Kernel function. Artificial neural network works as like as human brains and neurons between layers transfer datasets and process them during the run time, where in the recent paper the layer number of the created network is one for hidden layer and one for the output layer and 10 neurons have been selected for hidden layer and one for the output layer. Network type of this system is feed-forward with back propagation and TRAINLM used as training function and LEARNGDM used for adaption learning function. Both approaches depicted reliable and acceptable results, where RMSE and R values for support vector machine, respectively 0.0774 and 0.8456, also artificial neural network resulted 0.0914 for RMSE and 0.8396 for R, which are reasonable outcomes. As the outcomes for training datasets were better than the results for testing datasets, both approaches showed acceptable performances because of over-training controlling, which is a serious and prevalent difficulty of soft computers. Support vector machine, with lower root mean square error and higher correlation coefficient selected as better application for ground level ozone prediction. These series of studies are supportive for calibration of measuring systems and due to their expensiveness, soft computing is the most reliable and affordable substitute for the past machines. Also the analysis of tolerances among the parameters illustrated that CO, Temperature and NO2 are the most effective where, PM2.5 had the least amount impact on O3 forecasting process.

Page 1 from 1