Keywords : SVM

Comparative Studying for Opinion Mining and Sentiment Analysis Algorithms and Applications

Rana Z. Alobaidy; Ghaydaa A.A. Al-Talib

AL-Rafidain Journal of Computer Sciences and Mathematics, 2018, Volume 12, Issue 2, Pages 13-23
DOI: 10.33899/csmj.2018.163578

The amount of the available data increases the ability to analyze and understand. The internet revolution has added billions of customer’s review data in its depots. This has given an interest in sentiment analysis and opinion mining in the recent years. People have to depend on machines to classify and process the data as there are terabytes of review data in stock of a single product. So that prediction customer sentiments is very important to analyze the reviews as it not only helps in increasing profits but also goes a long way in improving and bringing out better products.  In this paper , we present a survey regarding the presently available techniques and applications  that appear in the field of opinion mining , such as , economy , security , marketing , spam detection , decision making , and elections expectation.

Detection of desert percentage in Al-Hatra Region based on image contents

Saja Mallaaloo; Ghaydaa A.A. Al-Talib

AL-Rafidain Journal of Computer Sciences and Mathematics, 2014, Volume 11, Issue 2, Pages 37-51
DOI: 10.33899/csmj.2014.163755

In this paper we benefiting from Satellite imaging to retrieve information by using its contents, which is the pixels value of the image and by using the information of groups of pixels like texture, color gradation etc….then analyzing these information to extract spatial and temporal information of this images. Content Based Information Retrieval (CBIR) technique was used to retrieve image contents depending on visual objects of it. Support Vector Machine (SVM) technique was put into use by depending on more than one function like polynomial and RBF, then applying every one of them alone with the training image with different blocks size, then using block size and function that give best result from the training phase to be applied on the test images.
The Satellite imaging was classified into two areas; desert and none desert in order to find the desert percentage of each image and comparing increasing of the desert percentages in Al-Hatra Region as a typical desertification area in nenavah governorate on different temporal periods. The language used in building the system is Matlab R2011a.

An Investigation for Steganalysis in Color Images

Samah F. Aziz; Ahmed S. Nori

AL-Rafidain Journal of Computer Sciences and Mathematics, 2011, Volume 8, Issue 2, Pages 151-167
DOI: 10.33899/csmj.2011.163657

With science developing and techniques used in Information hiding, there are another techniques wall together for Steganalysis.
Steganography is considered as the new and the complementary system of Cryptography that took a long time in transferring secret and important messages through the networks Internet. Then there was the emergence of what complements Steganography as a science that analysis and discover the content of the secret messages and this science is Steganalysis.
This study tackled and manifested the ideas of analysis processes that can be followed to interpret the secret messages and discovering them either by means of knowing about their existence only or the capability of extracting them in full.
The work relied on two important technologies; the first is called the Support Vector Machine (SVM) and the second is called Fisher Linear Discriminator (FLD). The SVM technology has been used with the blind application idea while FLD has been used with the blind and non-blind application ideas using colored images which are PNG and BMP.
Results proved the high efficiency of the two technologies in detecting the image that includes the secret messages and comparisons were varied between the two technologies in terms of detection rate, fault and the execution time.