CSTY 2018

4th International Conference on Computer Science and Information Technology

October 27 ~ 28, 2018, Dubai, UAE

Learn More Submission
Accepted Papers
Improving ESL Writing Using AccurIT as an Online Word-Collocation Checker
Basem Y. Alkazemi
Umm Al Qura University, Saudi Arabia

Learners of ESL usually face considerable difficulties to write correct sentences that reflect the semantical meaning of the original form. Students tend to translate words of a sentence separately or think of a sentence in their mother language first then translate it to the second language. This might lead to inaccuracy of words collocations as the selected translated words cannot reflect the intended meaning of the sentence. In this paper we presents our prototype of a collocation checker namely, AccurIT, than can help students to check the semantic of their written translation. The tool follows a statistical approach that utilizes search engines to check the validity of a sentence then advise the student about the correctness and possible variations of the written sentence. AccurIT has been packaged as a plugin to MS-word to optimize its functionality in terms of checking for the semantics of sentences in addition to the regular checks that currently word processor is able to perform( i.e. spelling, grammar)

Countering Terrorism on Social Media Using Big Data
Ali Alzahrani1, Khalid Bashir1, Turki Alghamdi1 and Hanaa Aldahawi2
1Islamic University of Madinah, Saudi Arabia and 2King Abdulaziz University, Saudi Arabia

Terrorism and violence have been used by miscreant groups and individuals to disrupt the normal course of events. This is not a new phenomenon and has existed since long. However, with the information age new and innovative methods of spreading messages related to terrorism and expanding through recruitment on social media have been observed. This is alarming due to the very reach and speed of propagation on social media. To ensure safety, harmony and peace it is important that the use of social media for terrorism be contained. In the paper we discuss the various methods used by terrorists on social media to gain a farther reach. We identify that the inherent structure of social media, the amount of data and language understanding pose challenges as well as opportunities in control efforts. We propose a strategy for the restraint of terrorist activities by using data mining methods on big data created by social media in combination with natural language processing for language understanding and social network analysis for uncovering the underlying structure and association of terrorist groups and their activities.

Smart mobile application to recognize tomato leaf diseases using Convolutional Neural Networks
1 Rabat IT Center,ENSIAS, Mohammed V University in Rabat .2 Dept. Mathematics, University of New Mexico, New Mexico, USA

Deep Convolutional Neural networks(CNN) recently has recognized much advances in many fields, in particularly in computer vision such as classification, object detection, segmentation, etc. Evenifitstremendousadvancesincomputervision tasks, CNN face many challenges, such as computational burden and energy, to be used in mobile phone and embedded systems. The aim of this paper is to build a smart mobile application to recognize tomato leaf diseases that can be used by non-technical farmer. To build such application,our CNN model has been inspired from Mobile Net CNN model and trained on tomato leafs dataset.

Product Recommendation using Sentiment Analysis of Reviews: A Random Forest Approach
Gayatri Khanvilkar and Prof. Deepali Vora
Vidyalankar Institute of Technology Mumbai, India

People are now a day more attracted towards Social media and online shopping. Due to growth in social media all the fortune companies are working on Sentiment Analysis. Sentiment analysis in NLP (Natural Language Processing) has become a major area. This paper explores the performance of Machine Learning Algorithms such as Multinomial Nave Bayes algorithm, Logistic Regression, SVM Classifier, Decision Tree and Random Forest used for sentiment analysis. Comparative tabulation of above mentioned classifiers is created to analyze the performance of sentiment analysis. Random Forest is a flexible and easy to use supervised machine learning algorithm that produces a great result most of the time. In proposed system, Random Forest shows outstanding performance. The achieved polarity of reviews will be used to provide recommendation list to users. Keywords- Sentiment analysis, Recommendation System, Machine learning, Random Forest, Contentbased Recommendation I

Comparing Results of Sentiment Analysis Using Naive Bayes and Support Vector Machine in Distributed Apache Spark Environment
Tomasz Szandała
Wrocław University of Technology Wrocław Poland

Short messages like those on Twitter or Facebook has become a very popular opinions sharing tool among Internet users. Therefore microblogging websites are nowadays rich sources of data for opinion mining and sentiment analysis. However it is challenging because of the limited contextual information that they normally contains. Furthermore the greatest benefit can be achieved by collecting sentiment class in real time - when the post is published, in order to react as soon as possible. Nevertheless, most existing solutions are limited in centralized environments only. thus, they can only process at most a few thousand tweets. Such a sample, is not representative to define the sentiment polarity towards a topic due to the massive number of tweets published daily. Sample analysis has been performed using Machine Learning methodologies alongside with Natural Language Processing techniques and utilizes Apache Spark's Machine learning library, MLlib, on a labeled (positive/negative) corpus containing 4234 tweets regarding Presidential Election in USA in 2016. The analysis has been completed using distributed Apache Spark environment with simulated stream of data from Kafka database

A Belief Approach For detecting Spammed Links In Social Networks
Salma Ben Dhaou1, Mouloud Kharoune2 , Arnaud Martin2 and Boutheina Ben Yaghlane1
1Larodec, ISG of Tunis, Tunisia
2Univ Rennes, CNRS, IRISA, France

Nowadays, we are interconnected with people whether professionally or personally using different social networks. However, we sometimes receive messages that are not correlated to the nature of the relation established between the persons. Therefore, it became important to sort out our relationships. Thus, based on the type of links that connect us, we can decide if this last is spammed. Thereby, we propose in this paper a belief approach in order to detect the spammed links. Our method consists on modeling the belief that a link is perceived as spammed by taking into account the prior information of the nodes, the links and the messages that pass through them. To evaluate our method, we first add the noise to the messages, then to both links and messages. Second, we select spammed links and observe if our model manages to detect them. All the results were compared with those of the baseline. The experiments indicate the efficiency of the proposed model.

A Comparison Of Deep Learning Approaches For Vehicle Detection
Mohamed Ashraf1, Hossam E. Abd El Munim2, Ahmed Hassan Yousef2, and Sherif Hammad2
1Department of Mechatronics Engineering, Ain Shams University, Cairo, Egypt
2Department of Computer Engineering, Ain Shams University, Cairo, Egypt

Vehicle classification and detection for autonomous driving is now one of the biggest research topics in the industry. The objective of this process is to help the vehicle's embedded computer know what is around the vehicle, evaluate the situation and take decisions in real time. In this paper, we will focus on the classification process of the video-based vehicle detection. Different deep learning techniques are employed which is known as convolutional neural networks (CNN) architectures, as ResNet, InceptionV3, Inception-ResnetV2, MobileNetV2, NASNet and PNASNet architectures. Two different datasets are used for evaluation; Kitti dataset for car detection only, and MIO-TCD for different types of vehicles. Results showed that best performance was obtained by the Inception-ResnetV

Dynmic Inference Of personal Preference For Next-To-Purchase Items By Using Online Shopping Data
Yun-Rui Li1 Ting-Kai Hwang2 and Shi-Chung Chang1
1Department of Electrical Engineering, National Taiwan University, Taipei, Taiwan
2Department of Journalism, Ming Chuan University, Taipei, Taiwan

With more and more people shopping online, companies deal with customer data input not only in high volume but also dynamic. In order to attract target customers more effectively and to provide customers with more personalized services, how to automatically extract personal preference from the real-time data and make real-time recommendation has been growing in importance for businesses in the competitive modern society. Current data analysis methods for online shopping recommendation largely rely on historical transaction record. Analyses have indicating that next items a customer would like to buy not only depend on one's past historical records but on the item currently being put into the shopping cart. This paper designs an engine to combine each customer's past transaction and current shopping cart data to dynamically infer one's preference for the next items. The design, Transaction-Data Based Real-time Preference Inference Engine (TRPIE), consists of two innovative ideas. The first exploits the purchasing sequence information and turns one's purchase history into a temporal series of data, where a customer's dynamic purchasing behaviour information lies. The second is a design of a two-layer Recurrent Neural Network (RNN) for extracting personal purchasing preference pattern from the temporal series of data to infer preference of next items. A reference implementation of TRPIE design integrates existing tools such as Keras, tensorflowTM, sklearnTM, and MlxtendTM. Test results over real data from 1374 people show that prediction accuracy has doubled that obtained by a basket analysis method, which ignores sequentiality of purchasing items.

Music Sequence Prediction with Mixture Hidden Markov Models
1Tao Li, 2Minsoo Choi, 3Kaiming Fu, 4Lei Lin
1Department of Computer Science, Purdue University, USA
2School of Industrial Engineering, Purdue University,USA
3School of Mechanical Engineering, Purdue University, USA
4Goergen Institute for Data Science, University of Rochester, USA

Recommendation systems that automatically generate personalized music playlists for users have attracted tremendous attention in recent years. Nowadays, most music recommendation systems rely on item-based or user-based collaborative ?ltering or content-based approaches. In this paper, we propose a novel mixture hidden Markov model (HMM) for music play sequence prediction. We compare the mixture model with state-of-the-art methods and evaluate the predictions quantitatively and qualitatively on a large-scale real-world dataset in a Kaggle competition. Results show that our model signi?cantly outperforms traditional methods as well as other competitors. We conclude by envisioning a next-generation music recommendation system that integrates our model with recent advances in deep learning, computer vision, and speech techniques, and has promising potential in both academia and industry.

Parag Wadnerkar, R. S. Dalu
COE Amravati, India

A comparative study was used to outline the literature in the research topic. This paper presents an analytical study of agro manufacturing industries in region from year 2005 to 2018. The AHP considers a group of analysis criteria, and a group of other choices among that the most effective call is to be created. It's vital to notice that, since a number of the standards may be different, it's not true normally that the most effective choice is that the one that optimizes every single criterion, rather the one that achieves the foremost appropriate trade-off among the various criteria. The AHP generates a weight for every analysis criterion in step with the choice maker's try wise comparisons of the factors. the upper the burden, the additional vital the corresponding criterion. Next, for a fixed criterion, the AHP assigns a score to each option according to the decision maker's pair wise comparisons of the options based on that criterion. The higher the score, the higher the performance of the choice with relevance the thought-about criterion. AHP combines the standards weights and also the choices scores, so deciding a worldwide score for every possibility, and a ensuant ranking. This paper present need of current era to study SCM in agro manufacturing industries in vidarbha region.

Claudia Jacy Barenco Abbas, Rapahel Montandon and Renato Azevedo,
Campus Universitario Darcy Ribeiro Brasilia, Brazil

This paper is concerned with the development of interactive systems for smart meeting rooms. Automated recognition of video events is an important research area. We present an LTL model of basic objects and activities recognition in smart meeting rooms using object attribute details. There are still problems of misrecognizing objects in existing visual recognition methods because lack of enough feature attributive information details. This paper investigates morphological approach to increase recognition accuracy using variability in a limited area of moving object using object attribute details. The proposed methods are also compared to popular and recent methods of visual object and event recognition.

Vatsal Shah1 and Vivek Kumar2,
1Department of Research, ezDI, LLC, Louisville, USA

Complete and accurate clinical documentation in the medical record has a direct impact on the assignment of codes, more accurate levels of reimbursement, and is critical to the higher quality of patient care. This paper describes the development of a system which can automatically flag the cases if there is an opportunity of improvement in patient clinical documents. Automated Clinical Documentation Improvement (CDI) leverages the natural language processing (NLP) and contextual understanding of health record structure with additional business rules logic, helping CDI specialists identify critical documentation information that may be missing from the medical record. This results in more specific coding opportunity and better understanding of the clinical complexity for accurate reimbursement. This system helped increase CDI specialists' productivity by efficiently filtering cases which need more attention from them.