IRSPAS 2016

Permanent URI for this collectionhttp://repository.kln.ac.lk/handle/123456789/15651

Browse

Subject: search.filters.subject.Big data

Search Results

Now showing 1 - 3 of 3
  • Thumbnail Image
    Item
    Real-time big data video analytics for unorganized traffic environments
    (Faculty of Science, University of Kelaniya, Sri Lanka, 2016) Ranaweera Arachchi, I.; Jayalal, S.; Rajapakse, C.
    Traffic on local roads has reached such a level that it is necessary to address the issue of traffic congestion and seek complex transport solutions for the city. Increase of the number of vehicles on the road becomes one of the key reasons for increasing traffic congestion. Traffic congestion is associated with massive financial and manhour loss and therefore attempts to alleviate this has been of keen interest. The basis of almost all those approaches is traffic monitoring and analysis, leading to having an effective traffic management system. Most traffic management systems are applied in well-organized traffic environments such as highways, where driver discipline is high. But in unorganized urban environments as seen in Sri Lanka, road traffic behavior vary from the accepted standards. Driver and pedestrian indiscipline cause huge traffic congestions in urban areas. Hence in such a scenario, a system that monitors road traffic on different traffic environments is very useful. There are several existing techniques such as Magnetic Loops, Microwave RADAR, Infrared Detectors, Ultrasonic Detectors and Camera Based Systems. Traffic monitoring systems require short processing time, low processing cost and high reliability. Therefore, according to the literature, camera-based monitoring is the best-suited technique for traffic monitoring. Real-time video analytics are part of a centralized approach to modern traffic management which is defined as computer vision-based surveillance that provides algorithms for object detection, tracking, classification and trajectory analysis using real-time traffic surveillance video. It usually uses roadside cameras (CCTV) to obtain traffic information and transmit it to central servers, exhibiting real-time operability of big data. In this study, several approaches and algorithms for moving object detection, based on temporal differencing method, optical flow method, background filtering are compared and a novel real-time vehicle detection and classification algorithm based on background filtering will be proposed and re-engineered in order to be applicable to challenging unorganized traffic environments. The solution will classify vehicles individually and their trajectories in real time in unorganized traffic environments in order to analyze the behaviors of the drivers as well as pedestrians on the road. We use OpenCV which is a library of programming functions mainly aimed at real-time computer vision, for implementing and testing algorithms. Data will be collected via pre-recorded video clips from Kiribathgoda junction in the western province, for the testing purpose and real- time CCTV surveillance video is going to be used as the input for implementation. A comprehensive data analysis is required to be conducted to address the higher processing requirement of such videos. The solution will be validated for performance subsequently. The final objective of this research is to come up with an optimum algorithm for vehicle detection and classification in unorganized traffic environments which would help to analyze the behavior of road users. The solution will lead to reduced traffic congestion in the country by enhancing the efficiency and effectiveness of traffic monitoring and analyzing systems.
  • Thumbnail Image
    Item
    Detection of cyber bullying on social media networks
    (Faculty of Science, University of Kelaniya, Sri Lanka, 2016) Priyangika. S.; Jayalal, S.
    Social Media is becoming an integral part of people’s daily lives today. It is an effective way of sharing one’s life experiences, special occasions, achievements and other events with their friends and family. Although it is a fruitful way to communicate with groups, some people find themselves being insulted or offended by others who are involved in certain post or conversations. These insulations can be based on racism, using profanity or any other vulgar or lewd language. This cyber bullying needs to be monitored and controlled by the social media site owners since it will highly effect on the number and safety of the active site membership. Currently, there is no automated process of identifying offensive comments by the social network site itself. It can be only diagnosed by humans after reading the comments, flagging or reporting them to the owner of the site or blocking the offender. Considering the massive big data set generated in social media daily, automatically detection of offensive statements is required to reduce insulation effectively. For this purpose, text classification approach can be applied where a given text will be categorized as insulting or not, through learning from a pre-learned model. In order to develop the model, data was collected from the popular data repository site named www.kaggle.com. The dataset consists of comments posted on Facebook and Twitter. Firstly the dataset was divided into training data set and test data set. Then the collected data was preprocessed by removing the unwanted strings, correcting words and eliminating duplicate data fields. In the next step, features or keywords were extracted which are qualified to distinguish a statement as ‘insulting’ using N-grams model and counting methods. Feature selection is done using Chi- Squared test and finally apply classification algorithms for separating insulting comments and non-insulting comments from a dataset given. Machine learning algorithms such as Support Vector Machines (SVM), Naïve Bayes, Logistic Regression and Random Forest are used for this. Out of the classification algorithms, SVM is to be performed better than other algorithms since this is a two-class classification problem and a comment is to be classified only into two separate classes which are ‘insulting’ and ‘neutral’. With an exact separation of a given comment into ‘insulting’ and ‘neutral’ category, cyberbullying happening through offensive comments posted on social media sites can be detected.
  • Thumbnail Image
    Item
    An approach to personalize learning using big data analytics for higher education
    (Faculty of Science, University of Kelaniya, Sri Lanka, 2016) Jabir, A.; Rajapakse, C.
    The concept of BYOD (Bring Your Own Device) has gained popularity in studentcentered learning and higher education institutions make significant investments on improving the wireless network to enhance this. Virtual Learning Environment and Learning Management Systems were introduced and personalization of learning becomes the next milestone. The huge streams of data produced by these Wi-Fi networks makes ground for Big Data analytics to identify opportunities in educational environments to adopt personalized learning. The term ‘Personalization’ refers to the tailoring of content and recommending items by inferring what interests a user based on previous or current interactions with that user, and possibly other users. This research proposes an approach to personalize learning on an online learning platform by providing personalized recommendations of educational web resources, comparative feedback and allocate personalized bandwidths based on the concept of deprioritization (lowering priority ranks of heavy users). Concepts of Big Data analytics and data mining techniques will be used to satisfy the objectives. The approach consists of offline phase (modelling phase) and online phase (recommendation /deprioritization) phase. In the offline phase, models will be developed for recommendation and deprioritization separately. For recommendation a hybrid filtering method will be used. k-Nearest Neighbour, a user-based collaborative filtering technique, will be used with correlation based similarity measure with demographic filtering based on demographic classifiers (faculty, year, General/Special/Honors, GPA) to eliminate the cold start problem. To increase the efficiency and accuracy, k-means clustering will be used as an intermediate step to determine usage clusters to group users exhibiting similar browsing patterns and page clusters to discover pages with similar access patterns. For this the access logs of the University of Kelaniya’s Wi-Fi network will be utilized. The parameters for usage clustering would be the timestamp, web resource and category (education, social networking, gaming etc.) whereas the parameters for page clustering would be category and temporal concepts. In the online phase, first the cluster that the current active user belongs to will be identified and k-NN will be applied on that particular cluster to recommend web resources. These techniques also provide the basis for comparative feedback compared to top scorers of the same area of major. For personalized allocation of bandwidth a separate k-means clustering will be performed to identify heavy users during the offline phase. During the online phase deprioritization will be applied accordingly if the current user belongs to the heavy users cluster and there is a heavy traffic in the network. Cross validation will be used to evaluate the models.