Science
Permanent URI for this communityhttp://repository.kln.ac.lk/handle/123456789/1
Browse
6 results
Search Results
Item The Role of Social Media (Twitter) in Analysing Home Violence: A Machine Learning Approach(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Adeeba, Saleem; Banujan, Kuhaneswaran; Kumara, B.T.G.S.Home Violence (HV) has been a persistent issue across the globe, transcending economic status and cultural boundaries. The COVID-19 pandemic has further exacerbated this problem, bringing it to the forefront of public discourse. This study aims to analyse the impact of HV by utilising Twitter data and Machine Learning (ML) techniques, categorising tweets into three groups: (i) HV Incident Tweets, (ii) HV Awareness Tweets, and (iii) HV Shelter Tweets. This categorisation provides several advantages, such as uncovering new or hidden evidence, filling information gaps, and identifying potential suspects. Over 40,000 tweets were collected using the Twitter API between April 2019 and July 2021. Data pre-processing and word embedding were performed to prepare the data for analysis. Initially, tweets were categorised into HV Positive (containing relevant information) and HV Negative (noise or unrelated content) groups. Manually labelled tweets were used for training and testing purposes. Machine learning models, including Support Vector Machines (SVM), Naïve Bayes (NB), Logistic Regression, Decision Tree Classifier, Artificial Neural Networks (ANN), and BERT+LSTM, were employed for this task. Subsequently, HV Positive tweets were classified into the three aforementioned categories. Manually labelled tweets were again used for training and testing. Models such as Tf-IDF+SVM, Tf- IDF+Decision Tree, Tf-IDF+NB, and GloVe+LSTM were utilised. Several evaluation metrics were used to assess the performance of the models. The study’s results provide important new understandings of the prevalence, patterns, and causes of HV as they are reported on social media and how the general population reacts to these problems. The research clarifies how social media may help spread knowledge, provide assistance, and link victims to resources. These insights can be instrumental in informing policymakers, non-profit organisations, and researchers as they work to develop targeted interventions and strategies to address HV during and beyond the COVID-19 pandemic.Item Effectiveness of Machine Learning Algorithms on Battling Counterfeit Items in E-commerce Marketplaces(Department of Industrial Management, Faculty of Science, University of Kelaniya Sri Lanka, 2023) Gunawardhana, Kalinga; Kumara, B.T.G.S.; Rathnayake, R.M.K.T.; Jayaweera, Prasad M.For e-commerce marketplaces, counterfeit goods are a major issue since they endanger public safety in addition to causing customer unhappiness and revenue loss. Traditional techniques to identify fake goods in online marketplaces take too long and have a narrow reach, hence they are ineffective. Machine learning algorithms have become a potential tool for swiftly and precisely identifying counterfeit goods in recent years. The usefulness of two machine learning algorithms in identifying fake goods in online marketplaces is examined in this research. The study assesses the performance using a sizable dataset of descriptions, title, prices and seller names from many well-known e-commerce platforms. The study's findings show that machine learning algorithms significantly affect the detection of fake goods in online marketplaces.Item A data mining approach for the analysis of undergraduate examination question papers(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Brahmana, A.; Kumara, B.T.G.S.; Liyanage, A.L.C.J.Examinations play a major role in the teaching, learning and assessment process. Questions are used to obtain information and assess knowledge and competence of students. Academics who are involved in teaching process in higher education mostly use final examination papers to assess the retention capability and application skills of students. Questions that used to evaluate different cognitive levels of students may be categorized as higher order questions, intermediate order questions and lower order questions. This research work tries to derive a suitable methodology to categorize final examination question papers based on Bloom’s Taxonomy. The analysis was performed on computer science related end semester examination papers in the Department of computing and information systems of Sabaragamuwa University of Sri Lanka. Bloom’s Taxonomy identifies six levels in the cognitive domain. The study was conducted to check whether examination questions comply with the requirements of Bloom’s Taxonomy at various cognitive levels. According to the study the appropriate category of the questions in each examination, the paper was determined. Over 900 questions which obtained from 30 question papers are allocated for the analysis. Natural language processing techniques were used to identify the significant keywords and verbs which are useful in the determination of the suitable cognitive level. A rule based approach was used to determine the level of the question paper in the light of Bloom’s Taxonomy. An effective model which enables to determine the level of examination paper can derive as the final outcome.Item Social media mining for post-disaster management - A case study on Twitter and news(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Banujan, K.; Kumara, B.T.G.S.; Incheon PaikA natural disaster is a natural event which can cause damage to both lives and properties. Social media are capable of sharing information on a real-time basis. Post disaster management can be improved to a great extent if we mine the social media properly. After identifying the need and the possibility of solving that through social media, we chose Twitter to mine and News for validating the Twitter Posts. As a first stage, we fetch the Twitter posts and news posts from Twitter API and News API respectively, using predefined keywords relating to the disaster. Those posts were cleaned and the noise was reduced at the second stage. Then in the third stage, we get the disaster type and geolocation of the posts by using Named Entity Recognizer library API. As a final stage, we compared the Twitter datum with news datum to give the rating for the trueness of each Twitter post. Final integrated results show that the 80% of the Twitter posts obtained the rating of “3” and 15% obtained the rating of “2”. We believe that by using our model we can alert the organizations to do their disaster management activities. Our future development consists mainly of two folds. Firstly, we are planning to integrate the other social media to fetch the data, i.e. Instagram, YouTube, etc. Secondly, we are planning to integrate the weather data into the system in order to improve the precision and accuracy for finding the trueness of the disaster and location.Item Pedestrian detection using image processing for an effective traffic light controlling system(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Chathumini, K.G.L.; Kumara, B.T.G.S.Traffic congestion and road pedestrian accident are the two major issues that the Sri Lankan society faced toady. These two issues can be reduced by use of traffic light controlling system in an effective way. This research paper proposed a system to make effective PEdestrian LIght CONtrolled (PELICON) crossing system using image processing. The proposed system consists of three major parts. That is CCTV camera, the system, and pair of poles with standard traffic light system. First the system captures an image of pedestrians who are waiting to cross the road, using CCTV camera. Then the system processes the image to identify and detect the number of pedestrians. Finally, if the number of pedestrians exceeds a given threshold value or pedestrian waiting time is exceeded, then the logical part of the system works and produces a result to control the traffic light system. This system that uses PELICON crossing system could be more effective than a button clicking PELICON Crossing system.Item Data mining model for identifying high-quality journals(International Research Conference on Smart Computing and Systems Engineering - SCSE 2018, 2018) Jayaneththi, J.K.D.B.G.; Kumara, B.T.G.S.The focus in local universities over the last decade, have shifted from teaching at undergraduate and postgraduate levels to conducting research and publishing in reputed local and international journals. Such publications will enhance the reputation on the individual and the university. The last two decades has seen a rapid rise in open access journals. This has led to quality issues and hence chossing journals for publication has become an issue. Most of these journals focus on the monetary aspect and will publish articles that previously may not have been accepted. Some of the issues include design of the study, methodology and the rigor of the analysis. This has great consequences as some of these papers are cited and used as a basis for further studies. Another cause for concern is that, the honest researchers are sometimes duped, into believing that journals are legitimate and may end up by publishing good material in them. In addition, at present, it is very difficult to identify the fake journals from the legitimate ones. Therefore, the objective of the research was to introduce a data mining model which helps the publishers to identify the highest quality and most suitable journals to publish their research findings. The study focused on the journals in the field of Computer Science. Journal Impact Factor, H-index, Scientific Journal Rankings, Eigen factor Score, Article Influence Score and Source Normalized Impact per Paper journal metrics were used for building this data mining model. Journals were clustered into five clusters using K-Means clustering algorithm and the clusters were interpreted as excellent, good, fair, poor and very poor based on the results.