Browsing by Author "Herath, H.M.S.S."

Now showing 1 - 2 of 2
  • No Thumbnail Available
    Item
    A Preliminary Study on Absorption of Heavy Metal (Lead) From Synthetic Waste Water Using Mullet (Mugil spp.) Fish Scales
    (Uva Wellassa University of Sri Lanka, 2016) Herath, H.M.S.S.; Alakolanga, A.G.A.W.; Abeyrathne, E.D.N.S.
    Heavy metals are relative dense metals present in earth crust that are in high accumulation level in water bodies. Among several technologies to remove these heavy metals, biosorption is a novel concept that use in waste water facilities. This study was carried out tofind the feasibility of using mullet (Mugilidae spp.) fish scales as a biosorbant and find the best cost effective form of mullet fish scales to absorb heavy metal (Pb) in synthetic waste water.The effect of oven drying conditions, particle size, and dosage of fish scales and pH for the absorption of heavy metal (Pb) absorption were investigated by Atomic Absorption Spectrophotometer (AA240, 283.33 nm, Varian., Australia). The reusability was investigated with the use of nitric and hydrochloric acids. The highest mean absorption results were obtained for 3 g (98.70 0.73) of damaged 1-2 cm2 size (92.03 ± 0.51) with oven drying conditions of 80°C (99.10 ± 0.52) for 24 hours (93.93 ± 2.38). All parameters were indicated that there was a significant difference among the treatments (P < 0.05). In addition to that, absorption was not depended on pH and reusable ability was high with the nitric (19.83 ± 5.66) compared to hydrochloric acid (3.09 ± 0.37). But in cost effective manner new fish scales were more applicable. The observed reason for the efficient biosorption from the fish scales were the crystal structure, chemical composition of hydroxyapatite with porous structure and the highly ordered three dimensional structure of collagen. This research revealed that, there is a high potential to use mullet fish scales as a biosorbant for treat waste water in wastewater treatment facilities and it is a rapid, cost effective and high efficient biosorbant among other biosorbants. Keywords: Heavy metal, Biosorption, Cost effective, Hydroxyapatite, Collagen
  • No Thumbnail Available
    Item
    Sentiment Analysis in Sinhala Texts using Convolution Neural Networks
    (Uva Wellassa University of Sri Lanka, 2021) Herath, H.M.S.S.; Siyamalan, M.
    Because of the rapid development of the information communication technology, an enormous amount of data is produced, shared across the internet and other media. Opinion mining, also known as Sentiment Analysis, is a technique, which can be used to detect the opinion of a given sentence or to make a judgement based the given sentence, can play a major role in automatically analysing this data. In addition, the development of Natural Language Processing in Sri Lanka leads the Sri Lankan native user to browse web in their native language and to express their opinions in their mother tongue. But in most of the cases Sinhala language was named as the morphological rich,less resourced language. An automatic solution for the text categorisation and opinion mining could be very useful for analysing sentences from Sinhala language. This work explores a Convolution Neural Network (CNN) based sentiment analysis techniqe, where, each word of a sentence is converted into a numerical representation using a pretrained FastText word embedding model. These numerical representations obtained for each word of the sentences are then used to train the CNN, in order to predict the opinion of the given sentences at test time. This CNN model is trained and tested on a Sinhala news comments dataset which consists of 5010 comments. There were 2520 negative comments and 2490 positive comments. Dataset used in this project is collected by crawling Sinhala online news sites, mainly www.lankadeepa.lk.This data was initially preprocessed by removing non- Sinhala characters, punctuation marks and stop-words. Our model was trained and tested on 70% and 30% of the data respectively. Experiments report a testing accuracy of 85%. Keywords: Convolution Neural Network; Natural Language Processing; Sentiment Analysis; Text analysis for Sinhala Language