https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4766 2026-03-29T06:14:33Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4770 Title: Improving Performance of Statistical Machine Translation between Morphologically Rich and Low Resourced Language Pairs Authors: Pushpananda, B. H. R. Abstract: ABSTRACT Statistical and machine learning methods have trumped rule-based approaches to machine translation (MT). Unfortunately however, most language pairs do not have acceptable MT systems due to the unavailability of adequate amounts of parallel data. In addition to this problem, morphological richness is a confounding factor that needs to be addressed in developing a successful MT system, especially for agglutinative languages. When any one or both languages requiring to be translated are both data scarce and also morphologically rich, developing a good MT system for them is very difficult. In this research, we investigated several approaches and techniques for translating from and into both morphologically rich and low resourced languages, namely between Sinhala and Tamil. According to the literature, integrating morphological information to an MT process will is one way to improve MT between morphologically rich language pairs. However, the unavailability of language resources such as morphological analyzers and part-of-speech taggers hamper progress for translations between such language pairs. Since this affects both Sinhala and Tamil languages we integrated unsupervised learning and transliteration approaches to a phrase based statistical MT approach in an attempt to overcome these issues. Initially, we performed three sets of experiments employing different morphological representations namely; a “fully-morpheme-like”, a “semi-morpheme-like with merged suffixes” and a “semimorpheme- like without merged suffixes”. Based on these approaches, the fullymorpheme- like and the semi morpheme-like without merged suffixes approaches have shown improvements of approximately 2.9 and 1.7 BLEU points respectively compared to the baseline phrase-based approach. We were also able to reduce the out-of-vocabulary rate from 25% to 1% using these techniques. Based on a purely transliteration approach, we were only able to achieve a 0.44 BLEU point increment compared to the baseline approach. We then combined the fully-morphemelike approach with the transliteration (direct-mapping) approach to build a “joint” model which was able to achieve a 4.69 BLEU point improvement over the baseline and a 1.83 BLEU point improvement over the fully-morpheme-like approach. iii We have also carried out a critical appraisal of the kernel based MT (KBMT) framework which had shown some promise in early Sinhala-Tamil (SI-TA) translation work. We explored its performance on a large French-English (FR-EN) corpus and our much smaller SI-TA corpus. We introduced two novel approaches to filter the training examples based on the input using cosine similarity values. Overall, the KBMT approach gives lower quality translations compared to other approaches we have tried in this research for the morphologically rich language pair Sinhala and Tamil. We finally compared the results obtained from our kernel based translator, our phrase-based baseline, the baseline augmented by our fully-morpheme-like segmentation approach, and our “joint” model with results obtained from Google Translator for the SI-TA language pair. Both manual and automatic evaluation techniques were used to measure the quality of the resulting translations. Based on an automatic evaluation, the Google Translator gives the lowest BLEU score compared to our other models. However, expert manual evaluation shows that the quality of the translation output by the Google Translator is somewhat closer to our baseline approach. Overall, our unsupervised fully morpheme-like segmentation approach in its “joint” form has shown the best BLEU score in both automatic and manual evaluations and is significantly better than the Google translator for the Sinhala-Tamil language pair. 2017-01-01T00:00:00Z https://dl.ucsc.cmb.ac.lk/jspui/handle/123456789/4769 Title: Design and Evaluation of Mobile Learning Tools to Facilitate Guided-informal learning in the Domain of Agriculture Authors: Dissanayeke, D. M. U. I. Abstract: Abstract Mobile technology enhanced learning, which is commonly known as m-Learning, refers to the use of mobile devices along with the related technologies in offering learning contents, while also facilitating user interactions and collaborations in order to impart learning. This includes linking the learning community with each other to facilitate easier transferring of information irrespective of the location of the learner and the time of the day. Learners enjoy a great deal of autonomy in designing their own learning process, by making decisions such as whom to contact or which learning resources to be referred. Even though mobile technologies have been effectively used in disseminating critical information to agricultural communities in the past, the use of mobile phone as a learning device has not been adequately addressed. Information related to market prices and weather has been effectively communicated using Short Message Service (SMS) based systems; however these were mostly one-way or top-down approaches. Less emphasis was given on facilitating interactions and collaborations among the stakeholders, which is an important considerations in facilitating learning. Consequently the possibility of using mobile devices to facilitate learning was not adequately answered in the domain of agriculture. This research explores design and implementation of mobile learning to facilitate guided-informal learning among a group of young farmers in Kandy district, Sri Lanka. This thesis was written to address the research question ‘how to enhance the traditional agriculture learning systems by designing, developing and implementing mobile phone based informal learning opportunities?’ A design based research methodology was adopted, thus research interventions were designed, implemented and tested in collaboration with the practitioners, in real world settings. The research process included four phases, namely problem analysis, developing an m-learning solution, iterative testing and development of the learning solution in practice, and evaluation. Activity theory, which was often used in human computer interaction research, was used as the main inspiration and guiding theory in analyzing and designing the learning environment. The main concepts in activity theory, namely subject, object, tools, community, role and rules were used in conceptualizing the learning context and in studying the mobile learning design. Two m-learning interventions were planned considering the different types of mobile devices available with the study community: The first intervention was designed for basic feature mobile phones using the mobile SMS based Twitter platform. The learning community, which v includes instructors and learners, were linked to each other using Twitter web service. The instructor guided the learners in the learning process by posting questions, model answers and offering feedback. Learners interacted with the instructor and other learners using the m-Learning Approach (mLA) as well as personal calls, SMS, and face to face discussions in achieving pre-defined learning goals. The mLA offered a potential alternative when designing for communication and interaction facilitation among agricultural communities. It assisted in constructing knowledge, providing opportunities to interact with instructors, and obtain useful feedback in reinforcing learning through such interactions. The main drawbacks in the mobile SMS based Twitter version of the mLA were technical problems due to not receiving twitter messages, difficulties in moving towards higher order learning skills, and limited opportunities for collaborations due to SMS based platform & interface. The second intervention, was proposed to answer these drawbacks in the mLA, considering the future mobile learning environments. The second version of the mLA was designed for a Smartphone based learning environment, using android platform. This application, included similar features from the SMS based version of mLA such as questions, model answers and feedback, together with new features to facilitate collaborations between learners such as discussion forums. A prototype of smartphone based version of mLA was designed and tested with the same study community. The study suggests ‘instructor led informal m-learning systems’ as a solution to enhance the traditional agricultural learning systems. With proper training and guidance, the agricultural instructors can establish learning communities, and practice mobile based informal learning. M-Learning systems need to be designed to match with the available technologies among the target communities to assure easy adoption. The significance of the research lies in its efforts to design and implement a novel m-learning approach to facilitate guided informal learning among the young farmer communities in Sri Lanka. This research also contributes to the discussion on how to use of Twitter social media in creating low cost learning solutions for agricultural communities of practice. 2017-01-01T00:00:00Z