Visualisasi pohon sintaksis berasaskan model dan algoritma sintaks ayat bahasa Melayu
Previous works that produce syntactic tree output has disregarded additional relevant components such as sentence checking, sentence correction, the syntax tree visualization and the words attributes of each sentence. As such, this study aims at producing an algorithm for syntactic tree output enhan...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2018
|
| Subjects: | |
| Online Access: | https://etd.uum.edu.my/6890/1/DepositPermission_s92715.pdf https://etd.uum.edu.my/6890/2/s92715_01.pdf https://etd.uum.edu.my/6890/3/s92715_02.pdf https://etd.uum.edu.my/6890/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Previous works that produce syntactic tree output has disregarded additional relevant components such as sentence checking, sentence correction, the syntax tree visualization and the words attributes of each sentence. As such, this study aims at producing an algorithm for syntactic tree output enhancement from which the relevant output component mentioned above can be produced. The additional components namely sentence checking, sentence correction, syntax tree visualization (VPS) and word attribute are modelled into a package prior to translating them into a tangible output. In term of rules, previous studies have used phrase-structure rules (RSF) in analysing the Malay sentence. But RSF has been found to be a non-universal formula. Our work has brought us to the introduction of X-bar rules for BM VPS, which consequently becomes one of the contributions of this study. To achieve these objectives (the algorithm, the model and the X-bar rules), five phases of research methods involved namely identifying the research gap, the sentence and rules categorization, model and algorithm design phase, prototype development evaluation and conclusion phase. Parseval assessment method, which is an output evaluation method in natural language processing, was used for the evaluation. Point of analysis were the recall and precision valuation metrics. For VPS output, the average results obtained were 100% for recall and 97.8% for precision. For sentence correction, the results given were 100% for recall and 87.8% for precision. These results proved that the algorithm and model, for syntactic tree output enhancement, are generalisable enough to be tested on other languages. User evaluation on the prototype was also performed yielding in the average subjective satisfaction of 87.9% and a mean score of 6.157, based on semantic differential scales of 1 to 7. Cognitive assessment was also recorded, obtaining average cognitive score of 84.6% with a mean score of 4.230, on the scale 5. Analysis on those results indicated positive scores on the model-based product specifically on usefulness, ease of use, ease of learning, subjective satisfaction, and cognitive measures. It can be concluded that the algorithm and model proposed were useful for the development of the prototype. The prototype is therefore beneficial as an educational assistance to understand Malay sentences when provided with enhanced output on sentence checking, sentence correction, syntax tree visualization (VPS) and words attribute |
|---|