Semi-automated construction of Neglish-Malay machine readable dictionary for technical terms

Abstract

This project presents a method for semi – automated construction of English – Malay machine readable dictionary for technical terms. We proposed to use Keyword Density in order to classify the category for each term by measuring the weight of the term with Visual Studio using visual basic language. In the meantime, Cosine Similarity algorithm is used to measure the similarity between two sentence which are definition and sentence from the journal using C language. In order to calculate the category, 523 trainings data which is a set of journal for each term was collected. Then, we preprocessed the journal by using Brill’s Tagger with Penn-Tree Bank Tagger. We assigned 50 terms to test the algorithm. By using word extraction method the terms occurrence was counted. The total of the word in the category journal are also calculated. To categorize the term, we calculated the keyword density. For example sentence extraction, the data is used from the highest cosine similarity measurement between definition and sentence from journal. The sentence with the highest value was extracted as example sentence by the system. By using this algorithm, the Precision for the example sentence is 79%, Recall 90% and the F-Measure is 84%. It can be considered as a successful since the result is high. As a conclusion, based on the result, the proposed method shows a great potential with further improvement.