UTILIZATION OF SEMANTIC TEXT SIMILARITY IN DEVELOPMENT OF WORK COMPETENCY REPOSITORY IN INFORMATION TECHNOLOGY FIELD

• Main Author: Fadhel Ramadhan, Adam
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/64088
• Institution: Institut Teknologi Bandung

In Indonesia, competence can be measured through 3 things, namely a person’s knowledge and abilities also the person’s work attitude in the working environment. This competence is measured in a standard way called unit of competence. The making of competency units in Indonesia is still being carried out by different people and different teams. This causes some problems. The first problem is there are some competency units have the same meaning as other units even though they have different sentences. The second problem is that competency units are still made separate from each other. To overcome those problems, in this project a website is created. A website that utilizes semantic text similarity technology so that it can check beforehand the level of similarity between the unit that’s going to be made and existing units. In addition, the website is also capable to store all competency units that are already made before, so they are not separated from each other. In testing the model, pearson correlation coefficient (PCC) is being used to calculate its quality. Based on test results, the best semantic text similarity model has PCC value of 61.43%. Besides that, the website is also able to display a list of competency units that have most similarity with the search query. Among them there is a similarity level above 65%, between 50% - 65% and below 50%. The list will be displayed in order from most similar to least similar.