Pengekstrakan data berasaskan pendekatan ontologi: kes data jujukan hidrologi

Abstract

Information Extraction is a process that extracts information from existing system source and stores into a database. Previous researchers had focus on information extraction for HTML data using wrapper approach. The drawback from this approach is resiliency where wrapper fails to function when the file of interest’s structure changes. Ontology based information extraction is an alternative solution for this problem. In this research, ontology based information extraction used hydrological data from Jabatan Pengairan dan Saliran (JPS) as the case study. Ontology based information extraction for hydrology domain or also known as ‘EkstrakPro’ is divided into three main processes; which are ontology parser process, keyword and sequences recognition process, and a data mapping process. ‘EkstrakPro’ used two inputs; the hydrology data and ontology extraction. An important feature in ‘EkstrakPro’ is that ontology extraction, where unit object is introduced to simplify the ontology maintenance. The sequential recognition algorithm is to solve the time consuming issues for extracting sequential data. Five types of hydrological data are used in the experiment. These data are divided into three categories; (i) original data taken from gauging machine, (ii) the altered data and (iii) the different sizes of data. Based on these categories, the information extraction resiliency and time taken have been measured using a precise equation and O-notation. The results show that prototype ‘EkstrakPro’ can extract different structure hydrology data correctly by using only one algorithm. Using sequential recognition algorithm can also further reduce the time required for extraction of information. The result of the research proves that information extraction can be solved using ontology approach