Using Conditional Random Field in Named Entity Recognition for Crime Location Identification

Abstract

Electronic data or information comes in different forms, some are structured data and others unstructured data. The act of collecting such data is known as data mining. This paper will discuss the mining of crime data from electronic news sources in Malaysia, and how this data is further transformed to extract meaningful information from it. Furthermore, the paper will demonstrate how crime locations can be identified within the various news articles. This is significant because there are cases where a location name is mentioned in the news article but that is not the true crime location. To help achieve this, the system makes use of Named Entity Recognition (NER) algorithms. They are task with identifying locations in various sentences. To bring more accuracy to the work, the system will employ machine learning technique known as Conditional Random Field (CRF) to recognize if a sentence is referring to a crime location.