Traffic congestion is a significant problem for many developing countries. Among a variety of methods proposed to address this problem, extraction of road traffic conditions from microblogging platforms for dissemination to road users has gained popularity. Coupled with the use of Internet of Things (IoT) devices, up-to-date traffic condition reports from road users and road traffic authorities, can be effectively reconciled and communicated to the intended audience. However, the noisy and unstructured nature of Twitter texts causes a decline in the performance of conventional Named Entity Recognition (NER) techniques. NER techniques have traditionally been used to extract location information. In order to extract the corresponding traffic state information for each location, contextual information become important which are not implemented in conventional NER approaches. In this paper, a rule-based NER technique, which considers contextual clues, is proposed for the extraction of location and traffic state entities. The proposed approach has achieved significant improvement to the F1 score, 88.96 and 81.32, for location entity extraction from formal and informal sources, respectively.