In order to realize effective infotainment systems for vehicles, we need to have context-aware applications that use the latest (live) information for an enhanced user experience. Such up-to-date information is now abundantly available on the Internet, due to the explosive growth of the Web 2.0. In earlier times, it was difficult and expensive for vehicles to connect to the Internet. Recent advances in vehicular ad-hoc networks (VANETs) have enabled vehicles to connect to the Internet through road side infrastructures, with little to no additional cost. However, there are several problems with directly using Internet data in a vehicle, such as (1) Internet data sources have their own interfaces, which keep changing over time, needing frequent application updates, (2) information provided by multiple data sources needs to be preprocessed and fused before use, and (3) vehicles employ propriety platforms for infotainment systems, which makes an application update even more cumbersome. Furthermore, accessing multiple Internet sources may cause unnecessary overhead over the VANET bandwidth. In this thesis, we propose a cloud-based middleware framework for vehicular infotainment application development. The proposed framework follows service oriented architecture in which data filtering and fusion functionalities are delegated to the cloud. Data filtering and fusion reduce the data flow over VANET. Furthermore, because most the the processing is done on the cloud, the client becomes lightweight and loosely coupled with Internet resources and underlying platforms. We also propose a class based fusion method to combine information from multiple sources. The efficacy of the proposed framework is validated by developing an enhanced navigation (eDirection) application for the vehicle, as well as three infotainment applications: context-aware music, news, and weather.