Context-Aware Multimedia Content Recommendations for Smartphone Users

According to Gartner, a world leading information technology research and advisory company, 57.6% of all mobile phones sold in the last quarter of 2013 were smartphones (Gartner, 2013). Unlike feature phones, smartphones are replacing our desktops as they increasingly become more powerful in terms of processing capability, network connectivity, and multimedia processing support (Flora, 2010; Ricci, 2011). This development indicates global penetration and acceptance of smartphones as the primary platform for information access and processing. As mobile users go about their daily activities, they continuously browse the Web, seeking interesting Webbased multimedia content to consume, and occasionally also uploading their personal content. However, these users encounter huge volume of available Web-based content, which often does not match their preferences. These preferences change as mobile users move from one place to another, performing different activities. Therefore, it is important to keep track and learn mobile user’s contexts in which they perform such activities. This contextual information can be used to filter and to deliver relevant and interesting multimedia content, thereby assisting users to overcome frustrations of selecting from overwhelming set of potential multimedia content choices. Consequently, users can focus more on important activities, minimizing distractions and time wasted while browsing Web-based media. Context-aware recommendation (CARS) has become a major focus of researchers addressing information overload related problems (Adomavicius et. al., 2005). This process can suggest multimedia content to mobile users by considering user’s preferences and contexts in which such preferences are expressed. Many solutions of this kind, however, are limited to using static and explicit contextual information. For example, they rely on asking users to explicitly provide their current contexts in order to provide them with relevant items.. In fact, traditional recommendation systems do not consider context as an important factor in the recommendation process because they assume that user preferences are static. We define context-aware mobile multimedia recommendation (CAMR) as a special type of context-aware recommendations that uses mobile user’s contexts to compute media recommendations. CAMR is grounded in existing solutions and technologies. First, rapid development in the field of mobile and telecommunication networks has enabled ubiquitous communications whereby smartphone users can connect to the Web anywhere, anytime. With this development, mobile users can access multimedia content such as news, music, videos, etc. at their convenience. Second, mobile devices now come with cheap, built-in sensors, enabling ubiquitous context sensing (Kwapisz et. al., 2010). Sensors such as thermometers for sensing environment temperature, accelerometer for sensing movement, and GPS sensor for sensing location information, etc. now ship with smartphones. Third, context-awareness has enabled the ability to deliver personalized information based on user’s contextual situations. Information such as location, activity, time, weather, etc. can now be obtained readily in real-time from smartphones. Fourth, traditional information recommendation systems have matured, and are helping users to find relevant information (Adomavicius et. al., 2005). Thus these existing solutions can be explored to realize context-aware mobile multimedia recommendations. Therefore, CAMR builds on these core solutions, using mobile user’s preferences to suggest useful and interesting multimedia content, tailored to users contextual situations. Abayomi M. Otebolaku INESC TEC, Porto, Portugal