Personalized Web-Tasking (PWT) systems automate ordinary and repetitive web interactions while exploiting personal context to deliver personalized features. Among the personal context of a user, social context is all information obtained from the relationships with other users, which is relevant to the user's personalized web-tasks. Current approaches exploit the information of social media, or the explicit input of the user, and use it as is. In addition to this, PWT systems also benefit by inferring social relationships through reasoning over such information and other sources of context. For example, a calendar application might record events the user shares with other people, or the sensors on mobile devices can be used to identify others nearby. This information can be exploited to improve the execution of PWT applications including its personalization and context-adaptive capabilities. In this paper, we present our ongoing research and implementation to enable PWT systems with capabilities to exploit social context dynamically: (i) our extension of the PWT Ontology, and (ii) our SmarterContext inference rules approach to select relevant social context for PWT.