Recent efforts to mainstream social media-based and citizen-led political deliberations to complement traditional government-led e-Participation, must among others address a number of technical challenges. One of these challenges is how to automatically filter and extract meaningful information or knowledge from streams of social media data contributed by citizens to inform agenda setting or serve as feedback for specific policy issues. A central resource in developing an effective information extraction system is the Lexicon of terms and relationship among terms in the domain of interest. This poster summarizes our work in developing a lexicon of public services names as the kernel of the knowledge extraction component of our Social Software Infrastructure for spontaneous participation. Our lexicon-based knowledge extraction process is scalable and could be easily extended to capture other information of interest from social media or related platforms.