The world wide web including social sites, blogs, forums etc. generate a huge amount of data in the form of views, emotions, opinions and arguments regarding different products, social activities, brands and politics. This has prompted increasing interests in analytical domain. Thus, all these comments have a great influence on readers, vendors politicians. In political domain, political parties and other experts make use of facebook, twitter and blogs to communicate with the people and to check voice of public. Hence, an appropriate model which gives correct analysis and results about peoples' sentiments related to political parties and political diplomats is built. Data is captured from social media sites and pre-processing is carried out. Word2vec model is used for word embedding process. For final evaluation neural network is used. After detailed study and comparison efficient neural network model is chosen and positive/negative sentiments are identified. Experimental evaluation signifies the effectiveness of the proposed model. This system is also helpful for understanding people's response to particular political decision which will help in better decision making during elections and political campaigns. © 2018 IEEE.