The information available to users is overwhelming in today’s world. Therefore, it is essential to filter and convey only the essential information in a personalized fashion. We explored the automatic summarization of text as a means to address this problem. In addition, the current work explores two mechanisms: the shared attention and conceptual spaces in a an effort to extract abstract ideas from text and personalize them according to the users’ interests. The CNN_DM database was used as a source for both text and ground truth summarizations. User profiles were extracted from user generate commentaries in NYT, to provide insight into how individuals use abstraction. We utilized several recurrent neural networks with an attached attention mechanism. The results were comparable to the state of the art pointer generator network (0.145 f1 score). The shared attention RNN had an f1 score of 0.132. Moreover the Recurrent Neural Network equipped with a conceptual space mechanism scored 0.079 f1 on the same dataset. Summarization is the process of condensing the source text with loss of information and preservation of essential ideas. The existing methods of summarization, whether done by humans or automatic systems, create impersonal summarizations without the user profile in mind. In the current work we show that personalized summarization can be achieved by utilizing neural networks of cells equipped with attention mechanisms and by introducing semantic information via concept spaces. The models proposed here achieve similar performance as the state of the art while having user’s content as a guide to their interests.