In modern knowledge explosion society, Humanspsila ability to find the information they desire grows more slowly than the rate at which the new information becomes available. Recommendation system can reduce information overload and provide appropriate content to targeted user. This paper proposes a new recommending technique based on mining the userpsilas reading behavior, computing the user profile through 5W1H scheme. By comparing the user profile and the information representation, we can provide the customized content in suitable time to satisfy the userpsilas personalized requirement. For performance evaluation, we implement the technology to make experiments on real information and people. The results show that our technology and approach can acquire the userpsilas requirement, and recommend suitable information they really need.