As the e-commerce is gaining popularity various customer surveys of objects are currently accessible on the Internet. These surveys are frequently disordered, prompting challenges in knowledge discovery and object assessment. This article proposes an object feature positioning skeleton, which consequently recognizes the critical features of an object from online customer surveys. The critical object features are recognized focused around two perceptions: 1) they are normally commented extensively by customers and 2) customer suppositions on the critical feature significantly impact their general assessments on the object. Specifically, given the customer surveys of an item, we first extract the object feature by a shallow reliance parser and focus customer suppositions on these features through an opinion characterizer. We then create a probabilistic object feature positioning calculation to identify the criticalness of perspectives by at the same time considering feature recurrence and the impact of customer opinion given to every feature over their general opinion. The experimental results on 3 popular products demonstrate the effectiveness of our approach.