In contrast to pictures, linguistics learning on 3D purpose clouds employing a deep network is difficult thanks to the naturally unordered organization. Among the previous works being done, PointNet has attained promising results by directly learning on purpose sets. However, it doesn't take full benefits of a point's native neighborhood that contains fine-grained structural info that seems to be useful for higher linguistics learning.