Twitter Hash Tag Prediction Algorithm

1. Problem Definition Hash tag prediction is different from normal texts classification. Here we don’t know how many clusters we need to find. In addition, the tag set changes so frequently that it is almost impossible to effectively carry out classification or clustering, since a new tag would force us to establish a new class and a new classification rule. Our intuition is: if we can measure the correlation between various tweets as the mathematical metric we can treat the collected tweets as points in a high dimensional space, and construct a network by the latent space model. 2. Method 2.1 Theory An intuitive way to solve this problem is to use Euclidean distance between points as the measurement of their similarity. We developed our theory based on this distance. Since in a Euclidean Space, the distance is equivalent to the norm of a vector, we will focus our discussion on norms. Let be the standard bases (with unit norm) of a p-dimensional