A fundamental difficulty in recognizing human activities is obtaining the labeled data needed to learn models of those activities. Given emerging sensor technology,however, it is possible to view activity data as a stream of natural language terms. Activity models are then mappings from such terms to activity names,and may be extracted from text corpora such as the web.

Question answer session

a)Can we use this method without internet ?

Yes, you can. But in this paper, the authors generate the way to think about big useful of internet because the information on the internet is very useful to check our data set.

b)Why do we use unsupervised learning ?

From the web, for example a google, we mark the label for activities which do not have label. This is called unsupervised. We have to find the objects related to your acts.