From the Web Information Retrieval's project
In the navbar you can find all the link to the codes with their explanation.
There is the code used for the different functions created to classify the tweets, by the usage of the Google Knowledge Graph.
The Google Knowledge Graph provides to us different node depending on the query we submit, in particular the subject of each query is an entity, that can be a name of a VIP, a place, whatever! The list of nodes of an entity is called from the paper's authors lineage. A tweet may have different of those entities, that may be related together, so the assumption is that related entities could have common nodes.
In some cases we are not so lucky, because in our experiment we decided to simplify a little bit stuffs by no considering the context of a tweet, that can be web traffic in that day, events happened in that day, all the things that can influence users. Indeed...