Abstract: A user can designate a dataset and one or more n-grams in a graphical user interface displayed by a client computing device (“the client”). The client can send the designations to a server computing device (“the server”). The server can execute a query over the dataset based on the n-grams, and return statistics indicating occurrences of the n-grams in the dataset. The client can generate a graphical representation corresponding to the statistics. The user can select a portion of the graphical representation, which the client sends to the server. The server can respond with excerpts of documents corresponding to the selected portion and other n-grams within the documents. The client can then display a list of the excerpts and a list of other n-grams. The user can then perform further actions on the lists such as viewing the underlying documents or adding n-grams to the search query.