1. Build 

Simply run mvn package.

2. Perform cohort queries using COHANA.

2.1 Required documents:
    1 dataset: a csv file with "," delimiter (normally dumped from a database
        table)
    2 dimension file: a csv file "," delimiter. Each line of this file has two
        fields: the first field is the name of a column in the dataset, and the
        second field is a value of this column. Each distinct value of each
        column in the dataset shall appear in this dimension file once.
    3 table.yaml: a json file describing the schema of the dataset
    4 cube.yaml: a json file specifying the dimension and measure fileds
    5 runtime configure file: a yaml file specify the parameters for running query
        server. Currently, it is only required to specify the location of
        runtime directory (detailed in Step 2).

    We have provide an example for each of the three yaml documents in config
    directory. 

2.2 Compact the dataset

COOL evaluates cohort queries over a customized storage format. We have
developed a loader to convert the plain dataset in csv format into the desired
format. To convert, one can run the cool-loader.jar with the following 
five arguments:
    1 the table.yaml (the third required document)
    2 the dimension file (the second required document)
    3 the dataset file (the first required document)
    4 the output directory for the compacted dataset
    5 chunk size
For example, one can convert the dataset with the following command:
    java -jar cool-loader.jar "/path/to/table.yaml" "/path/to/dimension.csv"
    "/path/to/dataset.csv" "/path/to/output/directory" 65536