It's my pleasure to help you solve this problem.
That is quite easy with our UDF, which is user-defined functions. The only thing you need to do is to write a python parser and import it as a UDF into Xcalar. For your case, you need yield a dictionary, probably called "row", while reading your each one of your files. Then the key-value pair will be your keywords associated with your numbers.
When you import the dataset, kindly select the option "parse data with UDF" and choose your parser. Then you are all set.
Also you can click "refresh preview" to see if it is the exact way you want to parse your data. Once you are satisfied with the preview, you can import the entire dataset using your UDF and enjoy the flexibility and power provided by Xcalar.
Hope this helps.