Even though it is an old post, I would put the answer here for future reference.

Basically we need another data set to contain the split enumeration lists, because a single row in the original data set such as

1    key    |value1|value2|value4|

can become multiple rows

1    key    |value1|
1    key    |value2|
1    key    |value4|

Using a scripted data set should be quite straightforward. Here is a short tutorial for creating such a data set.
http://developer.actuate.com/be/documentation/ihub3/help/adg/adg65/index.html#page/ADG/ProgramDataAccess.1.3.html

For our purpose here, we will use a global object "data" to store the processed data. We only need to make slight changes to the scripted data set so that it reads data from the global "data" object in the "fetch" method. Of course we will have to first populate the global "data" object with the original Jazz data set, in the "onFetch" method. Here are the codes of the "ofFetch" method.

id = row["WI_ID"];
type = row["WI_TYPE"];
key = row["MEDIUM_STRING_KEY"];
value = row["MEDIUM_STRING_VAL"];
if (typeof(value) == "string") {
    valueSet=value.split("|");
    for (var i=1; i< valueSet.length-1; i++) {
        var newData = {};
        newData.id = id;
        newData.type = type;
        newData.key = key;
        newData.value = valueSet[i];
        data.push(newData);
    }
}

(Remember to skip the first and last items of the valueSet[] arrays since they are empty)
This is how it looks like with the original Jazz data set on top and the scripted data set at the bottom.

Hi Rodolfo,
I used this video to create BIRT report with enumeration: http://www.youtube.com/watch?v=8q29We9QoHE . You can also see the similar topic created by me: https://jazz.net/forum/questions/113221/using-custom-attribute-as-a-parameter-in-birt.
Let me know if it helps for you.

Regards,
Krzysztof Kazmierczyk