considering you have the following data { "user": { "userlocation": "California, Santa Clara", "id": 222222, "name": "Hortonworks", "screenname": "hortonworks", "geoenabled": true }, "tweetmessage": "Learn more about #Spark in #HDP 2.3 with @Hortonworks founder @acmurthy in this video overview http://bit.ly/1gOyr9w #hadoop", "createddate": "2015-07-24T16:30:33", "geolocation": null} your schema would look like so with JsonSerDe CREATE EXTERNAL TABLE tweets ( createddate string, geolocation string, tweetmessage string, `user` struct<geoenabled:boolean, id:int, name:string, screenname:string, userlocation:string>) ROW FORMAT SERDE 'org.apache.hive.hcatalog.data.JsonSerDe' LOCATION '/user/root/'; SELECT DISTINCT tweetmessage, user.name, createddate FROM tweets WHERE user.name = 'Hortonworks' ORDER BY createddate; ... View more

You did not include the python interpreter line in your python script and it has difficulty understanding its python. For what you're trying to achieve, you can skip streaming and just use Pig built-in filter function. It will perform better than streaming. http://pig.apache.org/docs/r0.15.0/ SSN_NAME = load 'students.txt' using PigStorage() as (ssn:long, name:chararray); /* do a left outer join of SSN with SSN_Name */ X = JOIN SSN by ssn LEFT OUTER, SSN_NAME by ssn; /* only keep those ssn's for which there is no name */ Y = filter X by IsEmpty(SSN_NAME); ... View more