Note that in topics.txt :-
-Topics numbered between 001 and 021 were used in Phase 1 experiments
-Topics numbered between 022 and 062 were used in Phase 2 experiments
-The qrels_MAP.txt file consists of binary relevance judgments (0 - nonrelevant; 1 -relevant)
-The qrels_nDCG.txt file consists of graded relevance judgments (0 - nonrelevant; 1 - relevant; 2 - highly relevant)
One could request for or buy the collection of SMS conversations from the Linguistic Data Consortium (LDC), UPENN (https://www.ldc.upenn.edu/)