Track Participants
Rutgers: expanded query with n-gram patterns that allow insertion of 0-2 characters (e.g.., “cat” searched as cat, ..at, c..t, ca..)
ANU: expanded query with corruptions found in a sample of the corrupted text & used span matching (allows mismatch)
GMU: expanded query using 20 most frequent 4-grams occurring in top 10 docs; exact match required
CLARITECH: used stochastic methods to correct text within sentence boundaries; then used normal CLARIT retrieval
ETH: computed an estimated frequency of word occurrence by allowing all initial substrings of words that are “sufficiently close” to query words to contribute to the document similarity: in effect this allows a vector of words to occupy a given position within a document