Exploring the Stability of IDF Term Weighting

TF IDF has been widely used as a term weighting schemes in today's information retrieval systems. However, computation time and cost have become major concerns for its application. This study investigated the similarities and differences between IDF distributions based on the global collection and on different samples and tested the stability of the IDF measure across collections. A more efficient algorithm based on random samples generated a good approximation to the IDF computed over the entire collection, but with less computation overhead. This practice may be particularly informative and helpful for analysis on large database or dynamic environment like the Web.