The technology for changing the nodes in an information retrieving system
using a computer. When information items are registered by allocating to
n nodes, steps are used to extract index information as a set of pairs of
index keys of information items and addresses of information items,
divide the index information into m (m>n) buckets and produce a
partial inverted file to be closed within each of the buckets. Here, m
and n are respectively integers of 1 (one) or above. When the allocation
of the search-targeted ranges to the nodes is altered, the allocation to
the buckets to each of the nodes is changed, and the partial inverted
file of each bucket and the inverted file of the existing indexes are
merged to produce new indexes, so that the indexes can be produced and
updated with high speed.