Abstract: A database indexing method and apparatus which includes an index structure for indexing a plurality of objects that are logically divided into fine slices of 8,000 records each, and the fine slices are grouped into coarse slices of 4,000 fine slices each. The indexes include fine and coarse keys, each of which corresponds to a particular data value and a particular fine or coarse slice. Bit vectors are used to determine which of the fine slices or objects identified by the fine slices have particular attributes, and these bit vectors can be stored in compressed format. The index structure can be used to index data where no data records exist. A word indexing technique for crossing page boundaries within a document is also disclosed, as well as a technique for enumerating and displaying virtual folders and their contents.

Abstract: A database having fixed length records stored together in record number order and an index structure for the database. The index structure comprises a separate index for each searchable field of the records. For purposes of indexing, the records are logically divided into fine slices of 8,000 records each, and the fine slices are grouped into coarse slices of 4,000 fine slices each. The indexes include fine and coarse keys, each of which corresponds to a particular data value and a particular fine or coarse slice. Associated with each key is a link that is used to determine which records contain the data value. For the fine keys, the link includes a pointer to a bit vector that has a single bit for each of the records within the fine slice associated with the key. For the coarse keys, the link includes a pointer to a bit vector that has a single bit for each of the fine slices contained in the coarse slice.

Abstract: A database having fixed length records stored together in record number order and an index structure for the database. The index structure comprises a separate index for each searchable field of the records. For purposes of indexing, the records are logically divided into fine slices of 8,000 records each, and the fine slices are grouped into coarse slices of 4,000 fine slices each. The indexes include fine and coarse keys, each of which corresponds to a particular data value and a particular fine or coarse slice. Associated with each key is a link that is used to determine which records contain the data value. For the fine keys, the link includes a pointer to a bit vector that has a single bit for each of the records within the fine slice associated with the key. For the coarse keys, the link includes a pointer to a bit vector that has a single bit for each of the fine slices contained in the coarse slice.