161Management of Big Semantic Data
two main reasons. On the one hand, HDT-encoded data sets are smaller than
its counterparts in NTriples and it improves decompression performance.
On the other hand, HDT-FoQ generates its additional indexing structures
(see section “Querying HDT-Encoded Data sets: HDT-FoQ”) over the origi-
nal HDT encoding, whereas RDF3X rst needs parsing the data set and then
building their specic indices from scratch. Both features share an important
fact: the most expensive processing was already done in the server side and
HDT-encoded data sets are clearly better for machine consumption.
Exchange and post-processing times can be analyzed together and because
of it the total time than a consumer must wait until the data ...