* Description

• Use no prior knowledge such as predicates and statistics information of the source endpoints -- up-to-date information of each source

• Contact each source with "ask" query in real time to determine the target source -- no predicates bound limitation

• Save the predicate information in cache for future use

• Join order: a heuristics-based cost estimation(based on a variation of the variable counting technique)

• Use exclusive group and bound join technology to minimize the number of requests sent to federation members

* source code:

FedX3.1

 http://www.fluidops.com/en/company/training/open_source

* execute:

example:

[fsearch@cl38 FedX]$ java -jar FedX3.1.jar -d bio2RDF.ttl -f STDOUT -q  "select  ?s1 ?o1 ?o2 where{ ?s1 <http://www.w3.org/1999/02/22-rdf-syntax-ns#type> <http://bio2rdf.org/omim_vocabulary:Gene>.?s1   <http://bio2rdf.org/omim_vocabulary:refers-to> ?o1. ?o1   <http://bio2rdf.org/omim_vocabulary:x-snomed> ?o2.
<http://bio2rdf.org/pharmgkb:PA446359> <http://bio2rdf.org/pharmgkb_vocabulary:x-snomedct> ?o2.}"

Result:

query1: return 5 results query2: return 40 results

query3: return 790 results

query4: return 9492 results

query5: return 111962 results

Query\time(ms)	time 1	time 2	time 3	time 4	time 5	average*
q1	11624	9469	10377	8978	9355	9545
q2	47152	43902	43569	42418	43393	43321
q3	113629	108622	109989	109297	109796	109426
q4	106218	108579	109989	107731	106713	107310
q5	792628	794677	806516	819745	812255	801519
q3s	109541	105442	109989	105650	106808	106972
q5s	27390	17186	11176	9099	7406	11216

*here we calculate the average by excluding the maximum in the five values.