diff --git a/live/live.default.ini b/live/live.default.ini index 3cd0e05fe7..60cde02774 100644 --- a/live/live.default.ini +++ b/live/live.default.ini @@ -12,9 +12,6 @@ uploaded_dump_date = 2012-04-01T15:00:00Z ; The place where application temporary files will be written (not public files) working_directory = /path/to/live/data/ -; Default extraction processing threads -ProcessingThreads = 1 - ;annotations are created ;works with SimpleDumpdestination and LiveUpdateDestination generateOWLAxiomAnnotations = true @@ -26,6 +23,10 @@ validateExtractors = false logpath = log/ ;rdfapi_include_dir = api/ + +; Default extraction processing threads +ProcessingThreads = 1 + ;********************* ; OAI Configuration ;********************* @@ -40,7 +41,6 @@ mappingsOAIUri = http://mappings.dbpedia.org/index.php/Special:OAIRepository mappingsOaiPrefix = oai:en.wikipedia.org:enwiki: mappingsBaseWikiUri = http://mappings.dbpedia.org/wiki/ - ;********************* ; OPTIONS FOR LANGUAGE ;********************* @@ -61,83 +61,28 @@ db_meta_ns = http://dbpedia.org/meta/ ;example for german, might still be changed ;dbpedia_ns = http://de.dbpedia.org/ - - - -;********************* -; OPTIONS FOR EXTRACTION FROM WIKIPEDIA DUMP -;********************* -wikiDB.host = localhost -wikiDB.user = root -wikiDB.password = password -wikiDB.refix = dbpedia_ - - -;for live extraction all db should be turned off -; except LiveMappingBased -LiveMappingBased.useTemplateDb=true -Geo.batchextraction = false -Geo.usedb = false -Persondata.usedb = false +graphURI = http://live.nl.dbpedia.org ;****************************************** ;Below configuration for the live extraction ;****************************************** -;when using OWLAxiomAnnotations and LiveUpdatedestinations: -;There are triples that are static: -;YAGO, WORDNET, UMBEL, OPENCYC, OWL:SAMEAS -;these will not be deleted -;also all extractors that are in status KEEP -;these are defined in rules in array metadata -;turn showconfig on to see those -;in current dbpedia there are string from other language -;editions these are treated separately -;please check below if all properties are set correctly for those -;then turn this option to true -;Note: I'm quite sure they are correct, but even I can make mistakes -iconfirmthaticheckedthestringsbelow = true -stringPredicateWithForeignlanguages[] = http://www.w3.org/2000/01/rdf-schema#label -stringPredicateWithForeignlanguages[] = http://www.w3.org/2000/01/rdf-schema#comment -stringPredicateWithForeignlanguages[] = http://dbpedia.org/property/abstract - -;addidtional vocabulary for static triples see above -;most of them are tackle by AlwaysFilterExtractor -;the following are just examples of how to add more -;in case new datasets are loaded -;example 1: predicateFilter[] = http://www.w3.org/2002/07/owl#sameAs -;example 2: objectFilter[] = http://dbpedia.org/class/yago/ -;separate by space " " -;example 3: predicateObjectFilter[] = http://www.w3.org/1999/02/22-rdf-syntax-ns#type http://umbel.org/umbel/ - - ;show the configuration of extractors at startup ;delays start for 5 seconds showconfig = false ;momentarily virtuoso only ;this is the configuration to load triples into the store -Store.type = virtuoso -Store.dsn = jdbc:virtuoso://localhost:1111 -Store.user = dba -Store.pw = dba -Store.SPARULdialect = virtuoso +store.class = virtuoso.jdbc4.Driver +store.dsn = jdbc:virtuoso://localhost:1177/charset=UTF-8/log_enable=2 +store.user = dba +store.pw = virtPass -;a sparqlendpoint is needed to create the diff -;sparqlendpoint = http://139.18.2.37:8890/sparql -;sparqlendpoint = http://dbpedia2.openlinksw.com:8895 -;sparqlendpoint = http://dbpedia.org/sparql -sparqlendpoint = http://localhost:8890/sparql -Sparql.use = odbc -graphURI = http://dbpedia.org -annotationGraphURI = http://dbpedia.org/meta -LiveUpdateDestination.strategy = primary -LiveUpdateDestination.useHashForOptimization = true +cache.class = com.mysql.jdbc.Driver +cache.dsn = jdbc:mysql://localhost/dbpedia_live_cache +cache.user = root +cache.pw = mysqlPass -;Blank node identifier generation -;Controls the prefix of blank nodes (the id generator is currently fixed) -;blankNodeIdPrefix = bn -;blankNodeIdFile = idStatus.dat ;dryRun doesn't update the store, but instead prints out the sparul dryRun=false @@ -170,89 +115,12 @@ fastFileHandling = true noglob = false -;**********EXTRACTOR CONFIGURATION********* -;**********ONLY AFFECTS LIVE********* -article.ActiveAbstract=ACTIVE - -article.Label=ACTIVE -article.ArticleCategories=ACTIVE -article.ExternalLinks=ACTIVE -article.Disambiguation=ACTIVE -article.MetaInformation=ACTIVE -article.Persondata=ACTIVE -article.Geo=ACTIVE -article.Wikipage=ACTIVE -article.Homepage=ACTIVE -article.LiveMappingBased=ACTIVE -article.Infobox=ACTIVE - -article.PageLinks=PURGE - -article.Abstract=KEEP -article.AlwaysFilter=KEEP -article.Image=KEEP -article.MappingBased=PURGE -article.InstanceType=PURGE -article.WordnetLink=KEEP - -category.Label=ACTIVE -category.SkosCategories=ACTIVE -category.MetaInformation=ACTIVE - - -redirect.Redirect=ACTIVE -redirect.MetaInformation=ACTIVE - -;***********DEBUG*************** -; the value false means no debugging -;does not delete the current record, so you can test -;the same page over and over -debug_loop = false -;run only once then exit -debug_die_after_one = false -debug_keep_remaining_triples_from_store = false - -debug_turn_off_insert = false; -debug_run_tests = false; - -;debug_do_not_load_into_virtuoso = false -;debug_keep_untouched_triples = false -;log_write_update_log = true -;log_write_added_deleted_triples = false - -;updatelog = files/updatelog/ -;rdfapi_include_dir = api/ - - - ;********************************** ;**OPTIONS FOR PUBLISHING UPDATES** ;********************************** osmReplicationConfigPath=./live - tmpPath=/tmp/lgd - -osmDb_hostName=localhost -osmDb_dataBaseName=lgd_test -osmDb_userName=postgres -osmDb_passWord=postgres - -rdfStore_hostName=localhost -rdfStore_userName=dba -rdfStore_passWord=dba -rdfStore_graphName=http://test.org - -tagMappings=TagMappings.xml - -entityFilter=LiveEntityFilter.txt.dist -tagFilter=LiveTagFilter.txt.dist - -publishDiffRepoPath=/home/mohamed/LeipzigUniversity/dbpedia_publish - sleepInterval=60 - -osmosisPath=/opt/osmosis/0.35.1/bin - sequenceNumber=1 ;*********************************** @@ -260,4 +128,4 @@ sequenceNumber=1 ;*********************************** ; OPTIONS FOR PUBLISHING STATISTICS* ;*********************************** -statisticsFilePath=/home/mohamed/publishdata/instancesstats.txt \ No newline at end of file +statisticsFilePath=/home/mohamed/publishdata/instancesstats.txt diff --git a/live/live.default.xml b/live/live.default.xml index c524f8a055..2d63f1bc45 100644 --- a/live/live.default.xml +++ b/live/live.default.xml @@ -1,7 +1,5 @@ - false - 10