izpis_h1_title_alt

Učinkovito poizvedovanje po povezanih podatkih s porazdelitvijo obremenitve
ID Robas, Jan (Author), ID Lavbič, Dejan (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (1,56 MB)
MD5: 65799A589FEC9CBA488EF9E093A374FA
PID: 20.500.12556/rul/0a522854-58f3-4d9f-9db9-07803876ccb4

Abstract
Na spletu so podatki predstavljeni na različne načine v različnih oblikah in zato med sabo niso združljivi. Isti problem se pojavi tudi pri spletnih programskih vmesnikih, saj je za povezovanje z zunanjimi podatki ponavadi potrebno narediti odjemalca, prilagojenega za dane podatke. Ta problem rešujejo povezani podatki. Uporabo povezanih podatkov zavirajo problemi s performancami poizvedovanja in posledično z dostopnostjo oddaljenih točk SPARQL. Z delci vzorcev trojčkov lahko pri poizvedovanju del obremenitve prestavimo na odjemalca, vendar pri tem prenesemo večjo količino podatkov. Obstoječa razširitev AMF zmanjša število zahtevkov in s tem posledično tudi količino podatkov pri določenih poizvedbah, medtem ko pri ostalih poizvedbah količino podatkov poveča. V tem delu predstavljamo našo razširitev, s katero smo poskusili zmanjšati tako število zahtevkov kot količino prenesenih podatkov tako, da smo v metapodatke dodali Bloomov filter, ki vsebuje trojčke, povezane z rezultati na prikazani strani delca vzorca trojčka. Našo razširitev smo primerjali z osnovno različico in z različico AMF ter pri našem testu dosegli dobre rezultate. Prispevali smo tudi popravek razširitve AMF, ki je že vključen v uradni repozitorij, in za konec razvili preprost grafični vmesnik, ki omogoča sestavljanje poizvedb SPARQL in njihovo izvajanje z uporabo naše razširitve.

Language:Slovenian
Keywords:povezani podatki, SPARQL, delci vzorca trojčkov, procesiranje na odjemalcu, spletne tehnologije
Work type:Master's thesis/paper
Organization:FRI - Faculty of Computer and Information Science
Year:2016
PID:20.500.12556/RUL-87033 This link opens in a new window
Publication date in RUL:16.11.2016
Views:1148
Downloads:439
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Efficient querying of Linked Data by distributing workload
Abstract:
Online data is presented in different ways and in various forms which are not mutually compatible. This problem is also present in Web APIs, because we usually have to implement a specialised client, suited for the kind of data the Web service is providing. This problem is solved with Linked Data. The problem with Linked Data is the query performance and the availability of remote SPARQL endpoints. With Triple Pattern Fragments we can execute SPARQL queries by transferring some workload to the client, but in contrast we have to transfer more data. The existing AMF extension reduces the amount of HTTP requests and consequently the amount of transferred data on some queries, while increasing the amount of transferred data with others. In this thesis we present our extension, where we try to lower the amount of HTTP requests and the amount of transferred data by extending the metadata with a Bloom filter, containing data, linked with triples on the current page of the Triple Pattern Fragment. We have compared our extension with the AMF extension and achieved encouraging results. We have also proposed a fix for the AMF extension, which is already included in the official repository. Finally, we have developed a simple graphical user interface that enables composition of SPARQL queries and their execution using our extension.

Keywords:Linked Data, SPARQL, triple pattern fragments, client-side processing, Web technologies

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back