Asukohapõhiste veebisündmuste algoritmid (Twitteri andmete näitel)
Date
2014
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Käesoleva magistritöö raames on katsetatud erinevate algoritmide jõudlust Twitteri andmete põhjal. Töö eesmärgiks on testida algoritmide pädevust suurte andmete protses-simisel (kaardil kuvamisel) ja püüda andmeid kiiresti lõppkasutajani viia nii, et nad oleks sobivalt filtreeritud ja klasterdatud. Andmeid on võimalik kaardil näha nii agregee-ritud kujul, kui ka agregeerimata. Veebirakendusega on võimalik otsida andmeid vabalt valitava raadiusega. Andmete klasterdamisel ja kauguste arvutamisel on katsetatud eri-nevaid algoritme ja võrreldud nende kiirusi. Andmebaasi päringute kiirendamiseks on katsetatud lisaks indekseerimisele ka riistvaralisi muudatusi. Andmeteks on Tartu Üli-kooli poolt pakutud Twitteri andmed, kus on üle 24 miljoni kirjet.
With this master's thesis the performance of different algorithms was tested using data from Twitter. The goal of this study was to test the performance of the algorithms when processing large-scale data (when displaying a map) and to try to get the data to the final user as quickly as possible with it being appropriately filtered and clustered. The data can be seen on the map in an aggregated state but also in a disaggregated state. With this web application it is also possible to search for a certain location with freely chosen radius in which only the points in a fixed radius are mapped. In order to cluster the points and cal-culate the distances, different algorythms were used and their speed compared. To quicken the database queries, changing hardware was also tried. The data used is Twitter data offered by The University of Tartu where there are over 24 million entities.
With this master's thesis the performance of different algorithms was tested using data from Twitter. The goal of this study was to test the performance of the algorithms when processing large-scale data (when displaying a map) and to try to get the data to the final user as quickly as possible with it being appropriately filtered and clustered. The data can be seen on the map in an aggregated state but also in a disaggregated state. With this web application it is also possible to search for a certain location with freely chosen radius in which only the points in a fixed radius are mapped. In order to cluster the points and cal-culate the distances, different algorythms were used and their speed compared. To quicken the database queries, changing hardware was also tried. The data used is Twitter data offered by The University of Tartu where there are over 24 million entities.