hermannkroll / entitycharacterization

EntityCharacterization

This repositoy contains our introduced BAFREC (BAlancing Frequency and Rarity in Entity Characterization) algorithm.

Getting Started

We can't deliver all necessary dependencies, because some are missing inside maven. This dependencies have to be downloaded manually and setup like described below.

Dependecies

ws4j

1. Download ws4j-1.0.1.jar from https://code.google.com/archive/p/ws4j/
2. Make sure its named "ws4j-1.0.1.jar"
3. Copy the file into lib/ws4j/ws4j/1.0.1/ws4j-1.0.1.jar

jawjaw

1. Download jawjaw-1.0.2.jar from https://code.google.com/archive/p/jawjaw/
2. Make sure its named "jawjaw-1.0.2.jar"
3. Copy the file into lib/jawjaw/jawjaw/jawjaw-1.0.2.jar

Virtuoso JDBC

(you don't need to setup a database containing the data)

1. Extract the Virtuoso JDBC driver like explained in http://docs.openlinksw.com/virtuoso/virtuosodriverpackaging/
2. Copy the jar inside lib/jdbc/virtjdbc/4.2/virtjdbc-4.2.jar

Cached database statistics

To prohibit setting up a DBpedia and LinkedMDB database, we deliver two cache files which contain all necessary statistics. The Virtuoso JDBC driver needs to be included, but no database connection is necessary. The connection is disabled as default.

Configuration

You can configure this software using the src/main/java/de/bs/tu/ifis/Config.java class.

You need to setup the following:

1. location of the ESBM benchmark data
2. location to save the results

Results

Bafrec's generated output for the ESBM benchmark can be found inside summ_bafrec.