README.md

NAISC - Automated Linking Tool

'Naisc' means 'links' in Irish and is pronounced 'nashk'.

Installation

The latest build can be downloaded from GitHub Releases

The latest versions can be downloaded with

install.sh

Alternatively Naisc can be installed with Maven, to compile and run the system run the following:

install.sh

Alternatively you may download the compiled JARs from the release and place them at the following paths

• https://github.com/insight-centre/naisc/releases/download/Ubuntu-latest/naisc-core-1.1-jar-with-dependencies.jar → naisc-core/target/naisc-core-1.1-jar-with-dependencies.jar
• https://github.com/insight-centre/naisc/releases/download/Ubuntu-latest/naisc-meas-jar-with-dependencies.jar → naisc-meas/target/naisc-meas-jar-with-dependencies.jar

Meas - Meas Evaluation and Analysis Suite

For developing models and training there is a web application that can be built by the following

./meas.sh

The Web interface will be available at http://localhost:8080

Alternatively you may download use the release version with

java -jar naisc-meas-jar-with-dependencies-1.1.jar

Command line operation

Naisc can be operated from the command line with the following script

./naisc.sh left.rdf right.rdf -c config.json -o alignment.rdf

This will output the alignment using the configuration to alignment.rdf

Offline training can be created using the training script, the dataset should be available under datasets/

./train.sh dataset -c config.json

Command line options

For linking (naisc.sh)

Option       Description
------       -----------
-c <File>    The configuration to use
-f <File>    Dump features
-n <Double>  Negative Sampling rate (number of
               negative examples/positive example)
-q           Quiet (suppress output)

For training (train.sh)

Option       Description
------       -----------
-c <File>    The configuration to use
-f <File>    Dump features
-n <Double>  Negative Sampling rate (number of
               negative examples/positive example)
-q           Quiet (suppress output)

Basic configurations

The following basic configurations are available:

1. config/jaccard.json: A simple Jaccard based string similarity
2. config/string-match.json: Uses string similarity metrics only
3. config/auto.json: The general purpose linker

Documentation

Javadoc for Naisc is available at https://uld.pages.insight-centre.org/naisc

There is an overview of the tool available here

For extending Naisc with new services please see the guide here

There is a quick video introduction on YouTube: