Marvin - semantic text annotator

Abstract

Marvin is a semantic text annotation tool that uses various external sources to annotate inputed text. Marvin text annotator can be also used as a java library

Installation

Marvin semantic text annotatior is a java program that can be also used as a java library for other application. This means it is an .jar file, which contains all the resources inside it. However, for use of Wordnet, Wordnet has to be installed from https://wordnet.princeton.edu/wordnet/download/current-version/

After the installation it is necessary to configure Wordnet path in file_properties.xml file. Tag that currently states following:

<param name="dictionary_path" value="C:\Program Files (x86)\WordNet\2.1\dict"/>

has to be changed with the correct path within the machine where Wordnet is installed.

Before running, you will also need to set the configuration of what resuources you would like to use in settings.cfg file. This file should be supplied with information on resources (what should be used, host and port of MetaMap if used, DBPedia SPARQL gateway if DBPedia is used, location of SKOS files and some information about your system and location for provenance)

There are no other requirements for the installation.

Running Marvin semantic text annotator

In order to run Marvin semantic annotator you can type in command line

java -jar Marvin.jar "Sentence to be semantically annotated."

Running Marvin as a library

How to run marvin as a library would be best described in a code snippet bellow. You need just to create object of MarvinSemAnnotator and invoke method "annotate" with the text to be annotated.

Marvin returns a linked list of Word objects and each Word object contains a LinkedList of annotations in list called wordmeanings.

public void Annotate(String valueToParse){
	MarvinSemAnnotator marvin = new MarvinSemAnnotator();
	LinkedList<Word> words = marvin.annotate(valueToParse);
	for(Word w:words){
		for(int s = 0;s<words.get(p).wordmeanings.size();s++){
			String source =  words.get(p).wordmeanings.get(s).Source;
			String ID =  words.get(p).wordmeanings.get(s).id;
			String URL = words.get(p).wordmeanings.get(s).URL;
			int startAt = words.get(p).wordmeanings.get(s).startAt;
			int endAt = words.get(p).wordmeanings.get(s).endAt;
			String appearing_word = words.get(p).wordmeanings.get(s).appearingWord;
			String Description = words.get(p).wordmeanings.get(s).Description;
			String Location = words.get(p).wordmeanings.get(s).Location;
			String EnvironmentDesc = words.get(p).wordmeanings.get(s).EnvironmentDesc;
			String AgentName = words.get(p).wordmeanings.get(s).AgentName;
			String AgentVersion = words.get(p).wordmeanings.get(s).AgentVersion;
		}
	}
}

Methodology

Methodology and technical solutions used in Marvin are explained in the following report: http://arxiv.org/ftp/arxiv/papers/1602/1602.00515.pdf

Reference

If you used Marvin, please cite:

Marvin: Semantic annotation using multiple knowledge sources, Nikola Milosevic, 2016, arXiv preprint arXiv:1602.00515 (http://arxiv.org/ftp/arxiv/papers/1602/1602.00515.pdf)

Credits

Developed by Nikola Milosevic - nikola.milosevic[at nospam] manchester.ac.uk

Feel free to contact me for any kind of information or queries regarding the project.

nikolamilosevic86 / Marvin