top

Applications (in English only)

One of the most notable results of the LT4eL project is a series of language technology tools, intended to be employed in an e-Learning context. The following are available:

Table of Contents

  • Keyword Extractor - A software assistant that will support authors, tutors and other content managers in by semi-automatically extracting appropriate keywords from a learning object. (WP2)
  • Glossary Candidate Detector - A tool to help authors create glossaries by presenting definitions from their learning objects. Authors can accept, reject, or modify the definitory contexts which are presented to them by the glossary candidate detector. The accepted definitions will be written to a glossary which accompanies a learning object and which can be edited by the author at any time. (WP2)
  • Grammars - Various grammars were developed on the basis of the training corpus
  • Ontology - The LT4eL ontology is a knowledge hierarchical repository of concepts in the domain of Information Technology for End Users (or in other words: Computer Science for Non-Computer Scientists). DOLCE Ultralite is used as upper ontology. The ontology contains 1002 domain concepts, 169 concepts from OntoWordNet and 105 concepts from DOLCE Ultralite. It also contains more than 100 object properties.  The ontology is an OWL file in XML representation. (WP3)
  • Lexicons
  • ILIAS - ILIAS is a powerful web-based learning management system that allows you to easily manage learning resources in an integrated system. The various language technology tools created in the LT4eL project are integrated into this system. (WP4)
(top)

Keyword Extractor

Description
The Keyword Extractor (KWE) is a software assistant that supports authors, tutors and other content managers by proposing appropriate keywords for a learning object. Users can select and add keywords and include them into the metadata for this learning object.

Prerequisites
To run the programme, a Java Runtime Environment is needed. The learning objects which are processed must be linguistically annotated and conform the LT4ELAna format.

Installation

The Keyword Extractor is part of the LTserver and cannot be used separately. To play with the tool, there is a command line interface. The procedure is that a user downloads the entire LTserver package, compiles it, but does not deploy a web application, but uses lt4el-cmd.jar package instead.

A description of the command-line interface can be found here:

In fact there are two possibilities for processing LOs:

  1. input file in LT4ALAna format
  2. input is text/html/doc/pdf and LPC is setup correctly for given language (then all conversions, tokenization, segmentation and tagging  is done inside LTserver). LPC can be used currently for Czech, English, Polish and Romanian.

Demo Videos
We have two demo videos available showing the keyword extractor and keyword search:

Download

(top)

Glossary Candidate Detector

Description
The Glossary Candidate Detector (GCD) is a tool to help authors create glossaries by presenting definitions from their learning objects. Authors can accept, reject, or modify the definitory contexts which are presented to them by the glossary candidate detector. The accepted definitions will be written to a glossary which accompanies a learning object and which can be edited by the author at any time.

Prerequisites
To run the programme, a Java Runtime Environment is needed. The learning objects which are processed must be linguistically annotated and conform the LT4ELAna format.

Installation

The Glossary Candidate Detector is part of the LTserver and cannot be used separately. To play with the tool, there is a command line interface. The procedure is that a user downloads the entire LTserver package, compiles it, but does not deploy a web application, but uses lt4el-cmd.jar package instead.

A description of the command-line interface can be found here:

In fact there are two possibilities for processing LOs:

  1. input file in LT4ALAna format
  2. input is text/html/doc/pdf and LPC is setup correctly for given language (then all conversions, tokenization, segmentation and tagging  is done inside LTserver). LPC can be used currently for Czech, English, Polish and Romanian.

Demo Videos
We have two demo videos available showing the glossary candidate extractor and the definition search:

Download

(top)

Grammars

Description
The grammars were developed on the basis of the training corpus in many iterations, where in each iteration the grammar was improved and/or the results were automatically evaluated quantitatively and manually evaluated qualitatively. The XML transducer lxtransduce is used to match the grammar against files in the LT4ELAna format.

Prerequisite: lxtransduce
The XML transducer lxtransduce developed by the language technology group at the University of Edinburgh is used to match the grammar against files in the LT4ELAna format. Lxtransduce has been developed especially for use in NLP applications. It supplies a format for the development of grammars which are matched against either pure text or XML documents. The grammars must be XML documents which conform to a DTD. In each grammar, there is one ‘main’ rule which calls other rules by referring to them. The XPath-based sub-rules are matched against elements in the input document. When a match is found, a corresponding rewrite is done.

Ontology

Description
The LT4eL ontology is a knowledge hierarchical repository of concepts in the domain of Information Technology for End Users (or in other words: Computer Science for Non-Computer Scientists). We have extracted the first set of domain concepts from learning materials. Then we have extended it with new concepts from other sources such as WordNet, Wikipedia, etc. Then we have connected the domain ontology with an upper-ontology in order to complete the hierarchy of concepts, but also to inherit all the information from the upper ontology, including the relations (properties). We have used DOLCE Ultralite as upper ontology.

The connection between the domain ontology and the upper ontology was established with the help of OntoWordNet. The ontology is an hierarchy in which the concepts are ordered on the basis of the relation superconcept-subconcept. Additionally, there are some other relations, such as part_of, used_for, etc. The ontology can be used for concept browsing and semantic search. The concept browsing functionality allows the user to view the concepts as a relation-based map. The semantic search functionality requires documents annotated with concepts and language specific lexicons, mapped to the ontology. In such cases the ontology becomes an invaluable source for cross-lingual searches.

 

The current ontology contains 1002 domain concepts, 169 concepts from OntoWordNet and 105 concepts from DOLCE Ultralite. It also contains more than 100 object properties. Each concept has a definition in English. Also all the concepts and properties are equipped with terms in several languages: Bulgarian, Czech, Dutch, English, German, Maltese, Polish, Portuguese, and Romanian. These terms can be used for representation of the ontology in the corresponding language and for annotation of texts with ontology information. The second task requires additionally development of appropriate grammars.

The ontology is an OWL file in XML representation. It can be viewed via any XML editor (e.g. CLaRK system). For graphical observation the user can rely on the Protege System.

Demo Videos
We have one demo video available showing the ontology:

Download

  • ontology Ontology (Zipped XML/OWL file, unzip and open in an ontology browser like protege)
(top)

Lexicons

Download

(top)

ILIAS

Description
ILIAS is a powerful web-based learning management system that allows you to easily manage learning resources in an integrated system. The software was developed to reduce the costs of using new media in education and further training and to ensure the maximum level of customer influence in the implementation of the software. ILIAS is published under the General Public Licence and free of charge. The LT4eL tools have been integrated into the ILIAS open source learning management system.

Prerequisite
Before installing ILIAS, you need the LT4eL web service server running. The integration report explains how to install the LT4eL server and ILIAS. Both can be downloaded from our sourceforce page.

Download & Installation of ILIAS
You can download the prototype implementation of ILIAS from our sourceforce page. Click on download and select the package ilias-lt4el.m30.tar.gz. Please use the tar command to extract the package to a folder within your web space: tar –xzvf ilias-lt4el.m30.tar.gz

The ILIAS project provides installation instructions for different kinds of operating systems at. You can use one of these or the instructions in the docs/INSTALL file for installing the prototype as well. After the usual ILIAS setup, login into ILIAS and open the ILIAS administration: http://<your.domain>/<your_path>/ilias.php?baseClass=ilAdministrationGUI

Enter the URL of the LT SOAP server and the paths of the lexicons and the ontology in the administration panel. Select Use Processing Chain for all languages that provide a configured linguistic processing chain. If you want to use the tools for other languages, you need to put pre-annotated files into the directory specified under LO Source data directory. After entering all values, press the Save button at the bottom of the administration screen. After that click on Create additional fields and tables and Load Ontology and Lexicons into DB. Now all functionalities described in the LT4eL WP4 D4.1 User Documentation should be available.