Limbă Limbă
Afişare conţinut web Afişare conţinut web
TERENCE is a three year long project for the design and development of the first Adaptive Learning System (ALS) for poor comprehenders, hearing and deaf, and for their educators. TERENCE aims at offering innovative usability and evaluation guidelines, pedagogical models, AI technologies, and an ALS for reasoning about stories, in Italian and in English.
Afişare conţinut web Afişare conţinut web

Afişare conţinut web Afişare conţinut web

Overview

TERENCE is a complex adaptive learning system, with learning material for poor comprehenders, hearing and deaf, that is partly designed by experts and partly semi-automatically generated. The part that is automatically generated is then revised manually by pedagogy experts via a personalised interface for them.

More precisely, the learning material is made of the following:

  1. English and Italian stories, organised into coherent units called books, and divided into different difficulty levels for the TERENCE learners;
  2. smart games for reasoning about each story, and divided into difficulty levels for the TERENCE learners;
  3. relaxing games for relaxing and motivating the learners after playing with the smart games.

Textual components of smart games are automatically generated by the TERENCE system by means of natural language and reasoning modules. The textual components of the learning material, e.g., stories and games, are then automatically assembled by a dedicated visualisation module.

This web page contains the links to

  1. the services for analysing texts of stories and generating textual components of smart games, as well as the service for assembling the textual and visual components of the learning material,
  2. repositories of textual and visual components for the learning material,
  3. as well as the available prototypes or releases of the system.

Services

Two natural language processors, one in English and one in Italian, analyse flat stories and annotate the factual events and some links between them, e.g., temporal links. The natural language processors can be invoked via the English annnotation service and the Italian annnotation service with input flat texts of stories in the respective languages.

The stories, annotated by the natural language processors, are then passed on to the reasoning services that do two things:

  1. they check the consistency of the annotations with respect to an Allen-like semantics over reals, repair possible inconsistencies, deduce further temporal annotations thereby producing so-called enriched stories; see enrichment service;
  2. they take in input the enriched stories and generate textual instances of smart games from them, according to the TERENCE game framework (see D2.3), in XML format; see the generation service.

Finally, the visualisation module automatically assembles textual components for stories and smart games together.

Repositories

The stories of TERENCE, in English and in Italian, are collected into books. The reader can access the (flat) texts of the books of Englishs stories as well as the books of Italian stories.

A PDF version of the stories and the games is also available for download here: 

The natural language services allow us to analyse flat stories and annotate them. The reader can access the repositories of the resulting annotated stories in English and of annotated stories in Italian.

The reasoning services allow us to process the annotated stories and generate textual components of smart games automatically. The reader can access the current versions of the generated textual smart games' repository.

In order to facilitate the exploitation of the TERENCE stories and smart games for research purposes, we prepared and published the TERENCE ontology and datasets in a Linked Open.Data (LOD) compliant format at http://datahub.io/dataset/terence as well as at the LinkedUp repository of educational LOD. Stories and games (and their metadata) can be now accessed and linked via unique URIs and a SPQRQL endpoint is provided for querying the dataset. Moreover, links to the TERENCE annotations of the stories and textual versions of the games, automatically generated, will be added very soon.

The textual stories and textual smart games, as well as other components of the TERENCE learning material, are illustrated by professional designers. Relaxing games are also developed by them according to the specifics set in D2.3. The reader can access the repositories of illustrations and relaxing games.

Prototypes and releases

The first high-fidelity flash prototypes of smart games were released in March 2011. The reader can also watch a flash video of the prototype of smart games released in March 2011.

The first release of the entire system for the small scale evaluation was released in June 2012. The reader can also watch a flash video of the June release.

The second release of the entire system for the large scale evaluation was made available at the end of February 2013. The final release of the entire system was made available at the end of the project, i.e., September 2013.