Amaryllis

AMARYLLIS

Goal

The main objective of this task is to exploit the AMARYLLIS experience in French collection provision and evaluation campaign managment in order to promote research in information retrieval using the French language.

This first experiment will provide participants with a different type of data collection (scientific texts rather than newspaper and news agency documents) and provides tools that can be used in the retrieval task (a controlled vocabulary in English and French).

The task will investigate the ability of systems to retrieve bibliographic references in a multi-disciplinary scientific database. The aim is to enable system testing and tuning using this type of data.

Preliminary Track Guidelines

The features of the experimental framework are:

a French scientific test collection
short documents, scientific writing style
scientific terminology, acronyms, etc
keywords listed in a French-English controlled- vocabulary.
a medium-sized collection (about 200 MB)

The INIST "controlled vocabulary" dictionary will be made available for participants.

The main task in Amaryllis 2002 will be French monolingual retrieval.

Several different runs are envisaged as follows:

to work only on free text (i.e : title and abstract)
to work with the controlled vocabulary
to work with both free text and controlled vocabulary.

An optional, additional task will be English -> French cross-language retrieval.

In this subtask, the topic set will be provided in English and participants can perform cross-language runs retrieving French documents using the English topic set and, optionally, the English keywords and dictionary provided. Participants in this task must also submit at least one monolingual run.

Participants will be able to test their systems in different contexts. Final decisions on the details of permitted runs will communicated with the release of the topics (1 April 2002). However, the Track Coordinators would appreciate early statements of intention to participate.

Conditions for Participation

Participants can also test their system in different contexts of use which must be clearly described (document fields used, topic languages, dictionary fields used). The specific conditions under which each experiment is performed must be described in a file, which has the same name as the results file plus the suffix 'desc'.

The description file should contain the following information:

TL: Topic Languages - value = FR or EN
FU: Fields Used for index with 3 possible values
- TEXT if title and abstract are exclusively used as indexing fields ( <ti> and <ab> in document set).
- KW if keyword field only is used ( <mc>for French and <kw> for English in documents set).
- BOTH if free text and keywords are jointly used.
CD: value = YES if Controlled Dictionary is used or NO if not.
CO: any other COmments or precisions.

The Track Coordinators would appreciate early statements of intention to participate.

Data Collection

The Amaryllis database consists of over 150 000 bibliographic documents for approximatively 200 MB. It includes biological sciences, chemical sciences, engineering sciences, humanities and social sciences, information science, medical sciences, physical and mathematics sciences, sciences of the universe.

Example of document:

<doc>
	<docid>	005768 </docid>
	<tit>	Dermatite atopique : Actualisation </tit>
	<ab>	La dermatite atopique (DA) est une pathologie fréquente du jeune enfant, dont le traitement classique repose sur des soins locaux adaptés. La meilleure compréh ension physiopathologique de la DA a réactualisé le rôle des antig&egrav e;nes de l'environnement dans cette pathologie multifactorielle, modifiant ainsi la prise en charge diagnostique et thérapeutique, en particulier dans les DA sévèr es. L'enquête allergologique (allergènes alimentaires, pneumallergènes de contact) a un intérêt évident dans les DA graves, résistantes a un traitement, en cas d'aggravation d'une DA jusque là bien contrôlée, ou en cas de suspicion cliniques.
	</ab>
	<mc>	Dermatite atopique, Enfant, Nourrisson, Article synthèse </mc>
	<kw>	Atopic dermatitis, Child, Infant, Review </kw>
</doc>

Example of topic in French:

<record>
	<qid>	22 </qid>
	<dom>	Psychologie pharmacologie </dom>
	<query>	Traitement de la dépendance aux boissons alcoolisées </query>
	<info>	Pour être pertinent, un document devra traiter de tous les types de traitement de l'alcoolisme (chimiothérapie et/ou psychothérapie). Les documents traitant des aspects santé publique (prévention, organisation de la santé) seront aussi pertinents. A l'inverse, les documents portant sur les complications médicales de l'alcoolisme et leurs traitements, sur la personnalité et la représentation sociale des alcooliques ne seront pas pertinents
	</info>
	<concept>	Alcoolisme, Traitement, Chimiothérapie, Psychothérapie </concept>
</record>

Example of dictionary entry:

<term>
	<fr> Dermatite exfoliatrice Ritter von Rittersheim</fr>
	<en> Ritter disease</en>
	<synofr> Ritter von Rittersheim maladie</synofr>
</term>

In order to participate in this task, please contact one of the Amaryllis Task Coordinators

Patrick KREMER or Laurent SCHMITT

INIST-CNRS
2 Allée du parc de Brabois,
54500 VANDOEUVRE,
FRANCE