<p><u><strong>R&eacute;sum&eacute;</strong></u></p> <p>&Eacute;valuer un Syst&egrave;me de Recherche d&rsquo;Information en Ligne (SRIL) revient traditionnellement &agrave; interroger son acceptabilit&eacute;, voire son utilisation. L&rsquo;acceptabilit&eacute;&nbsp;d&rsquo;un SRIL&nbsp;se d&eacute;finit traditionnellement par&nbsp;la perception plus ou moins positive &agrave; propos de son utilit&eacute; et de son utilisabilit&eacute;.&nbsp;A ces deux dimensions traditionnelles nous ajoutons la d&eacute;sirabilit&eacute;, c&rsquo;est-&agrave;-dire la n&eacute;cessit&eacute; du syst&egrave;me et l&rsquo;envie d&rsquo;utilisation qu&rsquo;il suscite. Nous postulons donc pour &eacute;valuer un SRIL, il faut s&rsquo;int&eacute;resser &agrave; son utilit&eacute;, &agrave; son utilisabilit&eacute; et &agrave; sa d&eacute;sirabilit&eacute;. L&rsquo;utilit&eacute; du syst&egrave;me recouvre sa performance et l&rsquo;avantage relatif qu&rsquo;il procure. L&rsquo;utilisabilit&eacute; se rapporte quant &agrave; elle, &agrave; l&rsquo;efficience du syst&egrave;me, &agrave; son apprenabilit&eacute;, &agrave; sa m&eacute;morabilit&eacute; et &agrave; sa capacit&eacute; de gestion des erreurs. Enfin, la d&eacute;sirabilit&eacute; renvoie &agrave; l&rsquo;apparence esth&eacute;tique du syst&egrave;me et &agrave; son int&eacute;gration dans le contexte de l&rsquo;utilisateur. Nous avan&ccedil;ons &eacute;galement que toute relation entre ces trois dimensions est&nbsp;&agrave; priori&nbsp;possible et que ces probables relations sont interpr&eacute;tables suivant la rigueur que requiert une &eacute;valuation d&rsquo;un SRIL.</p> <p>&nbsp;</p> <p><u><strong>Abstract</strong></u></p> <p>Evaluating an Online Information Retrieval System (SRIL) traditionally comes down to questioning its acceptability, or even its use. The acceptability of an SRIL has traditionally been defined by the more or less positive perception about its usefulness and usability. To these two traditional dimensions we add desirability, that is to say the necessity of the system and the desire for use that it arouses. We therefore postulate to evaluate an SRIL, we must be interested in its utility, its usability and its desirability. The utility of the system encompasses its performance and the relative advantage it provides. Usability relates to the efficiency of the system, its learnability, its memorability and its ability to handle errors. Finally, desirability refers to the aesthetic appearance of the system and its integration into the context of the user. We also argue that any relationship between these three dimensions is a priori possible and that these probable relationships can be interpreted according to the rigor required by an assessment of an SRIL.</p>