Text retrieval techniques for office documents

This paper begins with an overview of the currently used methods for text retrieval, describing the characteristics of the environments they are respectively suited to. Next, the work accomplished within the context of the MULTOS (MULTimedia Office System) project is considered. MULTOS includes in particular features for retrieval of documents by partial specification of the content. The method developed is that of signatures with superimposed coding. Lastly, an evaluation of the performances of the implemented algorithm is presented. MULTOS is an ESPRIT project, whose participants are Battelle (D), IEI (I), Mnemonica (GR), Olivetti (I), Philips (NL), Research Centre of Crete (GR), Triumph-Adler (D).