<?xml version="1.0" standalone="yes"?>
<Paper uid="W01-1505">
  <Title>SiSSA - An Infrastructure for NLP Application Development</Title>
  <Section position="4" start_page="0" end_page="0" type="metho">
    <SectionTitle>
3 SiSSA at Work
</SectionTitle>
    <Paragraph position="0"> There are two main activities regarding the characteristics of SiSSA described in this paper: the development of projects and the integration of processors.</Paragraph>
    <Section position="1" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
3.1 Projects
</SectionTitle>
      <Paragraph position="0"> The creation and editing of projects takes place exclusively via the SiSSA graphical interface.</Paragraph>
      <Paragraph position="1"> First the user decides which linguistic activities are relevant to her/his project. Then s/he can browse the Processor Repository, searching for those which are suitable to realize each linguistic activity.10 Finally, s/he composes them into a project.</Paragraph>
      <Paragraph position="2">  some processors developed by the partners of the project: the morphological analyzer and the parser of NLGRADE (ILC, When it is necessary to test a given project on a text the SiSSA Manager prepares a suitable stream (IDataStream) and sends it to the processor selected as the first in the analysis chain. The processor interprets the metadata, executes the specified operation on the linguistic data and finally sends its output to all its observers; some of them can be required to perform further processing on the linguistic data. The output produced by a processor is sent to the SiSSA Manager as well, so that it can be shown to the user in a suitable form.</Paragraph>
      <Paragraph position="3"> In Figure 2 the starting page of the SiSSA system is shown. In the upper part of the window there are a few buttons that are present in all the pages of SiSSA. From left to right:  Figure 3 shows the applet that interactively monitors the status of the project currently active and displays it to the user. In the upper part of the window the details of the active project are shown: the processors (left), the connections between processors (middle), and the XSL filters (right). In the lower part of the window the messages coming from the processors are shown. The bottom bar shows which of the processors/filters is currently active (using the IStateMonitor interface described in Section 2.2).</Paragraph>
    </Section>
    <Section position="2" start_page="0" end_page="0" type="sub_section">
      <SectionTitle>
3.2 Integration of processors
</SectionTitle>
      <Paragraph position="0"> Differently from the activity of creation and editing of projects, only the final part of the work involved in the integration of processors is accomplished via the SISSA graphical interface (more written in C and running under Windows: (Prodanof et al., 1998; Prodanof et al., 2000)) and the preprocessor and the parser of GEPPETTO (ITC-irst, written in Common Lisp and running under Solaris: (Ciravegna et al., 1997; Ciravegna et  precisely, the registration in the Processor Repository of the availability of the processors).</Paragraph>
      <Paragraph position="1"> In order to make a processor SiSSA-compliant, the following steps are necessary: a0 to provide it with a wrapper so that it communicates via the CORBA IDLs of SISSA; a0 to make a translation between the processor's native input/output and the corresponding linguistic representation specified by process-data; a0 to register the processor in the Processor Repository using the SiSSA graphical interface; during this step the class of the processor, its corbaloc: URL and activation string have to be specified.</Paragraph>
    </Section>
  </Section>
class="xml-element"></Paper>