<?xml version="1.0" standalone="yes"?>
<Paper uid="E87-1011">
  <Title>A Multi-Purpose Interface to an On-line Dictionary</Title>
  <Section position="7" start_page="67" end_page="68" type="concl">
    <SectionTitle>
7 Conclusion
</SectionTitle>
    <Paragraph position="0"> We have sketched the requirements for, and the design of, a flexible interface to an on-line dictionary, capable of supporting the lexical requirements of a number of active research projects and adaptable to a range of applications. The programs have been developed in Lisp, and make heavy use of the interactive graphic capabilities of Xerox's Lisp workstations. Nothing, however, depends critically on this; the Interllsp-D interactive graphics simply make the task of constructing search specifications very easy for the end user. The design is sufficiently modular to allow easy modification, and the system would be capable of functioning on a conventional minicomputer or mainframe (as long as it supported random access to files) just as well as it does on a single user workstation.</Paragraph>
    <Paragraph position="1"> In order to make the lexical knowledge base available to all the projects requiring access to it, the system had to be adapted to fit into the local environment of networked workstations. The database manager was easily repackaged to reflect the model of one server catering for several clients over a network; the Remote Procedure Call Protocol (XSIS, 1981) provided the necessary functionality to incorporate the manager into a dictionar!t server node (of the kind discussed by Kay, 1984) -- this bypassed the need for costly flleservers and proved the integrity of the design. We also plan to develop a version of the system running in Franz Lisp under UNIX and accessing the M_RC dictionary database (Coltheart, 1981).</Paragraph>
    <Paragraph position="2"> While the system implements in effect a linguistically motivated DBMS constructed round a suitable machine-readable source, it stops short of full browsing capability (even though such a capability could easily be added by fully integrating AlshawFs definitions analysis program into the overall design). In this sense the lexical knowledge base discussed here differs  from the concept of a lexical database as described in Calzolari (1986), or underlying Miller's WORDNET (1985). Nonetheless, the methodology described here is sufficiently flexible and powerful to satisfy a substantial proportion of the needs of the computational linguistics community till a proper mix of database and browsing capabilities becomes available.</Paragraph>
  </Section>
class="xml-element"></Paper>