<?xml version="1.0" standalone="yes"?>
<Paper uid="W96-0512">
  <Title>An Architecture For Distributed Natural Language Summarization</Title>
  <Section position="4" start_page="45" end_page="47" type="metho">
    <SectionTitle>
3 Summarization architecture
</SectionTitle>
    <Paragraph position="0"> The interoperability problem is addressed using a proposed standard for exchange of information and knowledge KQML \[Finin et al. 1994\]. KQML aims at  the standardization of both a protocol and a message format for communication among independent processes over a wide-area network. KQML is used to create facilitators which provide the interface between heterogeneous applications which run on various machines and which are written in various programming languages. Such facilitators communicate through KQML performatives and exchange messages written in some content language. In our case, this is a simple template language, developed locally.</Paragraph>
    <Paragraph position="1"> Our architecture draws from work on Software Agents \[Genesereth and Ketchpel 1994\]. Our goal was to expand the model to incorporate natural language interfaces. We have used agents of various types in a modular way:  the modules through the intermediary of facilitators that convert from the template format to KQML and vice-versa. In our system, the role of data collectors is performed by the MUC systems and the facilitators connected to the World Book.</Paragraph>
    <Paragraph position="2"> Planner: it maintains contacts with the facilitators in order to keep the knowledge base of the summarizer up to date. It uses KQML subscription messages to learn in an asynchronous way about changes in the knowledge bases of other facilitators.</Paragraph>
    <Paragraph position="3"> The following example shows how the planner uses a KQML subscription message to subscribe to new messages related to E1 Salvador.</Paragraph>
    <Paragraph position="4">  Whenever a new message becomes available (E.g., Figure 2), the MUC facilitator will reply with an appropriate message.</Paragraph>
    <Paragraph position="5"> Summarizer(s): agents that are concerned with summarizing the data that they have collected over the network from different sources and producing natural-language reports for the end-user. The summarizer is connected with the user model and the user interface.</Paragraph>
    <Paragraph position="6"> Database servers: expert agents that have access to knowledge bases which are updated periodically and which contain information that is less likely to change over the course of a summarization session (e.g. heads of state, geographical and common-sense knowledge). In our case, such information comes from two sources: the CIA World Book \[CIA 1995\] and the ontologies supplied with the MUC conferences. An example from the World Book related to E1 Salvador is shown in Figure 4. The World Book facilitator parses the entries for each country into a Lisp-like format and provides access to them to the planner. Another instance of a database server is the facilitator connected to the node labeled Ontology in Figure 3. This represents the database containing the ontologies (including geographical locations, weapons, and incident types, available from the MUC conference). null Data collectors: agents that are connected to the real world through filters or use human experts who can feed real-time raw data such as sports scores, news updates, changes in stock prices, etc. They are connected to the rest of  Other KQML performatives, such as ask-all, ask-one, register, tell, or sorry have also been implemented.</Paragraph>
    <Paragraph position="7"> User Model: it keeps information about the user's interests (e.g. keywords, regions in the workl), preferences (how frequently he wants to get updates), and interaction history (what information has already been shown to him).</Paragraph>
    <Paragraph position="8"> Let's consider the case in which the user has already been notified abo,lt a terrorist act: A bombing took place on August 23rd, 1988 in the district of Talcahuano, Chile.</Paragraph>
    <Paragraph position="9"> The next time the system needs to refer to the same event, it can omit some information that it has already shown to the user (e.g., the fact that Talcahuano is in Chile), and can instead focus on information that has not been included previously.</Paragraph>
    <Paragraph position="10"> The Talcahuano bombing didn't result in any injuries. However, the Chapel of the Church of Jesus was damaged.</Paragraph>
  </Section>
  <Section position="5" start_page="47" end_page="47" type="metho">
    <SectionTitle>
4 Current Work and Direc-
</SectionTitle>
    <Paragraph position="0"> tions for Future Research Currently, our system can handle simple summaries consisting of 1-3 sentence paragraph which are limited to the MUC domain and to a few additional events for which we have manually created MUC-like templates. Several components related to interoperability are also fully implemented (e.g., the subscription package in KQML and the query-response interface to the MUC and World Book facilitators). We haven't yet connected the system to a working MUC component 2. The user model hasn't been implemented yet.</Paragraph>
    <Paragraph position="1"> A problem that we haven't addressed is related to the clustering of articles according to their relevance to a specific event. Another issue is domain-independence.</Paragraph>
    <Paragraph position="2"> Since the understanding and generation modules share only language-independent templates, we would try to implement a limited form of machine translation by summarizing in one language news written in another language.</Paragraph>
  </Section>
class="xml-element"></Paper>