readme.harvest

harvest是一个下载html网页得机器人

NOTE: As of May 2001, the Rainbow based summarizers are obsolete and
      included for demonstration purposes only.


This is an example how to use ``Rainbow'' to summarize Framemaker's Maker
Interchange Format (MIF) and Microsoft's Rich Text Format (RTF) documents.

The MIF and RTF summarizers work by first converting into SGML and then
using the Harvest SGML summarizer.  The SGML conversion step is performed
using a commercial software package (and DTD) called ``Rainbow'', which was
developed by Electronic Book Technologies.  Since converting these formats
to SGML is an ``up conversion'', the translation process is necessarily
imperfect.

Files converted from RTF/MIF to SGML by the rbmaker program look something
like this:

        <PARA PARATYPE="title">My Document Title</PARA>
        <PARA PARATYPE="heading 1">Introduction</PARA>
        <PARA PARATYPE="normal">The purpose of this ...<PARA>

We can separate these two paragraphs with the following translation table
entries:

        <PARA,PARATYPE=title>           title
        <PARA,PARATYPE=heading 1>       headings,keywords
        <PARA>                          body

Unfortunately, the type names title, heading 1, etc. are not standardized.
Therefore, our default translation table may not do an adequate job. To
learn which paragraph type names are contained in a document, you can add
this to the table:

        <PARA:PARATYPE>                 paragraph-types

and then examine the SOIF output.