\centerline{\bf Book Review}
\medskip
\noindent{\it Automatic Text Processing, the
Transformation, Analysis and Retrieval of Information by
Computer, Gerard Salton, 530pp, Addison-Wesley Publishing,
\quid36.85, {\sc isbn:} 0 201 12227 8.}
\smallskip
\noindent
As the subtitle implies, Salton covers much more than
simple `text processing' in this book. He covers a wide
variety of computer techniques which apply in some way to
textual material. Along the way, this includes the
automated office, text compression, encrytion, text
retreival, automatic indexing, and so on. The book is
structured that the four parts can be
read separately, and even within each part Salton indicates
which bits are `essential', rather than `optional'. He
classifies his readership into three main categories:
`trained computer science students and computer
professisonals', `computational linguitics and
language-processing researchers' and `students in library
or information science, or science and technology
programs'. For each group he outlines a rather different
reading pattern.

Given the encyclopaedic scope of the book this selection
is quite welcome. But the scope also means that some
material is a little dated. Some areas
have a dated feel anyway --- text compression is not one
of the most vigorous areas in computing, and many of the
references are quite old. But take \TeX, since I've heard
of it before: Salton's contact with \TeX\ (indexed as `TeX
editing program') is through the 1978 `\TeX\ and Metafont'
book. While the underlying box and glue model is
unchanged, the discussion of hyphenation \`a la \TeX\ is
obselete. His account of the {\sc unix} tools seems
better. \MF\ is also described in terms of the earlier
version --- and a rather more radical rethink took place
before it acheived its present incarnation. On the other
hand, Salton skips from |troff|/\TeX\ to {\it wysiwyg\/}
processors like MacWrite, and from there to systems using
`property sheets' (Lara, Bravo, etc.). The survey is
therefore a bit spotty.

His chapter on automatic indexing is interesting. I've
often wondered how to tackle this (or even if you
should). Part Four, covering text analysis
and language processing, was the highlight of the book
for me. At the completely trivial level, his discusison
of the {\sc unix} Writers Workbench reminded me that the
reason {\sc unix} people write such clear, unambiguous
prose is the availability of  such tools.

It is not a book to read from cover to cover.
For many of its topics it is a good place to start. The
references seem to cover the ground well, and there are a
fair sprinkling of fairly recent references too. Its
sheer scope means that some topics are examined  less
deeply, but it is still a useful, comprehensive synthesis
of the wider field of `text processing'.\looseness-1
\smallskip
\rightline{\sl Malcolm Clark}