1 | MG INFORMATION RETRIEVAL SYSTEM
|
---|
2 | ===============================
|
---|
3 |
|
---|
4 | The MG system is a suite of programs for compressing and indexing text
|
---|
5 | and images. Most of the functionality implemented in the suite is as
|
---|
6 | described in the book ``Managing Gigabytes: Compressing and Indexing
|
---|
7 | Documents and Images'', I.H. Witten, A. Moffat, and T.C. Bell; Van
|
---|
8 | Nostrand Reinhold, New York, 1994, ISBN 0-442-01863-0; US $54.95; call
|
---|
9 | 1 (800) 544-0550 to order.
|
---|
10 |
|
---|
11 | These features include:
|
---|
12 |
|
---|
13 | -- text compression using a Huffman-coded semi-static word-based scheme
|
---|
14 | -- two-level context-based compression of bi-level images
|
---|
15 | -- FELICS lossless compression of gray-scale images
|
---|
16 | -- combined lossy/lossless compression for textual images
|
---|
17 | -- indexing algorithms for large volumes of text in limited main memory
|
---|
18 | -- index compression
|
---|
19 | -- a retrieval system that processes Boolean and ranked queries
|
---|
20 | -- an X windows interface to the retrieval system
|
---|
21 |
|
---|
22 | As one example, a collection of 2 Gb of text (1,700,000 documents) can
|
---|
23 | be indexed (on a SPARC 10 Model 512) in about four hours and compressed
|
---|
24 | in a further four hours to make a database that in total occupies less
|
---|
25 | than 800 Mb, or 40% of the original size. This includes a full index to
|
---|
26 | every word and number in the original text. Boolean queries such as
|
---|
27 | ``managing AND gigabytes'' run in a few seconds, and ranked queries of
|
---|
28 | 30--50 terms are evaluated in 10--30 seconds.
|
---|
29 |
|
---|
30 | Details of these methods and further performance results appear in the
|
---|
31 | MG book.
|
---|
32 |
|
---|
33 | The MG system comes with ABSOLUTELY NO WARRANTY; for details see the
|
---|
34 | file COPYING.
|
---|
35 |
|
---|
36 | Instructions on how to build and install mg are in the file INSTALL.
|
---|
37 |
|
---|
38 |
|
---|
39 | ** CHANGES FROM BOOK
|
---|
40 |
|
---|
41 | For copyright reasons the stemmer used in this distribution of MG is
|
---|
42 | not the same as the one illustrated in Figure 3.8 on page 108 of the MG
|
---|
43 | book. This means that the numbers generated by the command ``mgstat
|
---|
44 | alice'' will not match those numbers in Figure A.1 on page 394.
|
---|
45 | Another stemmer was initially written as a simple stopgap for version
|
---|
46 | 1.0. That stemmer has been replaced by a stemmer based on the Lovin's
|
---|
47 | stemming algorithm for mg-1.1.
|
---|
48 |
|
---|
49 | The output format of ``mgstat'' has changed since Figure A.1 (page 394)
|
---|
50 | was prepared. The same information is displayed but formatted
|
---|
51 | differently.
|
---|
52 |
|
---|
53 | ** MG VERSIONS
|
---|
54 |
|
---|
55 | The current version is mg-1.2, September 1995. The changes from earlier
|
---|
56 | versions are listed in the file MODIFICATIONS. This can be accessed
|
---|
57 | with mg by building a database using ``mgbuild mods'' and can also be
|
---|
58 | accessed from the mg web page (see below).
|
---|
59 |
|
---|
60 | The mg-1.2 extensions include:
|
---|
61 |
|
---|
62 | -- Source modifications for use of GNU's autoconf.
|
---|
63 |
|
---|
64 | The mg-1.1 extensions include:
|
---|
65 |
|
---|
66 | -- A new highlighting mode.
|
---|
67 | The output mode ``hilite'' will highlight the query terms in the
|
---|
68 | retrieved text documents. The variable ``hilite_style'' can be set
|
---|
69 | to ``bold'' or ``underline''. It works best with the pager
|
---|
70 | ``less''. A .mgrc to use would include:
|
---|
71 | .set pager less
|
---|
72 | .set mode hilite
|
---|
73 | .set hilite_style bold
|
---|
74 |
|
---|
75 | -- A web site containing manual pages, documentation, and a
|
---|
76 | mgquery demo page (utilising cgi scripts).
|
---|
77 | See: http://www.kbs.citri.edu.au/mg
|
---|
78 | One of these pages ``about_mg.html'' is included in this
|
---|
79 | distribution.
|
---|
80 |
|
---|
81 | -- A revised mg_get script which uses a .mg_getrc file to map
|
---|
82 | specific collection names to filter types. (Modifications by Bruce
|
---|
83 | McKenzie). See mg_get.1 for more details.
|
---|
84 |
|
---|
85 | -- Code to perform merging of existing databases. This code
|
---|
86 | was created by Shane Hudson and is documented in the mgmerge.README
|
---|
87 | file found in the docs subdirectory. This code is maintained by
|
---|
88 | Shane Hudson ([email protected]).
|
---|
89 |
|
---|
90 | -- Revised man pages, including some new entries (thanks to Nelson
|
---|
91 | Beebe). See mg.1, mgintro.1, mgintro++.1.
|
---|
92 |
|
---|
93 | -- A real (rather than toy) stemmer.
|
---|
94 |
|
---|
95 | ** PORTABILITY
|
---|
96 |
|
---|
97 | Please refer to "README.port".
|
---|
98 |
|
---|
99 | ** CREDITS
|
---|
100 |
|
---|
101 | The MG development is largely the result of research collaboration
|
---|
102 | between:
|
---|
103 |
|
---|
104 | Tim C. Bell <[email protected]>
|
---|
105 | Ian Witten <[email protected]>
|
---|
106 | Alistair Moffat <[email protected]>
|
---|
107 | Justin Zobel <[email protected]>
|
---|
108 |
|
---|
109 | The bulk of the programming work has been carried out by:
|
---|
110 |
|
---|
111 | Stuart Inglis <[email protected]>
|
---|
112 | Craig Nevill-Manning <[email protected]>
|
---|
113 | Neil Sharman <[email protected]>
|
---|
114 | Tim Shimmin <[email protected]>
|
---|
115 |
|
---|
116 | In addition to these, the following people have contributed to the
|
---|
117 | development of the MG software:
|
---|
118 |
|
---|
119 | Lachlan Andrew <[email protected]>
|
---|
120 | Gary Eddy <[email protected]>
|
---|
121 | Hugh Emberson <[email protected]>
|
---|
122 | Kerry Guise <[email protected]>
|
---|
123 | Shane Hudson <[email protected]>
|
---|
124 | Linh Huynh <[email protected]>
|
---|
125 | Bohdan S. Majewski <[email protected]>
|
---|
126 | Bruce McKenzie <[email protected]>
|
---|
127 |
|
---|
128 | In addition to these, the following people have submitted bug reports
|
---|
129 | and suggestions/fixes:
|
---|
130 |
|
---|
131 | Rex Barzee <[email protected]>
|
---|
132 | Tim A.H. Bell <[email protected]>
|
---|
133 | Tim C. Bell <[email protected]>
|
---|
134 | Nelson Beebe <[email protected]>
|
---|
135 | Rodney Brown <[email protected]>
|
---|
136 | Rok Sosic <[email protected]>
|
---|
137 | Carl Staelin <[email protected]>
|
---|
138 |
|
---|
139 | Development of the MG system was supported by the Australian Research
|
---|
140 | Council; the Universities of Melbourne, Waikato, Canterbury, and
|
---|
141 | Calgary; RMIT; and the Collaborative Information Technology Research
|
---|
142 | Institute (Melbourne).
|
---|
143 |
|
---|
144 | ** BUG REPORTS
|
---|
145 |
|
---|
146 | Send bug reports to <[email protected]> and <[email protected]>.
|
---|
147 | Back-traces from gdb are always welcome but not mandatory :-)
|
---|
148 |
|
---|
149 | ** FURTHER READING
|
---|
150 |
|
---|
151 | A bibliography of MG related research work appears in the files
|
---|
152 | MG.Bibliography.ps and MG.Bibliography.bib on the ftp site and
|
---|
153 | is accessible through the Web page.
|
---|
154 |
|
---|
155 |
|
---|
156 |
|
---|