1 |
<HTML> |
2 |
<HEAD> |
3 |
<META NAME="GENERATOR" CONTENT="Adobe PageMill 2.0 Mac"> |
4 |
<TITLE> FS Consulting - MPS Search Interface </TITLE> |
5 |
</HEAD> |
6 |
<BODY BGCOLOR="#ffffff"> |
7 |
|
8 |
<TABLE BORDER=0 WIDTH=100%> |
9 |
<TR><TD VALIGN=TOP ALIGN=RIGHT COLSPAN=3> <A HREF="/"><IMG SRC="/demonstrations/search/images/banner.gif" ALT="Home" BORDER=0></A> </TD></TR> |
10 |
<TR><TD VALIGN=TOP ALIGN=CENTER> <H2>How to Search a database</H> </TD> </TR> |
11 |
</TABLE> |
12 |
|
13 |
<H3>Basic Searches:</H3> |
14 |
|
15 |
<P>For basic searching, you just need to supply a word or a set of words |
16 |
which will be used by the search engine to look documents. So you could |
17 |
look for 'animals', 'information retrieval systems' or 'user interface design'. |
18 |
The search engine will assume that the words are ORed together in the absence |
19 |
of other operators.</P> |
20 |
|
21 |
<H3>Wildcard Searches:</H3> |
22 |
|
23 |
<P>You may also do wildcard searched, so you could look for 'comput*' and |
24 |
the search engine will look for all words which begin with 'comput', so |
25 |
it will search for 'computer', 'computing', 'computational', etc.</P> |
26 |
|
27 |
<H3>Boolean Searches:</H3> |
28 |
|
29 |
<P>Boolean searches are also supported, so you may do things like 'lowland |
30 |
<B>AND</B> gorilla', or 'animals <B>NOT</B> gorilla'. Note that the operators |
31 |
need to be in <B>UPPER CASE</B>.</P> |
32 |
|
33 |
<P>You may also nest boolean searches using brackets, so you can do '(lowland |
34 |
<B>OR</B> gorilla) AND animals'.</P> |
35 |
|
36 |
<P>Booleans are implemented in a 'smart' way in that operators have precendence, |
37 |
so 'microsoft <B>AND</B> graphical user interfaces' will be evaluated as |
38 |
'microsoft <B>AND</B> (graphical <B>OR</B> user <B>OR</B> interfaces)'.</P> |
39 |
|
40 |
<H3>Proximity Searches:</H3> |
41 |
|
42 |
<P>Proximity searching is also available by using the <B>ADJ</B> operator. |
43 |
So to find all documents in which 'microsoft' and 'windows' are adjacent, |
44 |
you would use 'microsoft <B>ADJ</B> windows'.</P> |
45 |
|
46 |
<P>A simpler alternative syntax for proximity searches is to use quotes, |
47 |
so the search described above would look like "microsoft windows".</P> |
48 |
|
49 |
<H3>Fielded Searches:</H3> |
50 |
|
51 |
<P>Fielded searching is available for any database which supports it. To |
52 |
restrict a search to a particular field, you need to enter the field name |
53 |
followed by the equal sign, followed by the value you want to search, so |
54 |
to look for 'marchionini' in the 'author' field, you would use 'author=marchionini'.</P> |
55 |
|
56 |
<P>You can also combine field searches with brackets, so 'author=(marchionini |
57 |
AND florance)' would search for records which have 'marchionini' and 'florance' |
58 |
as authors.</P> |
59 |
|
60 |
<H3>Range Searching:</H3> |
61 |
|
62 |
<P>Range searching can also be performed on any numeric field, for example |
63 |
to search for all documents that were published after 1993 (assuming that |
64 |
the database had a 'date' field) you would use 'date>1993'.</P> |
65 |
|
66 |
<P><HR ALIGN=CENTER WIDTH=50%></P> |
67 |
|
68 |
<H3>Putting it all together:</H3> |
69 |
|
70 |
<P>So the following searches are legal:</P> |
71 |
|
72 |
<PRE> |
73 |
author=marchionini AND date>1993 |
74 |
|
75 |
title=encyclopedia AND author=marchionini |
76 |
|
77 |
(lowland AND gorilla) OR (animals AND plants)</PRE> |
78 |
|
79 |
<P>this is not the same as:</P> |
80 |
|
81 |
<PRE> |
82 |
lowland AND gorilla OR animals AND plants</PRE> |
83 |
|
84 |
<P>which would be evaluated as:</P> |
85 |
|
86 |
<PRE> |
87 |
lowland AND (gorilla OR animals) AND plants</PRE> |
88 |
|
89 |
<HR ALIGN=CENTER WIDTH=50%> |
90 |
|
91 |
<H3>Soundex Searches:</H3> |
92 |
|
93 |
<P>The use of 'soundex' is also supported in searches. This is useful for |
94 |
names, for example, when one is unsure of the spelling or if there are a |
95 |
number of spelling variants. For example if I were to look for documents |
96 |
containing the name 'tognazini' and I were unsure whether it is spelt with |
97 |
two 'n's and/or two 'z's, I would be able to use the soundex capability |
98 |
to search for the documents as follows 'soundex[tognazini]'. The server |
99 |
will expand the term and list all the variants in the query report. The |
100 |
user could then browse it to look at them and decide which is the correct |
101 |
spelling. You can also restrict the 'soundex' operation to certain fields |
102 |
such as 'author=soundex[tognazini]'.</P> |
103 |
|
104 |
<H3>Typographical Searches:</H3> |
105 |
|
106 |
<P>The use of 'typographical variation' is also supported in searches. This |
107 |
is useful when searching for words which could be mis-spelled, for example, |
108 |
if 'animals' is spelled 'anmials'. This operator looks for simple typos |
109 |
and juxtaposed letters. I would be able to use this capability to search |
110 |
for the documents as follows 'typo[animals]'. You can also restrict the |
111 |
'soundex' operation to certain fields such as 'author=typo[animals]'.</P> |
112 |
|
113 |
<H3>Using the Forms Interface to Search:</H3> |
114 |
|
115 |
<P>The forms provide a way of structuring the searches for the user so that |
116 |
they dont have to remember what field are available in the different datababase. |
117 |
The basic input field will accept any kind of query from the user, while |
118 |
the titled fields will accept text for searching on that particular field |
119 |
only. The user can fill in any of the fields and they will be ANDed together |
120 |
before searching.</P> |
121 |
|
122 |
<H3>Stemming:</H3> |
123 |
|
124 |
<P>Two types of stemming are available, one is a plural stemmer (which is |
125 |
used for both sample databases offered here), and the other is a Porter. |
126 |
Both stemmers will handle US and UK English.</P> |
127 |
|
128 |
<!--#include virtual="/demonstrations/search/snippets/footer.html" --> |
129 |
|
130 |
</BODY> |
131 |
</HTML> |