A searchbox query is made up of words and metadata separated by operators. Words are simply English (or any other language) words searchbox should look for in the documents; they appear as simple words in the query, as shown in Example 11.1, “A query containing a single word”.

searchbox

Moreover, as detailed in Example 11.2, “Some queries using wildcards” you can use wildcards when querying for words. Searchbox supports two wildcards: * (star or asterisk) and ? (question mark). The question mark means “any single character is allowed here” while the asterisk means “any sequence of any number of characters (including no characters at all) is allowed here”.

search*

search??

searchbox allows you to place a wildcard nearly anywhere in a word and not only at the end, as detailed in Example 11.3, “Some complex wildcards queries”; the only limitation is that the very first character of each word must be a real character and not a wildcard. Moreover you can use multiple wildcards in a word.

s*r*

s??r*

s*r

While words are, well, simple words, metadata are instead made of two parts: their key and their value; the key is the metadata name, so to say. You can ask for all the documents written in English with the query in Example 11.4, “Querying for specific metadata”

language:en

You can also query for documents containing metadata whose values belong to a specific range or even perform wildcard searches on the metadata value just as if it were a simple word. This is detailed in Example 11.5, “Selecting documents based on a range of metadata values ”

authordate:20050101:20051231

authordate:20050101:

authordate::20050101

authordate::????01*

There are two other important points in metadata queries, both related to the way metadata are indexed. These are discussed in the section called “Some final points on syntax”.

As stated above you are not limited to a single word or metadata in your queries: you can combine them with operators. The simplest operator is the phrase search (Example 11.6, “Phrase search”) which lets you search for a “phrase” (i.e. a sequence of adjacent words) instead of for a single word. Phrase search is triggered by surrounding words with a pair of double quotes (").

"focuseek searchbox features"

In fact the phrase search has an optional numeric parameter: the sloppiness. It is specified by following the closing double quote with a tilde (~) and a positive integer, the so called sloppiness value. No spaces are allowed between the quote and the tilde nor between the tilde and the integer. Valid values for the sloppiness are in the range from 0 to 2147483648. The sloppiness is the edit distance, i.e. maximum the number of word insertions and moves that searchbox will tolerate in the matched text when looking for the query phrase.

"focuseek features"~1

"focuseek features"~2

Note that a phrase search with a sloppines value of zero is equivalent to conventional phrase search.

You can compose even more complex queries using the so called boolean operators. There are three boolean operators in searchbox: & (an ampersand character, meaning “this word and this other word”^[15] or simply and), | (a vertical pipe character, meaning “this word or this other word” or simply or) and ! (an exclamation point, meaning “not this word“ or simply not). The first two are infix, binary operators, meaning that they accept two arguments and are placed after the first argument and before the second one, just as the regular algebra multiplication and addition operations. The not operator instead is prefix and unary, meaning that it accepts a single argument and the operator itself comes before the argument, just as the minus sign used for negative numbers in regular algebra. The or operator means you are looking for documents containing either one or both its arguments. The and operator means you are looking for documents containing both its arguments. The not operator means you are looking for documents which don't contain its only argument. Finally, you can omit blanks surrounding the operators ^[16] and the and operator (which is by far the most commonly used) can be omitted; thus focuseek & searchbox is the same as focuseek searchbox^[17]. A couple of examples (Example 11.8, “Some simple boolean queries”) will help make things clear.

searchbox & focuseek

searchbox & focuseek

searchbox focuseek

searchbox & focuseek & price

searchbox focuseek price

searchbox | focuseek

searchbox | focuseek | search 

!searchbox

focuseek & !searchbox

"focuseek searchbox" | "search tools"

focuseek & authordate:20050101:

In searchbox queries parentheses have the same meaning they have in regular algebra: they group things and help specifying what to do first. You can surround any word or any operator and its arguments with parentheses: the whole parenthesized expression will be considered as it was a single word.

(focuseek & searchbox) | (search & tools)

(focuseek | "search tool") & price

((focuseek | "search tool") & price) & authordate:20050101:

You might occasionally need to search for a character searchbox uses in its query syntax (e.g. & or :). This most usually occurs with metadata values as searchbox, as shipped, doesn't allow these characters into words. In order to do this you can prefix the character with a single \ (backslash character). Should you need to search for a backslash character simply insert two consecutive backslash characters.

pubdate:2005\:01\:01:2006\:12\:31

Now we get back to metadata queries. As described in the section called “Gathering "side metadata"” metadata passed to searchbox can be marked to be indexed exactly or to be normalized, tokenized or both. This influences your queries in two ways.

First, while "normal" text search in searchbox is case insensitive, metadata search is always case sensitive, both for key and value; thus authordate:20050101 and Authordate:20050101 return different results. However you can tell searchbox that you want to index a specific metadata in normalized form, meaning searchbox will store the metadata value in the same way it stores normal text^[18]. Thus if you store normalized metadata you can query them using lowercase values.

Second, but more important, metadata values are not split into words unless you specify to index the metadata as tokenized. Thus, if your document is marked by the metadata key:val1 val2 (i.e. a single metadata with a value containing a space) queries for key:val1 or key:val2 will return no results. See also Example 11.11, “Querying tokenized and normalized metadata”.

<fff id="docA">
  <!-- A lot of stuff is omitted here --!>
  <meta key="key1">value1 value2<meta>
  <meta key="key2" normalized="1">value1<meta>
</fff>

<fff id="docB">
  <!-- A lot of stuff is omitted here --!>
  <meta key="keyt1" tokenized="1">value1 value2<meta>
  <meta key="key2" normalized="1">VALUE1<meta>
</fff>

<fff id="docC">
  <!-- A lot of stuff is omitted here --!>
  <meta key="key1">value0 value1 value2 value3<meta>
  <meta key="key2">value1<meta>
</fff>

<fff id="docD">
  <!-- A lot of stuff is omitted here --!>
  <meta key="keyt1" tokenized="1">value0 value1 value11 value2<meta>
  <meta key="key2">VALUE1<meta>
</fff>

key1:value1

key1:value1 key1:value2

"key1:value1 key1:value2"

key1:value1\ value2

key1:value1\ value11

"key1:value0 value2"

key2:VALUE1

key2:value1

The ESP is accessible from any modern Jacascript enabled Web browser like IE 6, Firefox 1.5 or Safari 2 at the address:

http://<hostname>:<portnumber>

where hostname is localhost and portnumber is 2200 if you are working on the same machine where searchbox is installed on.

ESP is accessible both as anonymous and authorized user.

ESP can browse groups of archives that are previously organized in collection from the Control Panel. From the Where dropdown you can choose the collection where ESP will perform the query.

ESP shows results while you type into the search box starting from the fourth character. This feature is active only for single-word query, in case you use more concatenated words a standard query is performed. The incremental query can also be finalized pressing Return on the keyboard.

ESP can group query results using specific document metadata. Such metadata are automatically provided by embedded document analyzers. The list of user's available grouping methods can vary and can be configured from the Control Panel in to the Collection Tab.

Within each group of results ESP can use various sorting criteria. The Relevance criteria sorts results according the internal searchbox relevance algorithm. Other available methods are Time (sorting by increasing or decreasing timestamp) and MIME (sorting by document format) but more in general ESP can perform grouping by any metadata injected into documents by plugins and/or metadata templates.

ESP can also return results within a specific time interval. This control let users choosing the desidered time interval starting from the current date in a very immediate way. The document timestamp is set by default as equal to time when the document is fetched from the source.

Each group of results can be expanded or collapsed using the rotating arrow located at the left side of any group header bar. ESP automatically decide to show some groups as collapsed in order to show all available groups in a single page.

For each returned document more detailed info can be obtained. Clicking on the "i" sign the corresponding row is expanded showing the actual MIME type string and the original context where the search string has been found into the document.