The TACTweb Query Language

The TACTweb Query Language is very similar to that used in TACT's UseBase program. If you know this specification language, you already know how to specify queries in TACTweb.

The Basics

1. Word forms can be entered directly. To enter more than one form in a query, separate them by commas:

moon, sun (displays occurrences of the words "moon" and "sun")
"when (use a single" to specify a query language reserved word like "when")

2. Use the regular expression notation ".*" to specify a wildcard sequence:

abs.* (displays words beginning with "abs")
.*ent (displays words ending with "ent")
c.*ons (displays words beginning with "c", ending with "ons")
.*ite.* (displays words containing "ite")

To see more on regular expressions, click here.

3. Use the | operator to request a phrase:

my | lov.* (displays phrases beginning with the word "my" and followed immediately by any word starting with "lov")
my | (displays all words that immediately follow the word "my")

The discussion below introduces some variations.

4. Request co-occurrence patterns with the operators "&" or "~":

moon & star.* (displays all occurrences of "moon" that occur near to a word beginning "star")
moon ~ star.* (displays all occurrences of "moon" that do not occur near a "star" word)

For more information about the co-occurrence syntax, click here.

5. Any of the above selection tools can be followed by a "when" refinement:

moon.* ; when speaker = bottom (selects only those words starting with the letters "moon" when spoken by the character "Bottom")

Both the semi-colon and the word "WHEN" are required. For more information, click here.

The TACTweb Query in Detail

All TACTweb selection queries start with some type of initial selection called a source. Some also contain refinements or filters that modify the initial selection. The general form is:

 
<sources> ; <refinements>
 
 

Note that the refinement section is optional, but, if used, must be preceded by a semicolon.

Selection Sources

The source can consist of several selection items separated by commas:

 
source 1 , source 2 , ...
 
 

The following are legal sources:

• Vocabulary Words: Any vocabulary word can be given. If the word is reserved within the TACTweb query language, precede it with an (unmatched) '"', e.g. "when.
• Regular Expressions: Regular expressions allow you to select word forms based on letter patterns within that word. To see information on how to formulate regular expressions, click here.
• Frequency: The FREQ selection criteria can be used as a source or a refinement to select words according to frequency of occurrence. For more information about it click here.
• SIMIL: The SIMIL selector selects word types based on their similarity to a given pattern, e.g.: "SIMIL love 70%". This selects words with letter sequences that are more than 70% similar to "love".
• Phrase Selection: Any of the above selection tools can be combined into phrases by joining them with the "|" operator, e.g. "my | lov.*" chooses phrases beginning with the word "my" and followed immediately by any word starting with "lov". The ">" character can be used to select positions other than the first word in the phrase: e.g. "my | > lov.*" will save the "lov" words. The ">" character may be placed by itself before or after the phrase: "> | my | lov.*" chooses all word immediately before the phrase; "my | > | lov.*" chooses all phrases beginning with the word "my", followed by any word, and then followed by a "lov" word. The middle, unspecified, word is saved in the resulting list.
• Co-occurrence Selection: Any of the above selection tools can be combined into expressions involving the co-occurrence operators "&", "~". For more information about the co-occurrence syntax, click here.
• Exclusion: The "-" operator can be used: <source 1> - <source 2> produces a list of items that are in <source 1> but are not in <source 2>.

Selection Refinements

A refinement can follow any source or source group. It begins with a semicolon. More than one refinement can be given -- in which case separate each by a semi-colon:

 
<sources> ; <refinement> ; <refinement> ...
 
 

There are several different refinement operators to choose from. All refinement operators begin with a name, and some are followed by some further specification information.

• FREQ: The FREQ operator can be used as a refinement. For more information about it, click here.
• POS: The POS operator needs no further specification information. It transforms each word (type) entry it is given into a list of positions where that word occurs: e.g: "earth; pos"
• REGEXP: Regular expressions can be used as a refinement, e.g.: "freq 1; REGEXP a.*" chooses all word forms occurring once that start with the letter "a". For more information on regular expression syntax, click here.
• WHEN: Can be used to select positions based on structural information in the text. For more information, click here.
• SIMIL: The SIMIL operator (described above) can also act as a refinement, e.g.: "l.*; simil love 75%" will select all words beginning with "l" that are more than 75% similar to "love".
• SPAN: The SPAN operator creates a resulting list that consists of all word tokens "near" to the given inputted positions ("near" is defined by the Context associated with the query). If the configuration context was 5 words on either side, for example, the selection "love; span" would select all words that occur within 5 words of "love" in the text.

Web design Alex Stevens, content Geoffrey Rockwell or John Bradley. March 7, 1997