PHP Pregroup Parser : User's Manual

The Main Page

1. The first entry gives an access to the documentation.
2. With the second link, you can use the parser with your own grammar. In this case, a file containing the definition of the grammar will have to be selected. This file will be uploaded to the server each time the parser is started.
3. Alternatively, you can use one of the grammars distributed with the parser. In this case, you can enter a string or you can choose one (or all) example associated to the grammar.
4. The parser works with Unicode characters (UTF-8 encoding). Sometime, it is useful to have a grammar with only ASCII characters. This section presents ASCII version of the previous grammar. The lexicon uses several special sequences for several characters in the spirit of TeX or LaTeX. For instance π is coded by \pi.

The Parser Page

When Using Your Own Grammar

1. You must fill the first field with the name (or path) of the file that defines your grammar. The grammar will be uploaded to the server each time the parser will be started.
2. You can enter a string in the second field. The string is parsed using the given grammar when "Ok" is pressed.
3. Alternatively, you can enter the name of a file which contains one or several lines. Each line is parsed separately.
4. The last part is the parser option section which is presented bellow.

When Using A predefined Grammar

1. You can enter a string in the first field. The string is parsed using the predefined grammar when "Ok" is pressed.
2. If the grammar is also associated to a list of predefined examples, you can choose one of them in a list. The string is parsed when "Ok" is pressed. You can also click on the second button "Parse all examples" which starts the parser for each example in the list.
3. The last part is the parser option section which is presented bellow.

The Parser Options

• Show pregroup nets: When checked, the pregroup nets corresponding to correct analyzes are shown in the report.
• Show text nets: When checked, the pregroup nets are shown using only text characters. This option needs a monospace font for a correct rendering. If not checked, the nets appear as bitmap images (PNG format).
• Maximum number of viewed pregroup nets: This option gives a limit to the number of nets that is computed. In fact, even if the parser knows rapidly if there exists a net, the number of possible nets can be very high. This option gives a limit to the number of nets that are computed. If you do not want to limit their number, you must enter a negative number.
• Show only one pregroup net for each assignment of types: Sometimes, for the same list of types assigned to the words, it exist several different pregoup nets. In this case, the links are not connected at the same places. When checked, the report shows only one of these nets.
• Show Matrix: Before computing nets, the parser uses a tabular method based on majority partial composition that associates types to each sub-list of words of the input string. This option shows the intermediary result as a triangle matrix. Each cell correponds to the types that are computed for the sub-list of words that starts at the i-th word and ends at the j-th word.
• Show only useful types: A type that is computed for a sub-list of words of the input string (in the matrix) do not necessary appear in a correct analysis because the matrix is computed from the bottom to the top ans a useful type must also be reachable from the top. When this option is not checked, all the computed types computed from the bottom to this cell are shown in the matrix. If not checked, only types that appear in, at least, one correct analysis are presented in the matrix.
• All basic types correspond to a correct sentence:
• Scale of pregroup nets (percent):
• Alternative name of the lexicon on the server:
• XML input:
• Follow input tree (need XML input):
• Reset the form:

Last update: 2008-06-25