StudentPARSing Environment II
CGI-Web Demonstration Version

Introduction: This is a web demonstration version of the SPARSE II parsing program. SPARSE II (Student PARSing Environment) is a parsing program intended to be used as a pedagogical tool to help syntax students grasp the complexity of natural-language grammars and to begin developing their own models. It provides a true introduction to Natural Language Processing without requiring familiarity with Lisp or Prolog. The key feature of SPARSE II is that grammar rules are not integrated into the program: they are supplied by the user. This means that there is no underlying theoretical model in the program; it simply parses what it can with the rules available. The user is free to implement the rules according to any framework. For the more advanced user, SPARSE II is a stable platform for testing experimental grammars. The parse algorithm used by SPARSE II returns all possible parses (the web version is limited to 10), handles left recursion, optional elements and null constituents, provides feature unification using GULP notation, and displays parse trees. You can download SPARSE II to use offline if you like (it's much faster).

How to Use SPARSE II CGI-Web: This page loads with rules already in place and a sentence in the input box, so all you have to do to get started is click on the "Parse" button and see the results. When you are tired of looking at the results, use your browser's "Back" button to return here. The next step is to put your own phrase in the input box and see what it does. Before you do this you will want read the rules (see the rule box below) so that you will know what words are in the lexicon. The lexicon is not huge, but it will allow you to test all the structures the rules allow. Once you get comfortable with how SPARSE II CGI-Web works, you can add words and rules of your own to the rule box. In fact, you can put in a set that is entirely yours. Just cut and paste as you see fit. If you ever get totally fouled up, just click the "Reset Rules" button. Also, you will want to read the manual (it is pretty short) in order to understand the rule syntax. If you find any bugs or have a question that the manual won't answer, please contact the author, Clayton Darwin.

As a hint to get you started with modifying the rules, words in the lexicon are in the [square brackets] found in the first section of the rule box. You can add new words by following the same pattern: semicolon, bracket, word, bracket, semicolon, and a period at the end. If you want do more than that, read the manual. You can always reset the rules if you have a problem. Also, don't try to make these rules into something they are not: they are just here to demonstrate some of SPARSE II's capabilities. You'll have to write your own rules to do any real work.

Note: Unlike the full version of GULP, SPARSE does not support feature structures within feature structures, nor the use of { } to drop down into Prolog. SPARSE's parser is not the DCG top-down parser, so it is not clear what { } would mean if it were implemented.

Input to parse (up to 85 characters):

Parse Rules:
% ----- WORD RULES ---------------------------------------- d(num:plural) --> [the];[their];[]. d(num:sing) --> [the];[its];[a]. n(num:plural) --> [mice];[cats];[dogs];[forks];[hats]. n(num:sing) --> [mouse];[cat];[dog];[fork];[hat]. n(num:sing..gen:male) --> [rooster]. n(num:sing..gen:female) --> [hen]. v(num:sing) --> [is];[eats];[knows];[sees];[heads]. v(num:plural) --> [are];[eat];[know];[see];[head]. aux --> [do]. neg --> [not]. adj(gen:male) --> [handsome]. adj(gen:female) --> [pretty]. p --> [with];[in]. c --> [that];[]. % ----- STRUCTURAL RULES ---------------------------------- s(num:X..gen:Y) --> np(num:X..gen:Y),vp(num:X..gen:Y). np(num:X..gen:Y) --> d(num:X),n(num:X..gen:Y),[pp]. pp --> p,np. vp(num:X) --> [aux],v(num:X),[np],[pp]. vp(num:X) --> aux,neg,v(num:X),[np],[pp]. vp(num:X..gen:Y) --> v(num:X),[neg],adj(gen:Y). vp(num:X) --> v(num:X),cp. vp(num:X) --> aux,neg,v(num:X),cp. cp --> c,s.

---

This site uses SPARSE II CGI scripts written in SWI Prolog by Clayton M. Darwin
and is made available courtesy of the University of Georgia AI Center.