Back to results
US 2003/0154066 A1Published application
System and method for accurate grammar analysis using a part-of-speech tagged (POST) parser and learners' model
AssigneeSunflare Co., Ltd.
InventorsNaoyuki Tokuda, Liang Chen, Hiroyuki Sasai
Life of this patent
Estimating the term…
Abstract
An accurate grammar analyzer based on a so-called POST (part-of-speech tagged) parser and a learners' model for use in automated language learning applications such as the template-based ICALL (intelligent computer assisted language learning) system.
Claims2
A method of analyzing grammar using a part-of-speech tagged parser with a template-based computer-assisted learning system comprising the steps of:
- preassigning words and phrases of an input sentence with part-of-speech tags;
- regarding any phrase preassigned with a part-of-speech tag as one word;
- setting a probability of preassigned words having a given tag as 1;
- setting a probability of nonassigned words having a given tag as 0;
- obtaining a plurality of grammar trees;
- finding a combination within said plurality of grammar trees to maximize probability of the final grammar tree being any of a grammar tree with the following formula to choose the one with largest probability Ptree:
- 3 P tree ( T ) = ∏ rule i inT P rule i · ∏ tag j of word j inT ( P ( tag j | word j ) ) 2
- where Prulei denotes the probability of a rule to take on rulei, P(tagj|wordj) is the probability of the wordj being assigned to part-of-speech tag be tagj.
5 more elements
A method of applying a part-of-speech tagged (POST) parser in a template-automaton-based computer-assisted language learning system, comprising the steps of:
- reading a keyed-in sentence;
- checking the sentence with a standard spell check model, and correcting spelling errors;
- finding a best matched path having a highest similarity value with the input sentence, providing lexical error information, feedback information as well as a score of the input sentence;
- according to the error feedback information, finding a correct path in the template;
- applying the POST parser to obtain a syntactically bracketed grammar structure for the correct path;
- and
- drawing the parsed tree of the correct path and marking the errors at leaves of the relevant tree.
4 more elements
Dependencies are inferred from the claim text.