US 9,542,383 B2Granted

Example-based error detection system for automatic evaluation of writing, method for same, and error detection apparatus for same

AssigneeSK Telecom Co., Ltd.

InventorsSeunghwan Kim, Eunsook Lee, Seongmook Kim, Dongnam Kim, Sung Kim

Life of this patent

Estimating the term…

Loading patent…

Abstract

An error detection system for automatically evaluating writing includes: an example construction apparatus to collect example sentences including various literary styles, to break up the collected example sentences in units of morphemes, and to construct the example sentences in an example-based index database (DB); and an error detection apparatus to break up an input sentence in units of morphemes, to generate one or more morpheme sequences bound in arbitrary window sizes based on one or more of morphemes of the broken-up input sentence, to search the example-based index DB for each of the generated morpheme sequences, and to detect an error according to a frequency at which said each morpheme is arranged in a corresponding morpheme sequence among morpheme sequences searched for through the example-based index DB.

Claims11

Dependent claims

An error detection system for automatically evaluating writing, the system comprising:

an example construction apparatus, implemented by one or more processors and/or application-specific integrated circuits (ASICs), configured
to collect example sentences including various literary styles,
to break up the collected example sentences in units of morphemes, and
to construct the example sentences in an example-based index database (DB);
and
an error detection apparatus, implemented by one or more processors and/or application-specific integrated circuits (ASICs), configured
to break up an input sentence in units of morphemes,
to generate one or more morpheme sequences by binding one or more morphemes of the broken up input sentence in a forward direction or a backward direction in an arbitrary window size based on each morpheme to which the part of speech has been tagged, wherein the one or more morpheme sequences includes (i) forward morpheme sequences generated by binding morphemes positioned in the forward direction in the arbitrary window size and (ii) backward morpheme sequences by binding morphemes positioned in the backward direction in the arbitrary window size, and the arbitrary window size is determined based on a size of each morpheme sequence, and
wherein the one or more morpheme sequences are generated by varying the arbitrary window size,
to search the example-based index DB for each of the one or more morpheme sequences,
to extract a frequency at which each morpheme searched is arranged from the one or more morpheme sequences;
to calculate a frequency for occurring to arrange each morpheme of the broken-up input sentence in a corresponding morpheme sequence among the one or more morpheme sequences searched for through the example-based index DB, and
to detect an error section based on the calculated frequency, and by determining whether there is a section having a compatibility score for each morpheme that is lower than a threshold, or decreased by a predetermined value or more,
wherein the error detection apparatus is further configured to calculate the compatibility for each of the forward morpheme sequences and the backward morpheme sequences generated, and to calculate integrated compatibility reflecting the compatibility.

11 more elements

An error detection apparatus, including one or more processors and/or application-specific integrated circuits (ASICs), the error detection apparatus comprising:

an input sentence break-up unit, implemented by the one or more processors and/or the ASICs, and configured to break up an input sentence in units of morphemes;
a part-of-speech tagging unit, implemented by the one or more processors and/or the ASICs, and configured to tag a part of speech to each of the morphemes of the broken up input sentence;
and
an error detecting unit, implemented by the one or more processors and/or the ASICs, and configured
to generate one or more morpheme sequences by binding one or more morphemes of the broken up input sentence in a forward direction or a backward direction in an arbitrary window size based on each morpheme to which the part of speech has been tagged, wherein the one or more morpheme sequences includes (i) forward morpheme sequences generated by binding morphemes positioned in the forward direction in the arbitrary window size and (ii) backward morpheme sequences by binding morphemes positioned in the backward direction in the arbitrary window size, and the arbitrary window size is determined based on a size of each morpheme sequence,
to vary the arbitrary window size for generating the one or more morpheme sequences,
to search an example-based index database (DB) for each of the one or more morpheme sequences,
to calculate a frequency for occurring to arrange each morpheme of the broken-up input sentence in a corresponding morpheme sequence among the one or more morpheme sequences searched for through the example-based index DB, and
to detect an error section based on the calculated frequency by determining whether the calculated compatibility is lower than a threshold or whether the calculated compatibility has been decreased by a predetermined value or more, and
wherein the error detecting unit includes:
wherein the error detecting unit includes:
an N-gram generating unit, implemented by the one or more processors and/or the ASICs, and configured to generate, based on said each morpheme,
the forward morpheme sequences by binding one or more morphemes of the broken up input sentence in the forward direction in the arbitrary window size,
the backward morpheme sequences by binding one or more morphemes of the broken up input sentence in the backward direction in the arbitrary window size, and
a compatibility calculating unit, implemented by the one or more processors and/or the ASICs, and configured to
extract the frequency at which said each morpheme is arranged from morpheme sequences searched for by a search unit,
calculate for each of the forward morpheme sequences and the backward morpheme sequences generated by the N-gram generating unit, and
calculate integrated compatibility reflecting the compatibility.

15 more elements

An error detection method of automatically evaluating writing performed by an error detection apparatus, the error detection apparatus implemented by one or more processors and/or application-specific integrated circuits (ASICs) to execute the method comprising:

breaking up an input sentence in units of sentences and morphemes;
tagging a part of speech to each of the morphemes of the broken up input sentence;
generating one or more morpheme sequences by binding one or more morphemes positioned in a forward direction or a backward direction in an arbitrary window size for said each morpheme to which the part of speech has been tagged, wherein the one or more morpheme sequences includes (i) forward morpheme sequences generated by binding morphemes positioned in the forward direction in the arbitrary window size and (ii) backward morpheme sequences by binding morphemes positioned in the backward direction in the arbitrary window size, and the arbitrary window size is determined based on a size of each morpheme sequence;
varying the arbitrary window size for generating the one or more morpheme sequences;
searching an example-based index data base (DB) for each of the one or more morpheme sequences;
calculating a frequency for occurring to arrange each morpheme of the broken-up input sentence in a corresponding morpheme sequence among the one or more morpheme sequences searched for through the example-based index DB;
and
detecting an error section based on the calculated frequency by determining whether the calculated compatibility is lower than a threshold or whether the calculated compatibility has been decreased by a predetermined value or more, and
wherein, when the generated morpheme sequences include both a forward morpheme sequence and a backward morpheme sequence, the calculating of the frequency comprising:
extracting the frequency at which said each morpheme is arranged in respect to the forward morpheme sequence and the backward morpheme sequence,
calculating compatibility for each of the forward morpheme sequence and the backward morpheme sequence, and
calculating integrated compatibility reflecting the compatibility.

9 more elements

Dependencies are inferred from the claim text.

Facts

Application: 14/097,209Filed 4 Dec 2013 · continuation
Priority: KR 10-2011-010981226 Oct 2011
Published: US 2014/0093845 A13 Apr 2014
Granted: 10 Jan 2017
Term ends: Not estimatedThe source records don’t support an estimate.

Classification9 codes

G06F 40/268
Natural language analysis: morphological analysis
G06F 40/232
Natural language analysis: orthographic correction, e.g. spell checking or vowelisation
G06F 40/253
Natural language analysis: grammatical analysis; Style critique
G06F 40/44
Data-driven translation: statistical methods, e.g. probability models