# US 7,613,601 B2 · Method for predicting negative example, system for detecting incorrect wording using negative example prediction

> Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt

- Number: US 7,613,601 B2
- Document ID: US7613601B2
- Status: Granted
- Kind: B2
- Application number: 10/187,785
- Filed: 2002-07-03
- Published: 2003-06-26
- Granted: 2009-11-03
- Earliest claimed: 2001-12-26
- Assignees: National Institute of Information and Communications Technology
- Inventors: Masaki Murata; Hitoshi Isahara
- Web page: https://patentgrep.com/patent/US7613601B2

## CPC

- G06F 40/253: Grammatical analysis; Style critique
- Y10S 707/99937: Sorting
- G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N)
- Y10S: TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS

## Abstract

An negative example prediction processing method for predicting a likelihood of examples being negative for data where, with respect to a certain problem, it is not known whether the data is for a correctly worded positive example or for an incorrectly worded negative example. In this negative example prediction processing method, an unknown example x is inputted and a determination is made as to whether or not the example x exists in a positive example database provided in advance. If the example x does not exist, a typical probability of appearance p(x) for the example x is calculated, and a likelihood Q (x) of the example x being an negative example is calculated from the probability of appearance p(x).

## Claims

### Claim 1

A computer readable medium storing an executable code to perform on a computer a method of determining whether input data is correct, the method comprising:
storing a positive example data group D, including a number n of positive examples, wherein a positive example is known to be a correct example;
selecting example data x from the input data, which example data x consists of two independent parts a and b;
determining whether the example data x is stored in the positive example data group D;
calculating probabilities P(a) and P(b) based on appearance of part a and of part b of the example data x, respectively, in the positive example data group D, when the example data x is not stored in the positive example data group D, and determining a first probability P(x)=P(a)·P(b) that the example data x should be stored as a positive example in the positive example data group D; and
estimating a likelihood that the example data x is a negative example, any negative example being purposefully excluded from the positive example data group D, by calculating a second probability Q(x) using the first probability P(x), wherein Q(x)=1−(1−P(x))n the second probability representing likelihood that the example data x should be stored as any one of n positive examples in the positive example data group D when the example data x is not currently stored as a positive example;
determining the example data x is a negative example, if the second probability is larger than a predetermined value and the example data x is not currently stored as a positive example, and
outputting the example data x if the determining indicated that the example data x is a negative example.

### Claim 2 (depends on 1)

The computer readable medium storing an executable code to perform on a computer a method of determining whether input data is correct according to claim 1, the method further comprising:
arranging a plurality of example data x from the input data in a descending or an ascending order based on their respective second probability; and
outputting the plurality of example data x not present in the positive example data group D in the descending or the ascending order generated in the arranging of the plurality of example data x.

### Claim 3 (depends on 1)

The computer readable medium storing an executable code to perform on a computer a method of determining whether input data is correct according to claim 1, further comprising: displaying the input data wherein a plurality of example data are displayed with a changed color or brightness or displaying under different displaying conditions according to an ascending or a descending order based on their respective second probability.

### Claim 4 (depends on 1)

The computer readable medium storing an executable code to perform on a computer a method of determining whether input data is correct according to claim 1, wherein the outputting outputs the likelihood that the example data x is a negative example together with the example data x output as a negative example.

### Claim 5

A computer executable method of determining whether input data is correct, performed on a computer the method comprising:
storing a positive example data group D, including a number n of positive examples, wherein a positive example is known to be a correct example, in the computer;
selecting example data x from the input data, which example data x consists of two independent parts a and b;
determining whether the example data x is stored in the positive example data group D;
calculating probabilities P(a) and P(b) based on appearance of part a and of part b of the example data x, respectively, in the positive example data group D, when the example data x is not stored in the positive example data group D, and determining a first probability P(x)=P(a)·P(b) that the example data x should be stored as a positive example in the positive example data group D; and
estimating a likelihood that the example data x is a negative example, any negative example being purposefully excluded from the positive example data group D, by calculating a second probability Q(x) using the first probability P(x), wherein Q(x)=1−(1−P(x))n the second probability representing likelihood that the example data x should be stored as any one of n positive examples in the positive example data group D when the example data x is not currently stored as a positive example;
determining the example data x is a negative example, if the second probability is larger than a predetermined value and the example data x is not currently stored as a positive example, and
displaying the example data x on the computer, if the determining indicated that the example data x is a negative example.