Evaluation element recognition method, evaluation element recognition apparatus, and evaluation element recognition system

Patent metadata

Field	Value
Publication number	US-2017337182-A1
Application number	US-201715597501-A
Country	US
Kind code	A1
Filing date	May 17, 2017
Priority date	May 23, 2016
Publication date	Nov 23, 2017
Grant date	—

Abstract

A method, an apparatus and a system for recognizing an evaluation element are provided. The method includes receiving an input text; performing, using a first conditional random field model, first recognition for the input text to obtain a first recognition result, the first recognition result including a pre-evaluation element that is recognized by using the first conditional random field model; performing, using a second conditional random field model, second recognition for the input text to obtain a second recognition result, the second recognition result including a false positive evaluation element that is recognized by using the second conditional random field model, the false positive evaluation element being an element erroneously detected as an evaluation element; and recognizing, based on the first recognition result and the second recognition result, an evaluation element in the input text.

First claim

What is claimed is:

1 . An evaluation element recognition method comprising:
receiving an input text; performing, using a first conditional random field model, first recognition for the input text to obtain a first recognition result, the first recognition result including a pre-evaluation element that is recognized by using the first conditional random field model; performing, using a second conditional random field model, second recognition for the input text to obtain a second recognition result, the second recognition result including a false positive evaluation element that is recognized by using the second conditional random field model, the false positive evaluation element being an element erroneously detected as an evaluation element; and recognizing, based on the first recognition result and the second recognition result, an evaluation element in the input text.

2 . The evaluation element recognition method according to claim 1 ,
wherein before performing the first recognition for the input text using the first conditional random field model, the evaluation element recognition method further includes obtaining a plurality of sets of first training data, each set of the first training data including a text and an evaluation element labeled in the text; generating a feature of the first conditional random field model; and estimating, using the plurality of sets of first training data, a weight of the feature of the first conditional random field model, and generating, based on the estimated weight, the first conditional random field model.

3 . The evaluation element recognition method according to claim 2 ,
wherein the feature of the first conditional random field model is a word-level feature.

4 . The evaluation element recognition method according to claim 1 ,
wherein before performing the second recognition for the input text using the second conditional random field model, the evaluation element recognition method further includes obtaining a plurality of sets of second training data, each set of the second training data including a text and a false positive evaluation element labeled in the text; generating a feature of the second conditional random field model; and estimating, using the plurality of sets of second training data, a weight of the feature of the second conditional random field model, and generating, based on the estimated weight, the second conditional random field model.

5 . The evaluation element recognition method according to claim 4 ,
wherein the feature of the second conditional random field model is a sentence-level feature.

6 . The evaluation element recognition method according to claim 1 ,
wherein recognizing the evaluation element in the input text based on the first recognition result and the second recognition result includes determining that the pre-evaluation element is an evaluation element, when the pre-evaluation element is not the false positive evaluation element.

7 . The evaluation element recognition method according to claim 1 ,
wherein the first recognition result further includes a marginal probability p1 of the pre-evaluation element that is recognized by using the first conditional random field model, and the second recognition result further includes a marginal probability p2 of the false positive evaluation element that is recognized by using the second conditional random field model, and wherein recognizing the evaluation element in the input text based on the first recognition result and the second recognition result includes determining that the pre-evaluation element is an evaluation element, when a ratio between p1 and p2 is greater than a predetermined threshold, and determining that the pre-evaluation element is not an evaluation element, when the ratio between p1 and p2 is not greater than the predetermined threshold.

8 . An evaluation element recognition apparatus comprising:
a memory storing computer-readable instructions; and one or more processors configured to execute the computer-readable instructions such that the one or more processors are configured to
receive an input text;
perform, using a first conditional random field model, first recognition for the input text to obtain a first recognition result, the first recognition result including a pre-evaluation element that is recognized by using the first conditional random field model;
perform, using a second conditional random field model, second recognition for the input text to obtain a second recognition result, the second recognition result including a false positive evaluation element that is recognized by using the second conditional random field model, the false positive evaluation element being an element erroneously detected as an evaluation element; and
recognize, based on the first recognition result and the second recognition result, an evaluation element in the input text.

9 . The evaluation element recognition apparatus according to claim 8 ,
wherein before performing the first recognition for the input text using the first conditional random field model, the one or more processors are further configured to obtain a plurality of sets of first training data, each set of the first training data including a text and an evaluation element labeled in the text; generate a feature of the first conditional random field model; and estimate, using the plurality of sets of first training data, a weight of the feature of the first conditional random field model, and generating, based on the estimated weight, the first conditional random field model.

10 . The evaluation element recognition apparatus according to claim 8 ,
wherein before performing the second recognition for the input text using the second conditional random field model, the one or more processors are further configured to obtain a plurality of sets of second training data, each set of the second training data including a text and a false positive evaluation element labeled in the text; generate a feature of the second conditional random field model; and estimate, using the plurality of sets of second training data, a weight of the feature of the second conditional random field model, and generating, based on the estimated weight, the second conditional random field model.

11 . An evaluation element recognition system comprising:
an input apparatus configured to receive an input text; an analyzing apparatus; and an output apparatus configured to output a recognition result of the analyzing apparatus, wherein the analyzing apparatus includes a memory storing computer-readable instructions; and one or more processors configured to execute the computer-readable instructions such that the one or more processors are configured to
perform, using a first conditional random field model, first recognition for the input text to obtain a first recognition result, the first recognition result including a pre-evaluation element that is recognized by using the first conditional random field model;
perform, using a second conditional random field model, second recognition for the input text to obtain a second recognition result, the second recognition result including a false positive evaluation element that is recognized by using the second conditional random field model, the false positive evaluation element being an element erroneously detected as an evaluation element; and
recognize, based on the first recognition result and the second recognition result, an evaluation element in the input text.

Assignees

• Jiang Shanshan
• Dong Bin
• Zheng Jichuan
• Zhang Jiashi
• Tong Yixuan
• Ricoh Co Ltd

Inventors

• Jiang Shanshan
• Dong Bin
• Zheng Jichuan
• Zhang Jiashi
• Tong Yixuan

Classifications

• G06F40/44

  Statistical methods, e.g. probability models · CPC title

• G06F40/279 Primary

  Recognition of textual entities · CPC title

• G06F40/30 Primary

  Semantic analysis · CPC title

• G06F40/205 Primary

  Parsing · CPC title

• G06F17/2765

  Physics · mapped topic