JP2658997B2 - Sentence evaluation device using keywords - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an answer collation method for a computer-aided education (CAI) system.

[0002]

2. Description of the Related Art Conventionally, the answer collation of a CAI system is performed by:
Simple keyword matching or LADAR
(IEICE Educational Technology Research Report ET66-6 p
p. 39-44), a keyword matching method that extends a regular expression has been used.

[0003]

In the conventional symbol processing method, since only a binary state of matching or not with a keyword is determined, the degree of matching of sentences cannot be evaluated as a numerical scale. Was.

[0004]

A sentence evaluation apparatus according to the present invention is a system for evaluating the degree of coincidence of meaning in two sentences, in which a plurality of keywords selected from a model sentence are inputted and stored. A keyword classifying means for classifying and storing the keywords into categories according to meaning, and a conditional expression inputting means for expressing and storing a relationship between keywords belonging to each category as a conditional expression using a logical expression; Weighting input means for inputting and storing weights for each category, evaluated text input means for inputting and storing the evaluated text, and sequentially determining whether the evaluated text satisfies the conditional expression for each category Keyword matching means and, for all categories that satisfy the conditional expression, give points according to the weight of the category. And matching degree determination means for calculating a coincidence degree by calculating them, characterized in that it comprises means for displaying the degree of coincidence.

[0005]

In the present invention, the original sentence is classified into categories for each content, and the result of the sentence evaluation for each category is evaluated as a function of the weighting value. It is possible to provide as a numerical value of the degree of coincidence of sentences instead of the alternative of "Is it correct?"

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of a text evaluation apparatus according to the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing one embodiment of the present invention. As shown in FIG. 1, the present invention provides a keyword input unit 01 for inputting and storing a plurality of keywords selected from a model sentence, and a keyword classifying unit 02 for classifying and storing the keywords into categories according to their meanings. When,
A conditional expression input unit 03 for expressing and storing the relationship between keywords belonging to each category as a conditional expression using a logical expression, a weight input unit 04 for inputting and storing weights for each category, and a sentence to be evaluated Evaluated sentence input means 05 for inputting and storing, keyword matching means 06 for sequentially determining whether the evaluated text satisfies the conditional expression for each category, and, for all categories satisfying the conditional expression, the category Score according to the weight of the Accuracy Index (AI
Value), and a coincidence determining means 07 for calculating the degree of coincidence as the sum of the AI values, and a means 08 for displaying the degree of coincidence.
It is composed of the following, and performs keyword matching for the evaluated text for each category determined in the model text,
It is characterized in that the sum of the AI values obtained for each category is regarded as a sentiment evaluation result as the degree of coincidence of sentiment.

Referring to FIG. 1, a description will be given of a system for scoring a question answered by a student in a question set by a teacher.

[0009] The teacher pre-divides the items to be scored from the model answer into categories. The keyword input means 01 is for a teacher to select and input a keyword from a model answer. The keyword classification means 02
The teacher categorizes keywords according to the items to be graded. The conditional expression input means 03 is for the teacher to express the relation between the keywords with a conditional expression for each category divided by the keyword classification means 02 and to input. The weighting input means 04 is provided by the
A point weight is input for the category divided by. The sentence to be evaluated input means 05 is for the learner to input the sentence of the answer to the question. The keyword matching unit 06 determines whether a keyword satisfying the conditional expression input by the conditional expression input unit 03 is included in the evaluation sentence for each category. Matching degree determining means 07
Calculates the score (AI value) based on the weight input by the weight input means 04 for the category that satisfies the condition by the keyword matching means 06, and calculates the sum of the AI values of all the categories satisfying the condition. Is calculated. The evaluation result display means 08 includes the coincidence degree determination means 0
The value calculated in step 7 is presented to the learner.

FIG. 2 is a flowchart showing the processing performed by the keyword matching means 06 and the matching degree determining means 07. First, a keyword, a conditional expression (for each category) and a weight of a model answer are acquired (S1), and a sentence to be evaluated is acquired (S2). Next, it is checked whether the keyword in the model answer data exists in the evaluated text (S
3) If the keyword is included in the sentence to be evaluated, the keyword is stored in the keyword match array (S4).
The operation is performed for all keywords (S5). next,
Based on the keywords stored in the keyword match array,
It is checked whether the conditional expression in the model answer data satisfies the conditional expression for each category (S6). If the conditional expression is satisfied, a score (AI) value according to the weight of the category is calculated (S7), If not satisfied, set the AI value to 0
And the operation of adding to the evaluation value (S9) is performed for all categories (S10). When all categories have been checked, the evaluation value is set to the degree of coincidence (S11),
finish.

Next, a specific example of the above scoring system will be described.

First, the model answer is "check for signs and symptoms of the patient's illness and observe the mental state". The keywords are "sickness", "signs", "symptoms", "mental state", and "observation". Is 1. To look for signs and symptoms of the disease,
2. Observing the mental state is two points, and the conditional expression is Category 1. Disease and (sign or symptom) Category 2. Mental state and observation, weighting 60% for category 1, category 2
To 40%.

A description will be given of a scoring method in a case where the sentence to be evaluated is “check the symptoms of the disease and observe the health condition”. The keywords included in the evaluated text are “illness”, “symptoms”, and “observation”. When these keywords are applied to the conditional expressions of the above categories, category 1 is true because the keywords of "illness" and "symptoms" are included in the evaluated text, and category 2 is "mental state" in the evaluated text. Is not included in the expression, the result is false, and it can be seen that only the conditional expression of category 1 is satisfied. Assuming the above question is a perfect score of 10, the weight of category 1 is 60%
Therefore, the score is 60% of 10 points and 6 points.

[0014]

As described above, according to the present invention,
The evaluation of a sentence can be indicated by a numerical value called a coincidence which is a composite score of the evaluation result for each category. As a result, it is possible to know to what extent the evaluated text that is the answer of the learner satisfies the requirements for the model answer.

[Brief description of the drawings]

FIG. 1 is a block diagram showing one embodiment of the present invention.

FIG. 2 is a flowchart showing the operation of the present invention.

[Explanation of symbols]

 01 Keyword Input Means 02 Keyword Classification Means 03 Conditional Expression Input Means 04 Weight Input Means 05 Evaluated Text 06 Keyword Matching Means 07 Matching Determining Means 08 Evaluation Result Display Means

Claims (1)

(57) [Claims] 1. A system for evaluating the degree of coincidence of sentence in two sentences, a keyword input means for inputting and storing a plurality of keywords selected from a model sentence, and the keywords being classified into categories according to the sentence. Keyword classification means for classifying and storing; conditional expression input means for expressing and storing the relationship between keywords belonging to each category as a conditional expression using a logical expression; and weighting input for inputting and storing a weight for each category Means, input means for inputting and storing the text to be evaluated, keyword matching means for sequentially determining whether or not the text to be evaluated satisfies the conditional expression for each category, For a category, a score corresponding to the weight of the category is given, and the score is calculated by calculating them. A meaning evaluation device, comprising: a degree-of-match determination means; and a means for displaying the degree of coincidence.