JP2016085284A - Program, apparatus and method for estimating evaluation level with respect to learning item on the basis of person's remark - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a program, apparatus, and method capable of estimating an evaluation level with respect to learning items as a learning state, from a text based on persons' remarks.SOLUTION: A learning state estimation program causes a computer to function in such a manner as to estimate evaluation levels with respect to learning items from a text on the basis of persons' remarks. The learning state estimation program causes the computer to function as: learning word registration means for registering a plurality of learning words on the basis of the learning items; evaluation word registration means that registers a plurality of evaluation words that a person may remark about the learning items, and evaluation levels associated with respective evaluation words; word detection means that detects learning words from the text and detects evaluation words within a predetermined distance before and after the learning words; and evaluation level estimation means for associating the learning items based on the detected learning words with the evaluation level corresponding to the detected evaluation words.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a technique for analyzing a learning situation from text based on a person's remarks.

  Conventionally, test result data has been used to measure learning status such as learning progress and comprehension. However, the test is not only time consuming and costly to implement, but it also does not provide a detailed picture of the student's tasks and motivation. Therefore, in individual classes by private tutors and group classes in schools and cram schools, teachers or group leaders write reports as records of learning status after the classes.

  These test results data and learning status reports are used to interview student parents and to determine future course policies. However, there are differences in the content and amount of writing in the learning status reports depending on the individual teachers and group leaders. In addition, it was difficult to use the learning status report as an objective material compared to the test result data.

  For example, there is a technology for collecting lesson contents and supporting creation of an electronic notebook (see, for example, Patent Document 1). According to this technique, an input voice signal is converted into text by voice recognition, and a registered word is detected from the text. Then, the registered word is highlighted on the text, or the user is assisted by sending a warning signal. However, it is a human being that creates a summary note and is not automated.

  There is also a technique for extracting important words from the dialogue (see, for example, Patent Document 2). According to this technique, first, utterances in a dialogue are divided and common appearance words are extracted. The more frequently the word appears and the closer the utterance interval of the word, the more important words are extracted. However, it only extracts important words and does not summarize utterance contents.

  Further, there is a technique for objectively estimating the amount of knowledge of the questioner (see, for example, Patent Document 3). According to this technique, the utterance of a question or an explanation is discriminated, and the knowledge amount of the questioner is estimated based on the utterance state, the feature amount of the number of vocabulary used and the estimated information learned in advance. However, according to this technique, the utterance contents are not collected.

  There is also a technique for automatically generating a summary (see Non-Patent Document 1, for example). According to this technique, the summarization rate is as high as about 1/4, and it is not possible to summarize only the essential points.

JP 2014-085998 A JP 2011-248409 A JP 2013-167765 A

Yasutoshi Fujii, “Automatic Summarization of Lecture Speech Based on Extracting Important Sentences”, IPSJ Transactions Vol. 51 No. 3, 1094-1106 (Mar. 2010), [online], [October 18, 2014 Search], Internet <URL: http://ci.nii.ac.jp/naid/110007970712>

  According to conventional individual classes and group classes, it is not possible to estimate the degree of understanding that will be a learning result unless an examination is conducted after class. On the other hand, the implementation of the test is not only time consuming and expensive, but also cannot grasp the chronological learning status of the students.

  On the other hand, can the inventors of the present application estimate the evaluation level (for example, degree of understanding) for a learning item as a learning situation from a text based on a person's remarks? I thought. According to the technologies described in Patent Documents 1 to 3 described above, it is possible to extract important words or estimate the amount of knowledge from text based on a person's remarks, but to estimate the evaluation level for a learning item. I can't. Further, according to the technique described in Non-Patent Document 1, it is possible to summarize the whole sentence, but it is impossible to summarize and estimate the person's learning situation.

  Accordingly, an object of the present invention is to provide a program, an apparatus, and a method capable of estimating an evaluation level for a learning item as a learning situation from a text based on a person's remarks.

According to the present invention, from a text based on a person's remarks, a learning situation estimation program for causing a computer to function to estimate an evaluation level for a learning item,
Learning word registration means for registering a plurality of learning words based on learning items;
An evaluation word registration means for registering a plurality of evaluation words that will be spoken with respect to the learning item and an evaluation level associated with each evaluation word;
A word detection means for detecting a learning word from the text and detecting an evaluation word from within a predetermined distance between the front and rear of the learning word;
The computer is caused to function as an evaluation level estimation means for associating a learning item based on a detected learning word with an evaluation level corresponding to the detected evaluation word.

According to another embodiment of the learning status estimation program of the present invention,
When the learning word is detected by the word detection means, the learning word is recorded in association with a time-series code number,
The evaluation level estimating means preferably causes the computer to function so as to output the evaluation level in time series for each learning word.

According to another embodiment of the learning status estimation program of the present invention,
The evaluation level estimation means preferably causes the computer to function so that the evaluation level is lowered as the distance between the learning word detected by the word detection means and the evaluation word increases.

According to another embodiment of the learning status estimation program of the present invention,
The evaluation level estimation means derives, as an evaluation level, a value obtained by multiplying the reciprocal of the interword distance between the learning word and the evaluation word detected by the word detection means and the evaluation level of the evaluation word. It is also preferable to make the computer function.

According to another embodiment of the learning status estimation program of the present invention,
It further has a morphological analysis means for dividing the text into morphemes,
Preferably, the word detection means further causes the computer to detect a learning word and an evaluation word from the classified morphemes.

According to another embodiment of the learning status estimation program of the present invention,
The learning item further includes a learning word conversion table in which a problem number and a learning word are associated with each other,
The learning word registration means registers the problem number as a learning word,
It is preferable that the evaluation level estimation means further causes the computer to function so as to convert the problem number into a learning word using the conversion table for the detected problem number.

According to another embodiment of the learning status estimation program of the present invention,
The text is also preferably a text obtained by converting a speech signal uttered by a person by speech recognition.

According to another embodiment of the learning status estimation program of the present invention,
It is preferable that the word detecting means further causes the computer to function so as to use a time width between the learning word and the evaluation word as the distance between words.

According to another embodiment of the learning status estimation program of the present invention,
Learning image recording means for recording a learning image taken based on the learning item;
Image recognition means for extracting a learning word from the learning image by image recognition;
The learning word registration means registers the pronoun as a learning word,
It is also preferable that when the evaluation level estimation unit detects the indicating pronoun, the learning image acquisition unit acquires the learning image, and the image recognition unit further causes the computer to recognize a learning word by image recognition from the learning image. .

According to another embodiment of the learning status estimation program of the present invention,
A sound transition table in which the pitch / volume transition of the audio signal is associated with the evaluation level correction value;
Preferably, the evaluation level estimation means derives an evaluation level correction value from the pitch / volume transition of the audio signal using the sound transition table, and causes the computer to function so as to correct the evaluation level derived from the evaluation word.

According to another embodiment of the learning status estimation program of the present invention,
Preferably, the text further causes the computer to function so that a person has written his or her statements with a keyboard or electronic pen.

According to the present invention, an apparatus for estimating an evaluation level for a learning item from text based on a person's remarks,
Learning word registration means for registering a plurality of learning words based on learning items;
An evaluation word registration means for registering a plurality of evaluation words that will be spoken with respect to the learning item and an evaluation level associated with each evaluation word;
A word detection means for detecting a learning word from the text and detecting an evaluation word from within a predetermined distance between the front and rear of the learning word;
Evaluation level estimation means for associating a learning item based on the detected learning word with an evaluation level corresponding to the detected evaluation word is provided.

According to the present invention, using a device, a method for estimating an evaluation level for a learning item from text based on a person's remarks,
The device
A learning word registration unit that registers a plurality of learning words based on learning items;
An evaluation word registration unit that registers a plurality of evaluation words that will be remarked with respect to a learning item and an evaluation level associated with each evaluation word;
A first step of detecting a learning word from the text and detecting an evaluation word from within a predetermined distance between the front and rear of the learning word;
And a second step of associating a learning item based on the detected learning word with an evaluation level corresponding to the detected evaluation word.

  According to the program, apparatus, and method of the present invention, it is possible to estimate an evaluation level for a learning item as a learning situation from text based on a person's remarks.

It is a basic 1st functional block diagram in the learning condition estimation apparatus of this invention. It is explanatory drawing showing the example of the learning word registered into the learning word registration part of this invention. It is explanatory drawing showing the example of the evaluation word registered into the evaluation word registration part of this invention. It is the list | wrist produced | generated by this invention. It is a specific 2nd functional block diagram in the learning condition estimation apparatus of this invention. It is operation | movement explanatory drawing using the learning word conversion table of this invention. It is operation | movement explanatory drawing using the camera mounted in the apparatus of this invention. It is operation | movement explanatory drawing using the sound transition table of this invention. It is a functional block diagram of the evaluation degree estimation system in this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a basic first functional configuration diagram in the learning situation estimation apparatus of the present invention.

  FIG. 1 shows a learning status estimation device 1 that estimates an evaluation level for a learning item from text based on a person's remarks. The learning status estimation device 1 may function as a terminal or may function as a server. In addition, according to the embodiment, a tablet terminal or a smartphone that is preliminarily equipped with a user interface such as a microphone or a camera is also preferable.

  According to the learning state estimation device 1 of FIG. 1, the basic functional configuration of the present invention is a learning word registration unit 101, an evaluation word registration unit 102, a morpheme analysis unit 11, a word detection unit 12, And an evaluation level estimation unit 13. These functional components function as a learning situation estimation program that causes a computer installed in the apparatus to function. Moreover, the flow of processing of these functional components can be understood as a learning state estimation method.

  The learning state estimation device 1 inputs text (learning recording text) based on a person's remarks via an interface. The “utterance” may be a speech uttered by a person, or a character or a sentence described by a person. In addition, “text” is the text of teachers 'and students' remarks regarding individual learning and group lessons in learning items.

[Learning Word Registration Unit 101]
The learning word registration unit 101 registers a plurality of learning words based on learning items. The “learning word” refers to a word indicating a unit of a learning item or a problem, for example.

  FIG. 2 is an explanatory diagram illustrating an example of learning words registered in the learning word registration unit of the present invention. According to FIG. 2, a learning word as a learning item is registered for each grade of junior high school students.

[Evaluation word registration unit 102]
The evaluation word registration unit 102 registers a plurality of evaluation words that will be remarked with respect to the learning item and an evaluation level associated with each evaluation word.
An “evaluation word” is a word that represents a person's understanding of a learning item, such as “can / can't” or “correct / false”. Examples of the evaluation words include teacher's instruction words and emotion words. Further, the evaluation word and the evaluation level are not limited to those related to human understanding, and may be related to, for example, human interest / preference, motivation, depression, and the like. The “evaluation level” is set in a plurality of stages of two or more values.

  FIG. 3 is an explanatory diagram illustrating an example of evaluation words registered in the evaluation word registration unit of the present invention.

According to FIG. 3, the evaluation level is set in four stages.
Strong positive evaluation: +2
Normal positive rating: +1
Normal negative rating: -1
Strong negative evaluation: -2

  As another embodiment, the learning word registration unit 101 and the evaluation word registration unit 102 are not limited to the estimation of the learning state, and may register words according to uses other than learning.

[Morphological analyzer 11]
The morpheme analyzer 11 classifies the text input from the interface into morphemes. A “morpheme” refers to the smallest meaningful unit among the constituent elements of a sentence. The morpheme analysis is divided into morphemes using a dictionary in which “part of speech” and “reading” are registered for each “word”. The divided morphemes are output to the word detection unit 12.

[Word detection unit 12]
First, the word detection unit 12 detects the “learned word” registered in the learned word registration unit 101 from the text divided into morphemes by pattern matching. Next, an “evaluation word” registered in the evaluation word registration unit 102 is detected within a predetermined distance (proximity) between the front and rear of the learning word. The shorter the distance between words (the higher the degree of proximity), the closer the learning word and the evaluation word are.

  Here, the “inter-word distance” is not an inter-word distance indicating a semantic relevance using a so-called thesaurus, or an inter-language distance indicating a difference between foreign languages. The inter-word distance of the present invention may be, for example, the number of words between a learning word and an evaluation word.

  Further, the “distance between words” may be measured by limiting within one sentence (sentence). Simply, a pair of a learning word and an evaluation word often has a relationship between a subject and a predicate. The inter-word distance based on the number of words may be determined that the distance based on the order of the languages is short even if there is no utterance and the time is separated. On the other hand, in the case of sentence units, since one meaning is collectively included, a close relationship can be detected properly.

  Further, the evaluation word is detected from within a predetermined distance between “front” and “back” of the learning word. At this time, it is preferable to detect the evaluation word from “behind” the learning word. FIG. 1 shows an example of the relevance ratio for the transcribed text. The relevance rate is a result of determining whether the result output from the evaluation level estimation unit 13 in the apparatus 1 of FIG. Here, it can be understood that the relevance rate of the backward evaluation word is generally higher than that of the front evaluation word. Moreover, it can be understood that the inter-word distance has the highest relevance rate of about 6 to 10 characters.

  As another embodiment, when the learning word is detected, the word detection unit 12 preferably records the learning word in association with a time-series code number. For each learning item, evaluation words representing the degree of understanding and progress of students can be generated as a time-series list. Such a list can also be used in teaching policies for the student.

[Evaluation level estimation unit 13]
The evaluation level estimation unit 13 associates an evaluation level corresponding to the detected evaluation word with a learning item based on the detected learning word. As a result, for each learning item, an evaluation level representing the degree of understanding and progress of the student can be generated as a time-series list.

  FIG. 4 is a list generated by the present invention.

  FIG. 4A is a list a in which learning words, evaluation words, and evaluation levels detected from text are associated with each other. According to the list a, a time series number is assigned to the learning word, and an inter-word distance is assigned to the evaluation word. And the evaluation level is provided in the range of + 2-2 for every pair of a learning word and an evaluation word. According to FIG. 4A, two evaluation words are associated with each other, but it is sufficient that at least one evaluation word is associated with each other.

FIG. 4B is a list b in which an evaluation level and a reference evaluation word are associated with each learning item. The evaluation level of the learning item may be, for example, an average value at a plurality of evaluation levels.
Evaluation level of learning item = Σ _i ^N (Evaluation level x Inverse of distance between words) / N
N: Number of occurrences of the learning word in the learning item Thereby, the evaluation level can be derived so that the evaluation level becomes lower as the distance between the learning word and the evaluation word becomes longer.

  FIG. 5 is a specific second functional configuration diagram in the learning situation estimation apparatus of the present invention.

  5, the microphone 103, the text input interface (keyboard / electronic pen) 104, the display 105, the camera 106, the voice recognition unit 14, the learning word, in addition to the functional configuration unit shown in FIG. It further includes a conversion table 15, a learning image recording unit 16, an image recognition unit 17, and a sound transition table 18.

[Microphone 103]
The microphone 103 collects speech sounds of teachers and students in individual learning and group lessons. In order to suppress ambient noise and to be able to discriminate even if speech speech overlaps, it is preferably a proximity microphone. The collected voice is output to the voice recognition unit 14.

[Voice recognition unit 14]
The voice recognition unit 14 outputs a text obtained by converting a voice signal uttered by a person through voice recognition. The text is input to the word detection unit 12 via the morphological analysis unit 11. In order to improve the word recognition rate of speech recognition, it is preferable to provide a word dictionary relating to a learning situation that is a language model and to aggregate text corpora. It is also preferable to learn a speaker's acoustic sample as an acoustic model and recognize a specific speaker. On the other hand, according to the present invention, it is only necessary to detect a learning word and an evaluation word, and therefore a high recognition rate necessary for general speech recognition technology is not necessarily required. It is possible to use only a general-purpose dictionary without requiring speaker recognition.

[Text input interface 104]
The text input interface 104 inputs texts described by teachers and students in individual learning or group lessons. The text is input to the word detection unit 12 via the morphological analysis unit 11. For example, in the case of distance classes, teachers often write and explain parts that students do not understand in the class. For this purpose, it is also preferable to obtain written text.

  As another embodiment, when writing with the electronic pen on the text input interface 104, not only writing characters but also data values of writing speed or writing pressure can be acquired. The writing speed and the writing pressure indicate the strength of the posture explained by the teacher. It can be assumed that students have a low level of understanding. A data value of writing speed or writing pressure can also be used as a correction value for the evaluation level.

(Word detection unit 12)
As described above, when the distance between words is set to “number of words”, there is a case where language relations are erroneously determined because there is no human utterance and the distance according to the order of the languages is close even if they are separated in time. is there. Therefore, in the case of text generated by speech recognition, the word detection unit 12 uses the “time width” between the learning word and the evaluation word as the distance between words. The shorter the time span, the closer the learning word and the evaluation word are. However, the time needs to be synchronized for both the speech voice text and the written text.

  Even when an audio signal is used, it is preferable to detect an evaluation word from “behind” the learning word. FIG. 5 shows an example of the relevance ratio for the speech-recognized text. Here, it can be understood that the relevance rate of the backward evaluation word is generally higher than that of the front evaluation word.

[Learning Word Conversion Table 15] (Evaluation Level Estimation Unit 13)
The learning word conversion table 15 associates “question numbers” with “learning words” for learning items.

  During class, utterances are often spoken as problem words, such as textbooks, problem collections, drills, and page numbers, rather than units, small units, and names of study items. At this time, a learning word cannot be detected as the name of the learning item. Although only the problem number can recognize the teacher and student parties from each other, the third party cannot recognize the learning item. Therefore, a learning word conversion table 15 is provided so that the evaluation level estimation unit 13 can recognize a learning word corresponding to the problem number.

  FIG. 6 is an operation explanatory diagram using the learning word conversion table of the present invention.

  According to FIG. 6, the learned word registration unit 101 registers “problem number” in the learned word conversion table 15 as a learned word. The evaluation level estimation unit 13 uses the learning word conversion table 15 to convert the question number into a learning word for the “problem number” as the detected learning word. Thereby, the evaluation level corresponding to the detected evaluation word can be associated with the learning item based on the learning word.

  According to FIG. 6, it is assumed that the learning word “1 of 3” is detected from the text and the evaluation word “attention” for the learning word is detected. At this time, according to the learned word conversion table 15, “1 of 3” is converted to “currently progressive”. Accordingly, the evaluation word “attention” is associated with the learning word “currently progressive”.

[Display 105]
The display 105 can display a list as a result output from the evaluation level estimation unit 13. Moreover, the learning image when it is instruct | indicated as an instruction pronoun from the word detection part 12 among the learning images image | photographed by the learning word recording part 16 is displayed.

[Camera 106]
The camera 106 shoots, for example, a textbook on a desk or a question sheet that a student is learning as a learning image. The camera 106 takes a learning image intermittently every short time. The captured learning image is output to the learning image recording unit 16 together with the shooting time.

[Learning image recording unit 16]
The learning image recording unit 16 sequentially records learning images taken by the camera 106. In addition, when instructed by the word detection unit 12 as an indicating pronoun, the learning image is output to the display 105.

[Image recognition unit 17]
The image recognition unit 17 extracts a learning word from the learning image recorded in the learning image recording unit 16 by image recognition. Specifically, the learning word is recognized from the learning image output from the learning image recording unit 16 when instructed by the word detection unit 12 as an indicating pronoun.

  FIG. 7 is an operation explanatory diagram using a camera mounted on the apparatus of the present invention.

  According to FIG. 7, the pronoun pronouns “this” and “this” are registered in the learning word registration unit 101 as learning words. The learning words uttered during the lesson may be spoken by the pronouns such as “this” and “this” instead of the unit, the small unit, and the name of the learning item. Even if these demonstrative pronouns can be extracted as learning words, those other than the teacher and the student's party do not know what the learning content was.

  Therefore, the word detection unit 12 instructs the learning image recording unit 16 when “this” or “this” is detected (or at the time). The learning image recording unit 16 outputs the learning image at that time (or the same shooting time) to the display 105 and also outputs it to the image recognition unit 17. The image recognition unit 17 recognizes, for example, a problem number from the learning image. According to FIG. 7, when the learning word “this” is detected, “1 of 3” is recognized from the learning image at that time. The learning word corresponding to “1 of 3” is converted by the learning word conversion table 15. As a result, the learning word “currently progressive” is obtained from the pronoun pronoun “this”.

In the case where the teacher indicates with a writing tool or a finger, an image recognition technique for extracting the shape of the pen tip or the fingertip may be used.
In addition, when a character cannot be recognized from a learning image photographed by the camera 106, the learning image may be stored linked to an evaluation word.

  FIG. 8 is an operation explanatory diagram using the sound transition table of the present invention.

When the uttered voice is converted into text, for example, the evaluation level may be different for the homonyms. For example, even if there is a language “good” by morphological analysis, the evaluation levels are different as follows.
Positive meaning of “Good” Negative meaning of “Is it good” Meaning that it is not necessary to say “I do not need it” Meaning to confirm the understanding by raising the word “Good?” Also, the degree of emphasis and nuance There are many cases where I do not know.

In this case, it is preferable that the acoustic information is analyzed from the collected uttered voice, and the change of inflection due to the pitch, the magnitude of the volume, or the intonation of the utterance language is acquired, and the precise meaning is grasped. According to FIG. 5, the evaluation level estimation unit 13 refers to the sound transition table 18 in which the pitch / volume transition of the audio signal is associated with the evaluation level correction value. According to FIG. 8, the following evaluation level correction values are set for the pitch / volume transition with respect to the elapsed time.
If the pitch is positive and the volume is “Good!”: Correction value 1
In case of questionable “Good!” Pitch / volume: Correction value 0.5
The evaluation level estimation unit 13 derives an evaluation level correction value of “good” from the pitch / volume transition of the audio signal using the sound transition table 18. Then, the evaluation level derived from the evaluation word is corrected. Specifically, the final evaluation level is determined by multiplying the “good” evaluation level by the correction value.

  FIG. 9 is a functional configuration diagram of the evaluation degree estimation system according to the present invention.

According to FIG. 9, the function which estimates a learning condition as a system is implement | achieved by allocating the function structure part of FIG. 5 to a terminal and an estimation server, respectively.
The terminal equipped with the microphone 103, the text input interface 104, the display 105, and the camera 106 includes a voice recognition unit 14 and a learning image recording unit 16.
In contrast, the estimation server includes a learning word registration unit 101, an evaluation word registration unit 102, a morpheme analysis unit 11, a word detection unit 12, an evaluation level estimation unit 13, a learning word conversion table 15, and image recognition. Unit 17 and a sound transition table 18.

  As described above in detail, according to the program, apparatus, and method of the present invention, it is possible to estimate an evaluation level for a learning item as a learning situation from a text based on a person's speech.

  The learning status can be grasped by a list in which learning words, evaluation words, and evaluation levels are arranged in time series. For example, even if the same learning language is used, the understanding level is low in the first half of the lesson, but the understanding level is high in the second half, and the situation where the learning content is acquired can be grasped. Also, if there is a large number of utterances of a learning word, or if the learning word is detected over a long period of time, it is estimated that the teacher is teaching over time, and the understanding of the learning content may not progress I can grasp. Such a list representing the learning situation can also be used for subsequent learning policy determination.

  Various changes, modifications, and omissions of the above-described various embodiments of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 Learning situation estimation apparatus 101 Learning word registration part 102 Evaluation word registration part 103 Microphone 104 Text input interface 105 Display 106 Camera 11 Morphological analysis part 12 Word detection part 13 Evaluation level estimation part 14 Speech recognition part 15 Learning word conversion table 16 Learning image Recording unit 17 Image recognition unit 18 Sound transition table

Claims (13)

A learning situation estimation program that makes a computer function to estimate an evaluation level for a learning item from text based on a person's remarks,
Learning word registration means for registering a plurality of learning words based on learning items;
An evaluation word registration means for registering a plurality of evaluation words that will be spoken with respect to the learning item and an evaluation level associated with each evaluation word;
Detecting the learning word from the text and detecting the evaluation word from within a predetermined distance between the front and rear of the learning word; and
A learning situation estimation program that causes a computer to function as evaluation level estimation means for associating an evaluation level corresponding to a detected evaluation word with a learning item based on the detected learning word. When the learning word is detected by the word detection means, the learning word is recorded in association with a time-series code number,
The learning level estimation program according to claim 1, wherein the evaluation level estimation unit causes a computer to function so as to output the evaluation level in time series for each learning word. The evaluation level estimation means causes the computer to derive so that the evaluation level decreases as the distance between the learning word detected by the word detection means and the evaluation word increases. The learning situation estimation program according to claim 1 or 2, characterized by the above-mentioned. The evaluation level estimation means uses, as an evaluation level, a value obtained by multiplying the reciprocal of the interword distance between the learning word and the evaluation word detected by the word detection means and the evaluation level of the evaluation word. The learning situation estimation program according to any one of claims 1 to 3, wherein the computer functions so as to be derived. Morphological analysis means for dividing the text into morphemes,
5. The learning state estimation according to claim 1, wherein the word detection unit further causes a computer to detect the learning word and the evaluation word from the classified morphemes. 6. program. The learning item further includes a learning word conversion table in which a problem number and a learning word are associated with each other,
The learning word registration means registers the question number as a learning word,
6. The evaluation level estimating means further causes the computer to function so as to convert the problem number into a learning word using the conversion table for the detected problem number. The learning state estimation program according to item 1. The learning status estimation program according to any one of claims 1 to 6, wherein the text is a text obtained by converting a voice signal uttered by a person by voice recognition. 8. The learning state estimation program according to claim 7, wherein the word detection unit further causes the computer to function so as to use a time width between the learning word and the evaluation word as the inter-word distance. Learning image recording means for recording a learning image taken based on the learning item;
Image recognition means for extracting a learning word from the learning image by image recognition;
The learning word registration means registers a pronoun as a learning word,
When the evaluation level estimation means detects an indicating pronoun, the learning image acquisition means acquires a learning image, and the image recognition means further causes the computer to recognize a learning word by image recognition from the learning image. The learning situation estimation program according to claim 9 or 10, characterized in that A sound transition table in which pitch / volume transition of the audio signal is associated with an evaluation level correction value;
The evaluation level estimation means derives an evaluation level correction value from the pitch / volume transition of the audio signal using the sound transition table, and causes the computer to function to correct the evaluation level derived from the evaluation word. The learning situation estimation program according to any one of claims 7 to 9, wherein The learning status estimation program according to any one of claims 1 to 6, wherein the computer further functions so that the text is a text in which a person describes his / her speech with a keyboard or an electronic pen. An apparatus for estimating an evaluation level for a learning item from a text based on a person's remarks,
Learning word registration means for registering a plurality of learning words based on learning items;
An evaluation word registration means for registering a plurality of evaluation words that will be spoken with respect to the learning item and an evaluation level associated with each evaluation word;
Detecting the learning word from the text and detecting the evaluation word from within a predetermined distance between the front and rear of the learning word; and
An apparatus comprising: an evaluation level estimating unit that associates an evaluation level corresponding to the detected evaluation word with a learning item based on the detected learning word. A method for estimating an evaluation level for a learning item from a text based on a person's remarks using a device,
The device is
A learning word registration unit that registers a plurality of learning words based on learning items;
An evaluation word registration unit that registers a plurality of evaluation words that will be remarked with respect to a learning item and an evaluation level associated with each evaluation word;
A first step of detecting the learning word from the text and detecting the evaluation word within a predetermined distance between the front and rear of the learning word;
A second step of associating a learning item based on the detected learning word with an evaluation level corresponding to the detected evaluation word.

Images