WO2022124355A1 - Système d'analyse de résultats de correction de composition écrite en anglais - Google Patents

Système d'analyse de résultats de correction de composition écrite en anglais Download PDF

Info

Publication number
WO2022124355A1
WO2022124355A1 PCT/JP2021/045231 JP2021045231W WO2022124355A1 WO 2022124355 A1 WO2022124355 A1 WO 2022124355A1 JP 2021045231 W JP2021045231 W JP 2021045231W WO 2022124355 A1 WO2022124355 A1 WO 2022124355A1
Authority
WO
WIPO (PCT)
Prior art keywords
unit
english composition
storage device
vocabulary level
sentence
Prior art date
Application number
PCT/JP2021/045231
Other languages
English (en)
Japanese (ja)
Inventor
諭 内田
文勝 本間
圭悟 佐藤
千佳 廣田
真之介 田阪
浩一 西岡
裕司 前田
Original Assignee
株式会社新興出版社啓林館
国立大学法人九州大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 株式会社新興出版社啓林館, 国立大学法人九州大学 filed Critical 株式会社新興出版社啓林館
Publication of WO2022124355A1 publication Critical patent/WO2022124355A1/fr

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F40/00—Handling natural language data
    • G06F40/10—Text processing
    • G06F40/194—Calculation of difference between files
    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B19/00—Teaching not covered by other main groups of this subclass
    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B19/00—Teaching not covered by other main groups of this subclass
    • G09B19/06—Foreign languages
    • G—PHYSICS
    • G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
    • G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
    • G09B7/00—Electrically-operated teaching apparatus or devices working with questions and answers
    • G09B7/02—Electrically-operated teaching apparatus or devices working with questions and answers of the type wherein the student is expected to construct an answer to the question which is presented or wherein the machine gives an answer to the question presented by a student

Definitions

  • the present invention relates to a technique for analyzing a correction result by comparing an English composition with a correction sentence thereof.
  • a proofreading system that calibrates grammatical errors and spelling mistakes in English composition with a computer has already been put into practical use due to the development of artificial intelligence in recent years, and is being used in the business field as well.
  • the English learning system by the applicant of the present application which is exemplified in Non-Patent Document 1, is an example of a correction system that provides such a correction service.
  • users such as learners and teachers send the English composition of the Japanese sentence being asked to the system and receive the correction sentence by the corrector with excellent English ability from the system. be able to.
  • Patent Document 1 considers the difference in the position of the same word group, the difference in the conjugation form of the word, the difference in the part of speech of the word, etc. between the answer sentence created by the user and the correct answer sentence prepared in advance. Then, the technology for scoring the answer text is disclosed.
  • the present invention has been made in view of the above problems, and provides an English composition correction result analysis system that analyzes a correction result by comparing an English composition with the correction sentence and returns the analysis result to the user. With the goal.
  • the first aspect of the present invention is an English composition correction result analysis system, which includes an input receiving unit, a matching unit, and a result output unit.
  • the input reception unit accepts input of English composition and its correction sentence.
  • the associating unit associates the sentences constituting the English composition and the correction sentence with each other when the input is accepted.
  • the result output unit outputs the result of the mapping performed by the mapping unit.
  • the matching unit associates the sentences with each other so that the distance between the sentences based on the predetermined standard is minimized.
  • a user for example, a learner, a teacher, etc.
  • the second aspect of the present invention is the English composition correction result analysis system according to the first aspect, and the matching unit includes a single sentence and a plurality of sentences included in the English composition and the correction sentence.
  • the sentences are described in ascending order of distance between the single sentence and the plurality of sentences based on the predetermined criteria, until there are no unrelated sentences in either the English composition or the correction sentence. Make a correspondence between each other.
  • the correspondence between sentences can be obtained by a steady method in order, and the user who receives the result of the correspondence can find the correspondence between the sentences including a plurality of sentences and the distance is the minimum. , You can know.
  • the third aspect of the present invention is an English composition correction result analysis system according to the first or second aspect, which further includes a part-of-speech analyzer access unit and a part-of-speech search unit.
  • the part-speech analyzer access unit receives input of a sentence, analyzes the part-speech for each word constituting the input sentence, and accesses the part-speech analyzer that outputs the analysis result.
  • the part-speech search unit selects a set of words and their part-speech between the associated sentences that is in one sentence but not in the other sentence, or in both sentences. The search is performed based on the analysis result obtained by the analyzer access unit accessing the part of speech analyzer. Then, the result output unit also outputs the result of the search performed by the part of speech search unit.
  • part of speech includes a large classification of part of speech such as verbs and nouns, and further, a minor classification of part of speech such as verb tense, noun tense, and noun countability / uncountability (subordinate attributes of major classification). May be included.
  • the part-of-speech analyzer may be an external device.
  • the fourth aspect of the present invention is an English composition correction result analysis system according to any one of the first to third aspects, further comprising a grammar item storage device access unit and a grammar item search unit.
  • the grammar item storage device access unit accesses a grammar item storage device that stores one or more elements including a word and its part of speech in association with each other and the corresponding grammar item.
  • the grammar item search unit stores the grammar item in one sentence but not in the other sentence, or the grammar item in both sentences, among the associated sentences.
  • the device access unit searches based on the stored contents obtained by accessing the grammar item storage device. Then, the result output unit also outputs the result of the search performed by the grammar item search unit.
  • the user can know the grammatical items that are inconsistent or the grammatical items that match among the grammatical items that are in a corresponding relationship with each other.
  • "part of speech” includes a large classification of part of speech such as verbs and nouns, and further, a minor classification of part of speech such as verb tense, noun tense, and noun countability / uncountability (subordinate attributes of major classification). May be included.
  • the grammatical items are, for example, tense, progressive tense, passive voice, subjunctive mood, and the like.
  • the grammar item storage device may be an external storage device.
  • the fifth aspect of the present invention is the English composition correction result analysis system according to the fourth aspect, and the grammar item search unit is not only the grammar item but also the one or more corresponding to the grammar item.
  • the element of is also recognized as a grammatical item in both sentences only when the associated sentences match each other.
  • the sixth aspect of the present invention is an English composition correction result analysis system according to the fourth or fifth aspect, further including an explanatory data storage device and an explanatory data reading unit.
  • the commentary data storage device stores commentary data containing explanations for each grammar item in association with the corresponding grammar item.
  • the explanatory data reading unit reads the explanatory data corresponding to the grammar item in one sentence but not in the other sentence, which is searched by the grammar item search unit, from the explanatory data storage device. Then, the result output unit also outputs the explanatory data read by the explanatory data reading unit.
  • the user can obtain a corresponding explanation when there is a discrepancy in the grammatical items between the English composition and the corrected sentence.
  • the seventh aspect of the present invention is an English composition correction result analysis system according to any one of the first to sixth aspects, which is a vocabulary level storage device access unit, a vocabulary level determination unit, and a vocabulary level representative. It further includes a value calculation unit, a distance calculation unit, an evaluation data storage device, and an evaluation data reading unit.
  • the vocabulary level storage device access unit accesses a vocabulary level storage device that stores a vocabulary level for each word in association with the corresponding word.
  • the vocabulary level determination unit determines the vocabulary level for each word for all words or words satisfying predetermined conditions included in the English composition for which input is accepted, and the vocabulary level storage device access unit determines the vocabulary level for each word. This is done based on the storage content obtained by accessing the storage device.
  • the vocabulary level representative value calculation unit calculates a vocabulary level representative value which is a representative value of the vocabulary level determined by the vocabulary level determination unit.
  • the distance calculation unit calculates the total distance, which is the distance based on the predetermined standard, between the entire English composition for which input has been accepted and the entire correction sentence for which input has been accepted.
  • the evaluation data storage device stores evaluation data including evaluations according to the total distance and the vocabulary level representative value in association with the area for each area of the total distance and the vocabulary level representative value. ..
  • the evaluation data reading unit reads out the evaluation data corresponding to the calculated vocabulary level representative value and the calculated total distance from the evaluation data storage device. Then, the result output unit also outputs the evaluation data read by the evaluation data reading unit.
  • the user can obtain an evaluation based on the degree of difference between the English composition and the corrected sentence and the vocabulary level of the English composition.
  • the vocabulary level storage device may be an external storage device.
  • the representative value is, for example, an average value.
  • the predetermined condition is that, for example, the word is a content word.
  • the eighth aspect of the present invention is an English composition correction result analysis system, which is a vocabulary level storage device access unit, an input reception unit, a vocabulary level determination unit, a vocabulary level representative value calculation unit, and a distance. It includes a calculation unit, an evaluation data storage device, an evaluation data reading unit, and a result output unit.
  • the vocabulary level storage device access unit accesses a vocabulary level storage device that stores a vocabulary level for each word in association with the corresponding word.
  • the input reception unit accepts input of an English composition including a plurality of sentences and its correction sentence.
  • the vocabulary level determination unit determines the vocabulary level for each word for all words or words satisfying predetermined conditions included in the English composition for which input is accepted, and the vocabulary level storage device access unit determines the vocabulary level for each word.
  • the vocabulary level representative value calculation unit calculates a vocabulary level representative value which is a representative value of the vocabulary level determined by the vocabulary level determination unit.
  • the distance calculation unit calculates the total distance, which is the distance based on a predetermined standard, between the entire English composition for which input has been accepted and the entire correction sentence for which input has been accepted.
  • the evaluation data storage device stores evaluation data including evaluations according to the total distance and the vocabulary level representative value in association with the area for each area of the total distance and the vocabulary level representative value. ..
  • the evaluation data reading unit reads out the evaluation data corresponding to the calculated vocabulary level representative value and the calculated total distance from the evaluation data storage device.
  • the result output unit outputs the evaluation data read by the evaluation data reading unit.
  • the user can obtain an evaluation based on the degree of difference between the English composition and the corrected sentence and the vocabulary level of the English composition.
  • the vocabulary level storage device may be an external storage device.
  • the representative value is, for example, an average value.
  • the predetermined condition is that, for example, the word is a content word.
  • an English composition correction result analysis system that analyzes the correction result by comparing the English composition and the correction sentence and returns the analysis result to the user is realized.
  • FIG. 3 is a screen diagram showing an example in which the correspondence between sentences between an English composition and its correction sentence is displayed on the screen of a user terminal using the system of FIG. 1.
  • FIG. 1 is a block diagram illustrating the configuration of an English composition correction result analysis system according to an embodiment of the present invention.
  • This English composition correction result analysis system (abbreviated as "analysis system” as appropriate) 101 is a learner's English composition (also referred to as “learner's English composition” as appropriate) and its correction by the corrector ("corrector English”). It is a system that analyzes the correction result by comparing it with "composition” and returns the analysis result to users such as learners.
  • the analysis system 101 has a computer and is used by a user as a server connected to the network 120 in the illustrated example.
  • the network 120 is, for example, the Internet.
  • the learning system 141 In addition to the analysis system 101, the learning system 141, the corrector terminal 143, the user terminal 151, and the system administrator terminal 153 are connected to the network 120.
  • the user terminal 151 is a computer of a user who uses the learning system 141 and the analysis system 101. In addition to the learner, the learner's parents, teachers, etc. can also be users.
  • the system administrator terminal 153 is an administrator's computer that manages the learning system 141 and the analysis system 101.
  • the corrector terminal 143 is a corrector's computer.
  • the learning system 141 sets a question selected by the learner through the user terminal 151 from the questions of the English composition prepared in advance, receives the English composition answered by the learner, and uses the received English composition as the corrector terminal 143.
  • the learning system 141 further sends the received English composition and the corrected sentence to the user terminal 151 and also sends the received English composition to the analysis system 101 for analysis processing.
  • the content of the question may be a translation of a given Japanese sentence into English, or may be a request for creation of the content, such as "Write within 100 words about your future dreams.”
  • the corrector does not necessarily have to be fluent in Japanese, and may correct based only on the learner's English composition.
  • the user terminal 151 accesses only the learning system 141 and uses the analysis system 101 through the learning system 141.
  • the user terminal 151 may use the learning system 141 by installing a dedicated application, or may learn by accessing the program provided by the learning system 141 in the form of an ASP (application service provider).
  • System 141 may be used.
  • the analysis system 101 and the learning system 141 are individually connected to the network 120 as independent servers, but both may be constructed in a single server. In this case, the exchange of data between the analysis system 101 and the learning system 141 is performed without going through the network 120.
  • the device groups 161, 163, 165 used by the analysis system 101 are connected to the network 120 as a server.
  • the device group 161, 163, 165 includes a part-speech analysis device 161, a grammar item storage device 163, and a vocabulary level storage device 165.
  • the analysis system 101 having a computer includes devices 1 to 17 that are connected to each other so as to exchange data and each of which fulfills an individual function.
  • the device groups 1 to 17 include a device composed of only hardware and a device equivalently realized by reading a program by a computer.
  • the device groups 1 to 17 include an input receiving unit 1, a mapping unit 2, a result output unit 3, a part-of-speech analysis device access unit 4, a part-of-speech search unit 5, a grammar item storage device access unit 6, and a grammar item search.
  • the analysis system 101 associates the sentences included in the two English compositions with each other in order to enable an appropriate contrast between the learner's English composition and the corrector's English composition.
  • the mapping is performed, for example, by using the Levenshtein distance and selecting the other sentence with the shortest distance from one sentence as the corresponding sentence.
  • the Levenshtein distance is a well-known method for assessing the similarity between sentences. The higher the similarity between sentences, the smaller the distance. If the sentences are the same, the distance will be zero.
  • the distance is calculated for various combinations of sentences. The distance is calculated in singular and multiple sentence units. Even if one single sentence corresponds to multiple other sentences, the distance between one single sentence and the other is also calculated so that the mapping is correct. In this way, sentence-based correspondence is made between the learner's English composition and the corrector's English composition.
  • the input reception unit 1 of the analysis system 101 receives the input of the learner's English composition and the corrector's English composition, which is the correction sentence, sent from the learning system 141.
  • the associating unit 2 associates the learner's English composition and the corrector's English composition with each other.
  • the Levenshtein distance is calculated for various combinations of the singular and plural sentences contained in the learner's English composition and the singular and plural sentences included in the corrector's English composition.
  • the sentences are associated with each other in ascending order of the Levenshtein distance. The correspondence is performed until there are no unmatched sentences in either the learner's English composition or the corrector's English composition.
  • the result output unit 3 outputs the result of the association made by the association unit 2 to the learning system 141.
  • the learning system 141 sends the result of the association to the user terminal 151.
  • the user terminal 151 displays the result of the received correspondence on the screen.
  • the user for example, a learner, a teacher, etc.
  • the user can know the correspondence between the learner's English composition and the corrector's English composition.
  • the English composition 1 exemplified in Table 1 is an English composition tentatively assumed as a learner's English composition. Further, the English composition 2 exemplified in Table 2 is an English composition tentatively assumed as a corrector English composition.
  • the sentences included in each English composition are numbered 1, 2, ..., And a, b, c, ... In the order in which they are arranged.
  • a sentence that combines multiple sentences is used as a collection of sentences. to add.
  • An upper limit may be set in consideration of the calculation load as to how many sentences should be considered as the number of sentences constituting the combined sentence. In the examples of Tables 3 and 4, the upper limit is set to "3". That is, the combination of up to three sentences is considered. Further, regarding the combination of a plurality of sentences, in the illustrated example, only consecutive sentences are targeted for combination. In reality, the combination of discontinuous sentences rarely shows a good correspondence between the learner's English composition and the corrector's English composition. Therefore, it can be said that the method of combining only consecutive sentences causes almost no problem in reality, can avoid wasting calculation time, and is a practical method.
  • Table 14 illustrates the final result of the mapping.
  • Table 15 exemplifies the final results including sentence 1 constituting English sentence 1 (learner English composition) and sentence 2 constituting English sentence 2 (corrector English composition) by rearranging the sentences as appropriate. There is. In the example of Table 15, the sentences are rearranged according to the order of English sentences 2. The final results illustrated in Table 15 are output from the result output unit 3 to the learning system 141.
  • FIG. 2 is a screen diagram showing an example in which a learner's English composition and its corrector's English composition are displayed on the screen of the user terminal 151.
  • FIG. 3 is a screen diagram showing an example in which the correspondence between the sentences between the learner's English composition and the corrector's English composition is displayed on the screen of the user terminal 151.
  • the learner's English composition and the corrector's English composition illustrated in FIG. 2 are not the same as the learner's English composition and the corrector's English composition illustrated in Tables 1 and 2.
  • the learning system 141 sends the learner English composition received from the user terminal 151 and the corrector English composition received from the corrector terminal 143 to the user terminal 151.
  • the screen diagram of FIG. 2 shows an example in which the learner's English composition and the corrector's English composition sent from the learning system 141 are displayed side by side on the same screen so that they can be easily compared with each other.
  • the learner's English composition is displayed on the left side, and the corrector's English composition is displayed on the right side.
  • the "English composition analysis” button is displayed at the lower end of the screen.
  • the learning system 141 sends the learner English composition and the corrector English composition to the analysis system 101 for analysis processing.
  • the analysis system 101 starts the analysis of the sent learner's English composition and the corrector's English composition.
  • the analysis system 101 outputs the final result of the mapping illustrated in Table 15 to the learning system 141, the learning system 141 sends the received final result to the user terminal 151.
  • the sentences having a corresponding relationship between the learner's English composition and the corrector's English composition are displayed side by side with each other.
  • the user can clearly grasp the correspondence between the sentences between the learner's English composition and the corrector's English composition.
  • the "number 1", “number 2", and "distance” exemplified in Table 15 are not displayed. In this case, these data may not be included in the final result of the association sent by the analysis system 101 to the learning system 141.
  • the correspondence between sentences is performed until there are no unmatched sentences in either the learner's English composition or the corrector's English composition.
  • the unassociated sentences disappear at the same time.
  • the learner's English composition inserts a sentence that is completely unrelated to the question, the sentence that cannot be associated disappears first in the corrector's English composition, and the sentence that cannot be associated is the learner's English. It may remain in the composition.
  • a sentence that cannot be associated is also displayed in the learner English composition column on the left side, and the corrector English composition column on the right side is "irrelevant to the subject". ", Or a blank is displayed. Strikethrough may be superimposed on the unassociated sentence displayed in the learner English composition column on the left side.
  • Such processing is performed by, for example, the mapping unit 2. Sentences that cannot be associated are not subject to the following part-of-speech analysis and grammatical analysis.
  • the analysis system 101 further performs part-of-speech analysis by using the result of associating sentences with each other.
  • the difference between the part-speech used in one sentence and the part-speech used in the other sentence is logically calculated for the sentences corresponding to each other between the learner's English composition and the corrector's English composition. Then, based on the presence or absence of the difference, the analysis result is returned to the user as whether there is an error in the part of speech.
  • Part of speech analysis includes not only the major classification of part of speech such as verbs and nouns, but also the lower attributes of the major classification, that is, the minor classification of part of speech, for example, the verb time system for verbs and the singular / plural nouns for nouns.
  • Nouns such as countable and uncountable, can also be analyzed. That is, a broadly defined part of speech including a subclass of part of speech may be the subject of analysis. In the following explanation, we will take an example of analyzing such broadly defined part of speech.
  • the part of speech analysis device 161 is a device that, when a sentence is input, analyzes the part of speech for each word constituting the sentence and outputs the analysis result.
  • a part-of-speech analyzer 161 that fulfills such a function is already provided on the Internet as a server. TreeTagger (https://www.cis.uni-muenchen.de/ ⁇ schmid/tools/TreeTagger/) is one example.
  • the part-speech analyzer access unit 4 inputs the sentence sets associated with each other by the associating unit 2 into the part-speech analysis device 161 to obtain the result of analyzing the part of speech word by word for these sentence sets.
  • Which set of sentences to enter can be specified at the same time, for example, when the user requests part-of-speech analysis. For example, when the user clicks on any of the sentence sets displayed in FIG. 3, a pull tab or a small screen appears, and menus such as "part of speech analysis” and "grammar analysis” are displayed. By selecting "part of speech analysis", the user can request the part of speech analysis and specify the set of sentences to be analyzed at the same time.
  • the part-speech search unit 5 is located in one of the pairs of words and their part-speech between the associated sentences based on the analysis result of the part-speech acquired by the part-speech analyzer access unit 4. Search for pairs that are not in the other sentence, or in both sentences.
  • the result output unit 3 outputs the result searched by the part of speech search unit 5 to the learning system 141.
  • the learning system 141 sends the sent search result to the user terminal 151.
  • the user terminal 151 displays the received search result on the screen. As a result, the user who receives the search result can know the disagreement group or the matching group among the pairs of the word and its part of speech among the sentences in the corresponding relationship.
  • the part-of-speech analyzer access unit 4 inputs "text01" and "text02" into the part-speech analyzer 161 to obtain the analysis results exemplified in Tables 17 and 18, respectively.
  • the part of speech analyzer 161 to be used is the above-mentioned TreeTagger as an example.
  • the part-speech symbol is a symbol for identifying a part-speech defined in the part-speech analyzer 161.
  • part of speech means the above-mentioned part of speech in a broad sense.
  • the "meaning of the part-of-speech symbol” expressed in Japanese is a phrase that expresses the meaning of the part-of-speech symbol "VBD" in Japanese.
  • the part-of-speech analyzer access unit 4 separately accesses another server that is associated with the part-of-speech symbol and stores its meaning, or FIG.
  • the part-of-speech symbol meaning data storage device 16 provided as a storage device in the analysis system 101 stores the part-of-speech symbol and its meaning in association with each other, and the part-of-speech analyzer access unit 4 stores this part-of-speech.
  • the data storage device 16 may be accessed separately.
  • the part-of-speech symbol and its meaning data can be stored in the part-of-speech symbol meaning data storage device 16 through the input receiving unit 1 by, for example, the system administrator operating the system administrator terminal 153.
  • the part of speech search unit 5 is found in both the part of speech in "text01", the part of speech in "text02", and both "text01” and “text02” from the analysis results exemplified in Tables 17 and 18.
  • Search for part of speech In the search, "prototype of word” and “meaning of part of speech symbol” are not referred to. However, it refers not only to the “part of speech symbol” corresponding to the part of speech in a broad sense, but also to the "word”. Only when both the "part of speech symbol” and the "word” match, it is judged as "match".
  • FIG. 4 shows an example in which the analysis results of the words constituting each of the sentences corresponding to each other between the learner's English composition and the corrector's English composition and their part of speech are displayed on the screen of the user terminal 151. It is a screen view which shows. In the illustrated example, both sentences are displayed at the top of the screen. At the top of the screen, the meaning of each word and its part-speech symbol is displayed along with the sentence. In each case, the sentences in the learner's English composition are displayed on the left side and the sentences in the corrector's English composition are displayed on the right side for easy comparison. In the middle of the screen, the meanings of words and their part-speech symbols that exist only on one side of the corresponding sentences are displayed.
  • the left side shows the meanings of words and their part of speech symbols only in the sentence in the learner's English composition
  • the right side shows the words and their part of speech only in the sentence in the corrector's English composition.
  • the symbol is displayed.
  • the meanings of the words and their part-speech symbols in both of the corresponding sentences are displayed. From the illustrated screen, the user can clearly understand how the learner's English composition was corrected, where the problem was, what was good, and so on.
  • the analysis system 101 further performs grammatical analysis using the result of associating sentences with each other.
  • the grammar analysis the difference between the grammar item used in one sentence and the grammar item used in the other sentence is calculated for the sentences corresponding to each other between the learner's English composition and the corrector's English composition.
  • a logical operation is performed, and the analysis result is returned to the user as having / without an error in the grammar based on the presence / absence of a difference.
  • the grammatical items to be compared are specified by selecting an appropriate one from a large number of predetermined grammatical items based on the combination of part of speech and the like.
  • Grammar items are, for example, passive voice, verb + to infinitive, present perfect, and so on.
  • the procedure of grammar analysis will be described with reference to FIG.
  • the user terminal 151 requests a grammar analysis by a user's click operation or the like, this request is input to the input receiving unit 1 of the analysis system 101 via the learning system 141 and transmitted to the grammar item storage device access unit 6.
  • the user can specify the set of sentences to be analyzed at the same time. For example, when the user clicks on any of the sentence sets displayed in FIG. 3, a pull tab or a small screen appears, and menus such as "part of speech analysis” and "grammar analysis” are displayed. By selecting "grammar analysis", the user can request grammar analysis and specify a set of sentences to be analyzed at the same time.
  • the grammar item storage device access unit 6 accesses the grammar item storage device 163.
  • the grammar item storage device 163 stores one or more combinations of elements for each word including a word and its part of speech in association with the grammar item represented by the combination.
  • a collection of regular expressions for grammar items has been published as part of the CEFR-J project. (Link: http://cefr-j.org/PDF/sympo2020/CEFRJGP_GRAMMATICAL_ITEM_LIST.pdf).
  • 501 kinds of grammatical items are recorded in this regular expression collection in association with a regular expression of one or more combinations of elements for each corresponding word.
  • the elements for each word include "words", “part of speech symbols”, and “prototypes of words” exemplified in Tables 17 and 18.
  • Regular expression means a well-known expression form that concisely expresses a number of sets of elements using metacharacters.
  • the regular expressions included in the above regular expression collection are exemplified in Table 23 below.
  • the grammar item storage device access unit 6 accesses the grammar item storage device 163 to select a sentence associated with the grammar item storage device 163 from among the combinations of elements stored in the grammar item storage device 163 for each word. Search for the one that matches the combination of elements for each word included in the set, and pick up the corresponding grammar item.
  • the stored contents of the grammar item storage device 163 can also be stored in the storage device of the analysis system 101.
  • the grammar item storage device access unit 6 may access the storage device instead of the grammar item storage device 163.
  • the grammar item search unit 7 is based on the grammar item acquired by the grammar item storage device access unit 6, and the grammar item that is in one sentence but not in the other sentence, or both, between the associated sentences. Search for grammar items in the sentence.
  • the result output unit 3 outputs the result searched by the grammar item search unit 7 to the learning system 141.
  • the learning system 141 sends the sent search result to the user terminal 151.
  • the user terminal 151 displays the received search result on the screen. As a result, the user who receives the search result can know the grammatical items that do not match or the grammatical items that match between the sentences in the corresponding relationship.
  • the grammar item storage device access unit 6 may acquire the analysis result already obtained by the part of speech analysis device access unit 4 from the part of speech analysis device access unit 4 without newly accessing the part of speech analysis device 161. Next, the grammar item storage device access unit 6 expands the analysis results exemplified in Tables 17 and 18 into the original sentence formats of "text01” and "text02” as illustrated in Table 22. That is, the sentences “text01” and “text02” are expanded into the form of "word_part of speech symbol_word prototype" for each word.
  • Table 23 exemplifies two examples of the regular expressions recorded in the above-mentioned regular expression collection.
  • Each regular expression is recorded in association with a grammatical item, and each is described by one or more sets of elements whose contents are "word_part of speech symbol_word prototype". Therefore, the grammar item storage device 163 for storing a regular expression collection represents one or more sets of elements in which each element contains a "word”, a "(broadly defined) part of speech", and a "prototype of a word”. It is equivalent to remembering grammatical items in association with each other.
  • the grammar item storage device access unit 6 accesses the grammar item storage device 163, and for each of the 501 types of regular expressions, the corresponding characters in the expanded sentences of "text01" and "text02" exemplified in Table 22. Explore columns. The results of the search are illustrated in Tables 24 and 25.
  • the data describing the parent-child relationship is stored in the grammar item storage device 163 or analyzed as illustrated in FIG.
  • the grammar item storage device access unit 6 grasps such a parent-child relationship as illustrated in Tables 24 and 25. be able to. In the latter case, the grammar item storage device access unit 6 accesses the parent-child relationship data storage device 17 in addition to the grammar item storage device 163.
  • the parent-child relationship data storage device 17 can store data through the input reception unit 1 by the system administrator operating the system administrator terminal 153.
  • the grammar item search unit 7 sets the regular expression only in "text01", the regular expression only in "text02", and "text01" and "text02". Search for regular expressions on both sides.
  • the search not only the regular expression identification number but also the corresponding character string (“matched phrase” in the table) is referred to. That is, it is judged as "match” only when both the regular expression identification number and the corresponding character string match.
  • it is possible to avoid determining "match” when it should not be determined as “match”, such as when there is a different character string in which only the regular expression identification number matches.
  • Tables 26 to 28 exemplify the results of the search.
  • the regular expression “115" has a parental relationship with the regular expressions "107” and "219”. Therefore, the grammar item search unit 7 may delete the regular expressions "107” and "219” that are related to each other from the regular expressions that exist only in "text02".
  • the search results exemplified in Tables 26 to 28 are sent to the user terminal 151 through the learning system 141.
  • FIG. 5 is a screen diagram showing an example in which the analysis results of the grammatical items of the sentences corresponding to each other between the learner's English composition and the corrector's English composition are displayed on the screen of the user terminal 151. ..
  • both sentences are displayed at the top of the screen.
  • the sentences in the learner's English composition are displayed on the left side, and the sentences in the corrector's English composition are displayed on the right side.
  • the grammar items of each sentence are displayed in the upper part of the screen.
  • the grammar item is displayed together with the corresponding character string (“matched words / phrases” in Tables 24, 25, etc.). In the middle of the screen, grammar items that exist only in one of the corresponding sentences are displayed.
  • the grammar items found only in the sentences in the learner's English composition are displayed on the left side, and the grammar items found only in the sentences in the corrector's English composition are displayed on the right side.
  • the grammar items on both sides of the corresponding sentence are displayed. From the illustrated screen, the user can clearly understand where the grammatical problem was in the learner's English composition, what was good, and so on.
  • this function can be realized by preparing an explanatory data storage device 8 that stores grammatical items and explanatory data such as explanatory videos thereof in association with each other.
  • the explanatory data can be stored in the explanatory data storage device 8 through the input receiving unit 1 by the system administrator operating the system administrator terminal 153.
  • this request is input to the input receiving unit 1 of the analysis system 101 via the learning system 141 and transmitted to the explanation data reading unit 9.
  • the request includes, for example, data specifying sentences such as "text01" and "text02" exemplified in Tables 26 and 27, and data specifying grammatical items in sentences such as numbers [0] and [1]. Is done.
  • the explanation data reading unit 9 acquires the regular expression identification number of the grammar item for which explanation is required from the data included in the request from the data obtained by searching the grammar item search unit 7 exemplified in Tables 26 and 27. , The explanatory data corresponding to the acquired regular expression identification number is acquired from the explanatory data storage device 8.
  • the acquired commentary data is sent from the commentary data reading unit 9 to the user terminal 151 via the result output unit 3 and the learning system 141.
  • the request transmitted from the user terminal 151 to the input receiving unit 1 may include a regular expression identification number that specifies a grammatical item.
  • the commentary data reading unit 9 acquires the commentary data corresponding to the regular expression identification number from the commentary data storage device 8 without referring to the data searched and obtained by the grammar item search unit 7. Can be done.
  • the analysis system 101 can also return to the user an evaluation of the learner's English composition based on both the distance between the learner's English composition and its corrector's English composition and the vocabulary level of the learner's English composition.
  • this function includes a vocabulary level storage device 165, a vocabulary level storage device access unit 11, a vocabulary level determination unit 12, a vocabulary level representative value calculation unit 13, an evaluation data storage device 14, and evaluation data. This can be achieved by providing the reading unit 15.
  • the vocabulary level storage device 165 stores a large number of words in association with each other and the vocabulary level.
  • the vocabulary level for example, the vocabulary level defined in CEFR-J can be used.
  • the CEFR-J project provides vocabulary level data in spreadsheet format that describes the vocabulary levels defined by the CEFR-J for a large number of words (http://www.cefr-j.org/download.html). #cefrj_wordlist).
  • the vocabulary level storage device 165 stores vocabulary level data provided by, for example, CEFR-J.
  • the vocabulary level storage device 165 is already connected to the network 120 as a server, and if available, it can be used. If there is no available vocabulary level storage device 165, a new vocabulary level storage device 165 may be provided and the system administrator may operate the system administrator terminal 153 to store the vocabulary level data.
  • the storage device included in the analysis system 101 may store the vocabulary level data. In this case, the vocabulary level storage device access unit 11 accesses the storage device included in the analysis system 101.
  • the learning system 141 When the user terminal 151 sends an evaluation request to the learning system 141 by the user performing a predetermined operation, the learning system 141 inputs the sent request to the input reception unit 1 of the analysis system 101.
  • the input reception unit 1 conveys the received request to the vocabulary level determination unit 12 and the distance calculation unit 10.
  • a user's predetermined operation is performed by, for example, providing a button for requesting an evaluation on any of the screens illustrated in FIGS. 3 to 5 or all of them, and clicking the button. ..
  • the vocabulary level determination unit 12 Upon receiving the request, the vocabulary level determination unit 12 causes the vocabulary level storage device access unit 11 to access the vocabulary level storage device 165 to acquire the vocabulary level for each word included in the learner's English composition. Since the learner's English composition has already been transmitted from the input reception unit 1 to the correspondence unit 2, etc., the vocabulary level determination unit 12 can acquire the learner's English composition from, for example, the association unit 2.
  • the vocabulary level determination unit 12 may acquire the vocabulary level only for the words satisfying a predetermined condition, instead of acquiring the vocabulary level for all the words included in the learner's English composition. For example, the function words may be omitted from the words, and the vocabulary level may be obtained only for the content words.
  • the data provided by the CEFR-J project also describes the part of speech for each word, and it is possible to distinguish between function words and content words by referring to the part of speech.
  • the vocabulary level for each word acquired by the vocabulary level determination unit 12 is transmitted to the vocabulary level representative value calculation unit 13, so that the representative values are calculated.
  • the representative value is, for example, an average value, a median value, or the like.
  • the distance calculation unit 10 calculates the distance between the learner's English composition and the corrector's English composition.
  • the distance calculated here is not the distance between the learner's English composition and each sentence included in the corrector's English composition, but the distance between the entire learner's English composition and the entire corrector's English composition.
  • the distance is, for example, the Levenshtein distance. Since the distance calculation unit 10 has already received the learner's English composition and the corrector's English composition from the input reception unit 1, it is possible to calculate the distance between them.
  • the evaluation data reading unit 15 reads out the evaluation data corresponding to the calculated vocabulary level representative value and the calculated distance from the evaluation data storage device 14.
  • the evaluation data storage device 14 stores evaluation data according to both the vocabulary level representative value of the learner's English composition and the distance between the learner's English composition and the corrector's English composition.
  • FIG. 6 shows an example of evaluation data.
  • the evaluation space composed of the vocabulary level representative value and the distance is divided into four areas, and individual evaluation data is associated with each area. It is also possible to divide the evaluation space into a larger number of areas. Also, the area is not necessarily limited to a rectangular area.
  • the evaluation data can be stored in the evaluation data storage device 14 through the input receiving unit 1 by the system administrator operating the system administrator terminal 153.
  • the evaluation data reading unit 15 transmits the read evaluation data to the result output unit 3.
  • the result output unit 3 transmits the received evaluation data to the learning system 141, and the learning system 141 sends the received evaluation data to the user terminal 151.
  • the storage content of the vocabulary level storage device 165 may be stored in a storage device (not shown) included in the analysis system 101. Further, when the user requests an evaluation based on the learner's English composition and the corrector's English composition without requesting the correspondence between the sentences, the learning system 141 sets the learner's English composition and its correction.
  • the person's English composition is sent to the input reception unit 1 of the analysis system 101, and the input reception unit 1 transmits these data to the vocabulary level determination unit 12 and the distance calculation unit 10, so that the evaluation data reading unit 15 Can acquire evaluation data and convey it to the result output unit 3.

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Educational Administration (AREA)
  • Educational Technology (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Audiology, Speech & Language Pathology (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • General Engineering & Computer Science (AREA)
  • Document Processing Apparatus (AREA)
  • Electrically Operated Instructional Devices (AREA)

Abstract

La présente invention concerne un système d'analyse de résultats de correction de composition écrite en anglais qui est pourvu d'une unité d'acceptation d'entrée, d'une unité d'association et d'une unité d'affichage de résultats. L'unité d'acceptation d'entrée accepte une entrée consistant en une composition écrite en anglais et en un texte corrigé correspondant. L'unité d'association associe les uns aux autres les textes constituant la composition écrite en anglais et le texte corrigé dont l'entrée a été acceptée, entre la composition écrite en anglais et le texte corrigé. L'unité d'affichage de résultats affiche les résultats de l'association effectuée par l'unité d'association. En outre, l'unité d'association effectue l'association entre les textes afin d'augmenter la distance, sur la base de critères prédéterminés, entre des mots et une pluralité de textes inclus dans la composition écrite et des mots et une pluralité de textes inclus dans le texte corrigé, jusqu'à ce qu'il n'y ait aucun texte non associé dans la composition écrite en anglais ou dans le texte corrigé. Cette configuration permet d'obtenir un système d'analyse de résultats de correction de composition écrite en anglais qui analyse une composition écrite en anglais et un texte corrigé correspondant en les comparant l'un à l'autre et qui renvoie le résultat d'analyse à l'utilisateur.
PCT/JP2021/045231 2020-12-13 2021-12-09 Système d'analyse de résultats de correction de composition écrite en anglais WO2022124355A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
JP2020-206387 2020-12-13
JP2020206387A JP7084003B1 (ja) 2020-12-13 2020-12-13 英作文添削結果分析システム

Publications (1)

Publication Number Publication Date
WO2022124355A1 true WO2022124355A1 (fr) 2022-06-16

Family

ID=81973370

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2021/045231 WO2022124355A1 (fr) 2020-12-13 2021-12-09 Système d'analyse de résultats de correction de composition écrite en anglais

Country Status (2)

Country Link
JP (1) JP7084003B1 (fr)
WO (1) WO2022124355A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005345987A (ja) * 2004-06-07 2005-12-15 Hitachi Electronics Service Co Ltd ウィークポイント学習システム及びコンピュータ・ソフトウエア
WO2006134759A1 (fr) * 2005-06-15 2006-12-21 Waseda University Dispositif d’évaluation de phrases et programme d’évaluation de phrases
JP2017167413A (ja) * 2016-03-17 2017-09-21 独立行政法人大学入試センター 採点補助システム
JP2020140442A (ja) * 2019-02-28 2020-09-03 株式会社沖データ 文書添削方法及び文書添削装置
JP2020177387A (ja) * 2019-04-17 2020-10-29 株式会社Ｎｔｔドコモ 文出力装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2007286409A (ja) * 2006-04-18 2007-11-01 Hitachi Ltd 採点支援装置
JP5695440B2 (ja) * 2011-02-22 2015-04-08 株式会社教育測定研究所 言語解析システム及び言語解析方法
JP5729058B2 (ja) * 2011-03-22 2015-06-03 大日本印刷株式会社 外国語教材作成システム
JP7067751B2 (ja) * 2017-09-28 2022-05-16 株式会社デジタル・ナレッジ 教材オーサリングシステム

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2005345987A (ja) * 2004-06-07 2005-12-15 Hitachi Electronics Service Co Ltd ウィークポイント学習システム及びコンピュータ・ソフトウエア
WO2006134759A1 (fr) * 2005-06-15 2006-12-21 Waseda University Dispositif d’évaluation de phrases et programme d’évaluation de phrases
JP2017167413A (ja) * 2016-03-17 2017-09-21 独立行政法人大学入試センター 採点補助システム
JP2020140442A (ja) * 2019-02-28 2020-09-03 株式会社沖データ 文書添削方法及び文書添削装置
JP2020177387A (ja) * 2019-04-17 2020-10-29 株式会社Ｎｔｔドコモ 文出力装置

Also Published As

Publication number Publication date
JP2022093215A (ja) 2022-06-23
JP7084003B1 (ja) 2022-06-14

Similar Documents

Publication Publication Date Title
Garnier et al. The PHaVE List: A pedagogical list of phrasal verbs and their most frequent meaning senses
JP5362353B2 (ja) 文書中のコロケーション誤りを処理すること
Rumsey How To Find Information: A Guide For Researchers: A Guide for Researchers
US6442523B1 (en) Method for the auditory navigation of text
US9208144B1 (en) Crowd-sourced automated vocabulary learning system
US8332208B2 (en) Information processing apparatus, information processing method, and program
US9384678B2 (en) System and method for generating questions and multiple choice answers to adaptively aid in word comprehension
US20090306959A1 (en) Personal text assistant
US20130149681A1 (en) System and method for automatically generating document specific vocabulary questions
US20080033712A1 (en) Method of learning a second language through the guidance of pictures
KR102050266B1 (ko) 영어 학습 서비스를 제공하는 서버, 방법 및 컴퓨터 프로그램
US9311297B2 (en) Method and system for outputting information
US20210042300A1 (en) Mapping data resources to requested teaching objectives
JP2005157524A (ja) 質問応答システムおよび質問応答処理方法
US20100094855A1 (en) System for transforming queries using object identification
Sahu et al. Evaluating performance of different grammar checking tools
JP4018673B2 (ja) 多肢選択言語試験問題自動作成プログラム
JP2006126319A (ja) テスト問題配信システム
WO2022124355A1 (fr) Système d'analyse de résultats de correction de composition écrite en anglais
Laursen et al. Using an AD-HOC corpus to write about emerging technologies for technical writing and translation: The case of search engine optimization
JP5877775B2 (ja) コンテンツ管理装置、コンテンツ管理システム、コンテンツ管理方法、プログラム、及び記憶媒体
Palumbo IS EVERYTHING A “DICTIONARY”? Exploring users’ views of online language resources
Kao et al. Design and development of the sentence-based collocation recommender with error detection for academic writing
US12026157B2 (en) Narrowing synonym dictionary results using document attributes
US20030040902A1 (en) System and method of learning a foreign language

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 21903454

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 21903454

Country of ref document: EP

Kind code of ref document: A1