JP2018163660A - Method and system for readability evaluation based on english syllable calculation method - Google Patents

Abstract

A readability evaluation method and system based on an English syllable calculation method are provided. By obtaining a target document, standardization processing is performed using a preset standardization method, conversion is performed according to conversion conditions, the obtained standardized word is scanned to calculate the total number of syllables, and finally Then, the calculation is performed based on the readability derivation formula to obtain the readability coefficient and the reading difficulty coefficient of the target document. Supports readability analysis in various ways to obtain the target document. The result of the readability analysis can be directly displayed via the display terminal, and is intuitively fed back to the user. Since the feasibility of the readability calculation is improved, the calculation process is simple, easy to implement, and the calculation result is accurate, the system can be easily spread and used in many electronic learning products. [Selection] Figure 1

Description

The present invention relates to the technical field of information processing in English, and more particularly to a readability evaluation method and system based on an English syllable calculation method.

Measuring the readability of English sentences is the most important index for objectively judging the difficulty of the sentences. Based on this index, the difficulty level of a sentence can be quantitatively explained objectively, quickly and accurately. The determination of the index has significant significance in the following fields.

Objectively and accurately select and confirm the English sentences and their difficulty levels embedded in the student materials. Different readability indicators of English texts make the selection of English texts in different grade materials from kindergarten to graduate school more scientific and accurate, thereby learning English materials that vary in difficulty and level according to grade Can do. At present, the difficulty level of sentences in English materials of different grades is determined entirely by the subjective judgment of the editor, and the knowledge, ability, preference, and educational objectives vary from editor to editor. As a result, English language materials selected as materials do not have objective, scientific, quantitative and uniform grounds and cannot meet the needs of students at different grades, different regions, different ranks and different levels. , You cannot scientifically customize materials and materials suitable for student learning.

In interpersonal communication, it is possible to attract readers and customers more quickly and widely by quickly and accurately transmitting highly readable texts or sales information. Medical,
In all industries and departments where all information is transmitted in English as a medium, such as transportation and public services, people can understand more easily readable information, making interpersonal communication faster and more accurate.

In the current Internet era, he writes for Weibo, a short mailer, WeChat, e-mail, blog, web page text, government online reports, network news, advertisements, brochures, etc. The written text has a more rapid, accurate and efficient transmission effect within a reasonable range of high readability through verification of the readability index.

Due to factors such as the complexity of pronunciation of English itself, evolution, and the influence of foreign words, artificial intelligence calculations for English syllables are one of the most difficult problems in the world. In the field of computational linguistics, the calculation of syllables is
Data-driven (machine learning) method and language rule compilation algorithm (March
and, Y. and Damper, RI (2000) .A multi-strategy approach to improving pronunc
iation by analogy. Computational Linguistics, 26 (2): 195-219, Damper, RI and
Marchand, Y. (2006). Information fusion approaches to the automatic pronunciati
on of print by analogy.Information Fusion, 7: 207-20, Marchand, Y. and Damper,
RI (2006). Can syllabification improve pronunciation by analogy of English.
Natural Language Processing, 13 (1): 1-24) Data-driven methods have the advantages of low cost, simple modeling, and fast implementation, but are highly dependent on the type of training corpus and the size of the corpus, so one type of stylistic training model can be Cannot be used correctly for type style. Expert knowledge is required for methods based on language rule compilation algorithms. However, since the level of experts is different, the rules for editing and calculating syllables are cumbersome, and it is necessary to test repeatedly, so it is now possible to compile syllable calculation rules comprehensively and completely. There is no uniform method yet, and the conventional method cannot capture and compile all the rules for different users. Therefore, in the actual application, the calculated readability index is different, which may cause misunderstanding of the reader. On the other hand, readability evaluations are based on Weibo, short mail, WeChat,
Evaluation results for e-mails, blogs, and the like are not displayed, and the search engine does not have a function for displaying search results for user readability.

  The object of the present invention is to solve at least one of the above technical drawbacks.

Accordingly, one object of the present invention is to provide a readability evaluation method and system based on an English syllable calculation method, and to perform readability evaluation on an English sentence to be read by a user using the English syllable calculation method. The task is to analyze the level of difficulty and to scientifically customize materials and materials suitable for student learning.

In order to achieve the above object, as one embodiment, a readability evaluation method based on an English syllable calculation method provided by the present invention includes a target document acquisition terminal, a standardization processing server, a readability computing server, and a readability feedback terminal. Is implemented on a readability feedback platform including
A target document is acquired by a predetermined acquisition method. The target document is divided into sentences based on the punctuation marks to obtain the total number of sentences. Step S1 performed by the target document acquisition terminal that performs word division processing on each divided sentence with a regular expression to obtain the total number of words;
Step 2 performed by the standardization processing server that performs standardization processing according to a preset standardization method using each word in the target document obtained in step S1 as an input word, and obtains a plurality of standardized words;
Using all the standardized words obtained in step S2 as new input words, the number of syllables is calculated according to preset conversion conditions. Step 3 performed by the standardization processing server in which the sum of the number of syllables of each input word is set as the total number of syllables
The following two preset readability derivations for the total number of sentences, the total number of words, and the total number of syllables
Substituting for, readability coefficient (Flesch Readability) and reading difficulty coefficient (Flesch-Kin)
step 4 performed by the readability computing server for calculating the caid level);
Step 5 performed by the readability feedback terminal for simultaneously displaying the calculation result of the readability coefficient and the reading difficulty coefficient and the obtained target document and feeding back to the user.

The predetermined acquisition method is:
Search 1 based on keywords using a search engine that includes all websites that support English search and search the text snippet or document that meets the requirements from the index 2; And any of the other methods.

When the method 1 is used in step S1, the calculation result of the readability coefficient and the reading difficulty coefficient is calculated below the title or title of the target document or text snippet by using the target document or text snippet searched by the search engine and the keyword. Are displayed at the same time above.

When using method 2 in step S1, obtaining the target document entered by the user by detecting the window of the readability feedback platform and triggering the selection of the readability button, the target document is analyzed by the readability feedback platform; The readability coefficient and the reading difficulty coefficient are calculated, and the calculation results are automatically displayed above the title of the window in the readability feedback platform.

Preferably, the preset standardization method in step S2 includes conversion of a suffix character and addition of a prefix at the beginning.

In the conversion of the suffix character, the input word is divided into a general suffix character and a conversion required suffix character. The required conversion suffix characters are 5 suffixes: 2 suffixes, 3 suffix characters, 4 suffix characters, 5 suffix characters, and 7 suffix characters.
Includes one category. The general suffix characters are not converted. The last n characters in the suffix n characters are converted into the corresponding suffix conversion key. Here, n = 2, 3, 4, 5, and 7.

In the prefix addition, the input word is divided into a general prefix character and a conversion required prefix character. The conversion required prefix characters include four categories: prefix 2 characters, prefix 3 characters, prefix 4 characters, and prefix 5 characters. The general prefix character is not converted. The first m characters in the prefix m characters are converted into corresponding prefix conversion keys. Here, m = 2, 3, 4, and 5.

Preferably, the preset conversion condition in step S3 is such that when all of the input words are consonant alphabets that do not include “y”, the number of syllable outputs is made equal to the word length of the input word,
If the input word contains any vowel alphabet or the alphabet “y” and the word length is less than 3, output the number of syllables as 1,
If the input word contains any vowel alphabet or the letter “y” and the word length is equal to 3,
(1) When the first and third characters are both “y” of an arbitrary vowel alphabet or alphabet and the second is a consonant alphabet, the number of syllables is output as 2,
(2) If the first, second and third characters are any vowel alphabet or “y” of the alphabet, the number of syllables is output as 3,
(3) In cases other than the above (1) and (2), processing is performed under any condition that the number of syllables is output as 1, and an arbitrary vowel alphabet or “y” of the alphabet is included in the input word. If it is included and the word length is greater than 3, the input word is scanned from the beginning to the end. If a feature word string exists in the input word, the feature word string is replaced with a corresponding extended word string to obtain a replacement word. Then, the number of vowel strings in the replacement word is calculated. One or a plurality of “a, e, i, o, u, or y” combined with one vowel string (that is, a plurality of vowel strings in a replacement word are separated by a consonant alphabet excluding y) "A, e, i, o, u or y" to be combined
A single vowel string unless any consonant alphabet excluding y exists in the middle, regardless of the quantity of When the last character of the replacement word is “e”, the last character of the replacement word is deleted. When the first two characters of the replacement word are “mc”, the number of syllables is obtained by adding 1 to the number of vowel strings. If the first two letters of the replacement word are not “mc”, the number of syllables is made equal to the number of vowel strings.

Preferably, it is composed of three characters, the first character is an arbitrary alphabet from alphabet a to alphabet z, and the combination of the second character and the third character is:
One of “ia”, “io”, “iu”, “ae”, “eo”, and “ii”.
The extended word string is composed of 4 characters. The first word, the second character, and the fourth character of the extended word string are assumed to be the same as the first character, the second character, and the third character of the corresponding feature word string, respectively. Is defined. The feature word string is defined as the third character being the same as the first character of the feature word string.

The present invention further provides a system as a readability evaluation application platform based on the English syllable calculation method. The application platform includes a target document acquisition terminal, a standardization processing server, a readability computing server, and a readability feedback terminal. The target document acquisition terminal acquires a target document by a predetermined acquisition method, and transmits the target document to the standardization processing server,
Total text of the target document obtained after the target document is pre-processed by performing a pre-processing that performs a word division process on the target document based on punctuation marks, and a word division process using a regular expression for each divided sentence. The number and the total number of words are transmitted to the readability computing server.

The standardization processing server reads the target document, performs standardization processing on the target document according to the preset standardization method, acquires the total syllable number according to the preset conversion condition, and transmits the total syllable number to the readable computing server. A standardization processing module for obtaining a standardized word, and a scan conversion module for calculating the number of syllables according to a preset conversion condition for the standardized word. The sum of the number of syllables of each input word is taken as the total number of syllables.

The readability computing server calculates the total number of sentences, the total number of words, and the total number of syllables to obtain a readability coefficient and a reading difficulty coefficient.
Preset readability derivation formula including formula for calculating the reading difficulty factor (Flesch-Kincaid Level)
Assign to.

The readability feedback terminal is used for simultaneously displaying the calculation result of the readability coefficient and the reading difficulty coefficient and the acquired target document as an active window on the display and feeding back to the user.

Preferably, in the standardization processing module, a preset standardization method including conversion of a suffix character and addition of a prefix at the beginning is set.

In the conversion of the suffix character, the input word is divided into a general suffix character and a conversion required suffix character. The required conversion suffix characters are 5 suffixes: 2 suffixes, 3 suffix characters, 4 suffix characters, 5 suffix characters, and 7 suffix characters.
Includes one category. The general suffix characters are not converted. The last n characters in the suffix n characters are converted into the corresponding suffix conversion key. Where n
= 2, 3, 4, 5, 7.

In the prefix addition, the input word is divided into a general prefix character and a conversion required prefix character. The conversion required prefix characters include four categories: prefix 2 characters, prefix 3 characters, prefix 4 characters, and prefix 5 characters. The general prefix character is not converted. Prefix m
The first m characters in the character are converted to the corresponding prefix conversion key. Here, m = 2, 3, 4, and 5.

Preferably, a preset conversion condition is set in the scan conversion module. The preset conversion conditions are:
If all of the input words are consonant alphabets that do not contain “y”, make the number of syllable outputs equal to the word length of the input word,
If the input word contains any vowel alphabet or the alphabet “y” and the word length is less than 3, output the number of syllables as 1,
If the input word contains any vowel alphabet or the letter “y” and the word length is equal to 3,
(1) When the first and third characters are both “y” of an arbitrary vowel alphabet or alphabet and the second is a consonant alphabet, the number of syllables is output as 2,
(2) If the first, second and third characters are any vowel alphabet or “y” of the alphabet, the number of syllables is output as 3,
(3) In cases other than the above (1) and (2), the process is performed under any condition that the number of syllables is output as one.

If the input word contains any vowel alphabet or alphabet “y” and the word length is greater than 3, the input word is scanned from the beginning to the end. If a feature word string exists in the input word, the feature word string is replaced with a corresponding extended word string to obtain a replacement word. Calculate the number of vowel strings in the replacement word. One or a plurality of “a, e, i, o, u, or y” combined with one vowel string (that is, a plurality of vowel strings in a replacement word are separated by a consonant alphabet excluding y) Regardless of the number of “a, e, i, o, u, or y” to be combined, it is only 1 unless one of the consonant alphabets excluding y exists in the middle.
Two vowel strings). When the last character of the replacement word is “e”, the last character of the replacement word is deleted. When the first two characters of the replacement word are “mc”, the number of syllables is obtained by adding 1 to the number of vowel strings. If the first two letters of the replacement word are not “mc”, the number of syllables is made equal to the number of vowel strings.

Preferably, it is composed of three characters, the first character is an arbitrary alphabet from alphabet a to alphabet z, and the combination of the second character and the third character is:
A feature word string is defined as being one of “ia”, “io”, “iu”, “ae”, “eo”, and “ii”. It is composed of four characters, and the first character, the second character, and the fourth character are the same as the first character, the second character, and the third character of the corresponding feature word string, respectively. The extended word string is defined as the first character being the same as the first character of the feature word string.

The readability evaluation method and system based on the English syllable calculation method according to the present invention has at least the following merits over the conventional readability evaluation method.

1. Regardless of whether the document is a document input by the user, an electronic document scanned by a scanning device, or a document searched via a network, the readability analysis is performed by the method. These results can be directly displayed via the display terminal, and can be fed back intuitively to the user.

2. In the system according to the present invention, by making syllable calculation an important one in the readability calculation formula, the feasibility of readability calculation is improved and the calculation process is simple and easy to implement,
Since the calculation result is accurate, the system can be easily spread and used in many electronic learning products.

Additional aspects and advantages of the present invention are set forth in part in the following description. Some will become more apparent from the following description or will become apparent from the practice of the invention.

The above contents and / or additional aspects and advantages of the present invention will be clearly and easily understood in the description of the embodiments in conjunction with the following drawings.

It is a flowchart of the usage method based on the English syllable calculation method in readability evaluation concerning the present invention. It is an interface diagram of a user who inputs a target document in Embodiment 1 of the present invention. It is an interface figure which displays a readability calculation result and an original document simultaneously in Embodiment 2 of the present invention. It is another interface figure which displays the readability calculation result and the original document simultaneously in Embodiment 2 of this invention. It is a connection block diagram of the readability evaluation system based on the English syllable calculation method according to the present invention.

Hereinafter, embodiments of the present invention will be described in detail. Illustrative embodiments are illustrated in the drawings, wherein the same or similar reference numerals indicate the same or similar elements or elements having the same or similar functions. Hereinafter, the embodiments described with reference to the drawings are exemplary for explaining the present invention, and should not be construed as limiting the present invention.

As shown in FIG. 1, the readability evaluation method based on the English syllable calculation method according to the embodiment of the present invention includes the following steps.

Step 1: A target document is acquired by a predetermined acquisition method. Punctuation (punctuatio
n mark), the target document is divided into sentences to obtain the total number of sentences. Each divided sentence is subjected to word division processing using a regular expression to obtain the total number of words. The target document acquisition method may be any of the following, but is not limited thereto.

Method 1: A search engine is used to search based on keywords, and a document that satisfies the requirements is searched from the index.

  Method 2: The user inputs a document.

  Method 3: An electronic document is formed by means such as scanning a paper document.

Standardization processing is performed on each word of the input text or each word of the target document. Standardization processing is an important part of improving the accuracy of syllable calculations. "The CMU (Carnegie Mellon University, USA) Pronouncing Dictionary, Oxford Advanced Encyclopedia (Oxford Advanced En
Based on repeated tests on the three major dictionaries “glish Dictionary” and “Webster's Dictionary”, the accuracy of vocabulary syllable calculation is only about 48% to 65% when standardization is not performed. However, the calculation accuracy after the standardization process can reach 90%. The standardization process is to convert a complicated and irregular alphabet combination into a rule that is easy to identify and calculate. Each word after word division or each word of the target document found by the search is regarded as an input word, and the vocabulary standardized and converted in each step is set as an input word in the next step.

Step 2: Using each word in the target document obtained in Step 1 as an input word, standardization processing is performed according to a preset standardization method to obtain a plurality of standardized words. Preset standardization methods include the conversion of suffix characters (suffixes) and the addition of prefixes (prefixes) in front of word bases.

Convert suffix characters. The input word is divided into general suffix characters and conversion required suffix characters. The conversion required suffix characters include five categories: suffix 2 characters, suffix 3 characters, suffix 4 characters, suffix 5 characters, and suffix 7 characters. The general suffix characters are not converted. The last n characters (n = 2, 3, 4, 5, 7) in the suffix n characters are converted into corresponding suffix conversion keys. Suffix conversion keys are "consonant", "consonant + vowel + consonant", "consonant + vowel + consonant + vowel", "vowel + consonant + vowel + consonant" or "vowel + consonant + vowel + consonant + vowel"
Of the form.

The last three characters of the input word are “s # 39s” (# 39 is the US standard code for information exchange (A
Fig. 2 shows an apostrophe in SCII). ), "Ses", "sms", "ces"
, “Hes”, “ges”, “xes”, “yer”, “ile”, “cre”, “el
If it is “y”, “h # 39s” or “ole”, these three characters are converted to “dad”. Here, “dad” (that is, “consonant + vowel + consonant”) is a suffix conversion key corresponding to these three ending characters. When the last three characters of the input word are “tua” or “ua”, the conversion key converted from these three characters is “tada” (that is, “consonant + vowel + consonant + vowel”). The last three letters of the input word are the French suffix “ier”
"Or" que ", these characters are converted to" k "(ie," consonant "). The last three characters of the input word are “les”, and the fourth character from the last is “a”, “e”
, “I”, “o”, “u” and “y”, delete “les”,
Add the suffix “son” to the end of the word.

The last 7 characters of the input word are “eations”, “iations”, “oations”
”Or“ uations ”, these seven characters are converted to“ adada ”.

If the last two characters of the input word are “sm”, these two characters are replaced by “dob” (ie, “
Consonant + vowel + consonant ").

The last 5 characters of the input word are “ions”, “sians”, “cians” or “ti”
If “ans”, these five characters are converted to “dob”.

The last 4 characters of the input word are “eate”, “iate”, “oate”, “vour”,
If it is “uous” or “uate”, these four characters are converted to “adad”.
The last four characters of the input word are “sm # 39s”, “cian”, “eous”, “bled”
, “Gled”, “dled”, “kled”, “pled”, “tled” or “fl”
If “ed”, these four characters are converted to “dob”. The last 4 characters of the input word
In the case of “sion”, “ious”, “oate”, “ce # 39s” or “cial”, these four characters are converted to “dob”.

Add a prefix to the beginning of the word. Divide input words into general prefix characters and conversion prefix characters. The conversion required prefix characters include four categories: prefix 2 characters, prefix 3 characters, prefix 4 characters, and prefix 5 characters. Do not convert general prefix characters. The first m (m = 2, 3, 4, 5) characters in the prefix m characters are converted into corresponding prefix conversion keys. Prefix conversion keys are “consonant + vowel + consonant”, “consonant + vowel + consonant + vowel”,
The format is “consonant + vowel + consonant + vowel + consonant” or “consonant + vowel + consonant + vowel + consonant + vowel”.

If the first two letters of the input word are “io”, these two letters are changed to “dido” (ie,
"Consonant + vowel + consonant + vowel").

If the first 5 characters of the input word are “coinc”, these 5 characters are replaced with “didod”.
Convert to If the first five characters of the input word are “heroi”, these five characters are
to "idada".

If the first four characters of the input word are “nois”, “goin” or “voice”, these four characters are converted to “didod”.

If the first three characters of the input word are “rea” and the word length of the input word is greater than 4, these three characters are converted to “rera”.

Step 3: Using all the standardized words obtained in Step 2 as new input words, the number of syllables is calculated according to preset conversion conditions. The sum of the number of syllables of each input word is taken as the total number of syllables.
The preset conversion conditions are as follows.

If all of the input words are consonant alphabets that do not contain “y”, the number of syllable outputs is made equal to the word length of the input word.

If the input word includes an arbitrary vowel alphabet or “y” of the alphabet, and the word length is less than 3, the number of syllables is output as 1.

When the input word includes an arbitrary vowel alphabet or “y” of the alphabet and the word length is equal to 3, processing is performed under the following conditions.

(1) If the first and third characters are both “y” of an arbitrary vowel alphabet or alphabet and the second is a consonant alphabet, the number of syllables is output as two.

(2) When the first, second, and third characters are any vowel alphabet or “y” of the alphabet, the number of syllables is output as 3.

  (3) In cases other than (1) and (2) above, the syllable number is output as 1.

If the input word contains any vowel alphabet or alphabet “y” and the word length is greater than 3, the input word is scanned from the beginning to the end. If a feature word string exists in the input word, the feature word string is replaced with a corresponding extended word string to obtain a replacement word. Then, the number of vowel strings in the replacement word is calculated. One or a plurality of “a, e, i, o, u, or y” combined with one vowel string (that is, a plurality of vowel strings in a replacement word are separated by a consonant alphabet excluding y) "A, e, i, o, u or y" to be combined
A single vowel string unless any consonant alphabet excluding y exists in the middle, regardless of the quantity of When the last character of the replacement word is “e”, the last character of the replacement word is deleted. When the first two characters of the replacement word are “mc”, the number of syllables is obtained by adding 1 to the number of vowel strings. If the first two characters of the replacement word are not “mc”, the number of syllables is equal to the number of vowel strings.

The first character is an arbitrary alphabet from alphabet a to alphabet z, and the combination of the second character and the third character is “ia”, “i”.
A feature word string is defined as being one of “o”, “iu”, “ae”, “eo”, and “ii”. It is composed of four characters, and the first character, the second character, and the fourth character are the same as the first character, the second character, and the third character of the corresponding feature word string, respectively. The extended word string is defined as the first character being the same as the first character of the feature word string.

For example, when the input word is “abiocoen” including “io”, “bio” is a feature word string, and is replaced with the extended word string “bibo” to obtain the replacement word “abibocoen”. Since “abibocoen” includes four vowel strings “a”, “i”, “o”, and “oe”, and the first two characters are not “mc”, the number of syllables of the input word “abbiocoen” is four. It is.

For example, when the input word is “McGovern” that does not include the feature word string, the replacement word and the input word are the same. Since “McGovern” includes two vowel strings “o” and “e”, and the first two letters are “mc”, the number of syllables of the input word “McGovern” is three.

Further, when the input word is “grime” that does not include the feature word string, the replacement word and the input word are the same, and the last character of “grime” is “e”. The new replacement word after deleting “e” is “grim”, and since “grim” includes only one vowel string “i”, the number of syllables of the input word “girl” is 1. .

Step 4: Preset readability formula for total number of sentences, total number of words, and total number of syllables
Substituting for, readability coefficient (Flesch Readability) and reading difficulty coefficient (Flesch-Kin)
caid Level).

Step 5: In the readability feedback platform, the calculation result of the readability coefficient and the reading difficulty coefficient and the acquired target document are simultaneously displayed and fed back to the user. Display methods include, but are not limited to, the following several types.

Method 1: Using the target document or text snippet searched by the search engine and the keyword, the calculation result of the readability coefficient and the reading difficulty coefficient is simultaneously below the target document or text snippet or above the title. Display.

Method 2: In the readability feedback platform window, the user enters the target document and selects the readability button. The calculation result of the readability coefficient and the reading difficulty coefficient is automatically displayed above the title of the window in the readability feedback platform.

As shown in FIG. 5, the readability evaluation system based on the English syllable calculation method includes a target document acquisition terminal 1, a standardization processing server 2, a readability computing server 3, and a readability feedback terminal 4.

The target document acquisition terminal 1 is used for acquiring a target document by an acquisition method and transmitting the target document to a standardization processing server. The target document acquisition terminal performs preprocessing on the target document, performs sentence division processing based on punctuation marks, and performs word division processing on each divided sentence using a regular expression. The target document acquisition terminal transmits the total number of sentences and the total number of words of the target document acquired after the target document is preprocessed to the readable computing server.

The standardization processing server 2 includes a standardization processing module in which a preset standardization method is set, and a scan conversion module. Preset standardization methods include suffix character conversion and prefix prefix addition.

Convert suffix characters. The input word is divided into general suffix characters and conversion required suffix characters. The conversion required suffix characters include five categories: suffix 2 characters, suffix 3 characters, suffix 4 characters, suffix 5 characters, and suffix 7 characters. Do not convert general suffix characters. The last n characters in the suffix n characters are converted into the corresponding suffix conversion key. Where n = 2, 3,
4, 5, and 7. Add a prefix to the beginning of the word. Divide input words into general prefix characters and conversion prefix characters. Conversion required prefix characters are 4 prefix characters, 3 prefix characters, 4 prefix characters, 5 prefix characters.
Includes one category. Do not convert general prefix characters. The first m characters in the prefix m characters are converted into corresponding prefix conversion keys. Here, m = 2, 3, 4, and 5.

Preset conversion conditions are set in the scan conversion module. The preset conversion conditions are as follows.

If all of the input words are consonant alphabets that do not contain “y”, the number of syllable outputs is made equal to the word length of the input word.

If the input word includes an arbitrary vowel alphabet or “y” of the alphabet, and the word length is less than 3, the number of syllables is output as 1.

When the input word includes an arbitrary vowel alphabet or “y” of the alphabet and the word length is equal to 3, processing is performed under the following conditions.
(1) If the first and third characters are both “y” of an arbitrary vowel alphabet or alphabet and the second is a consonant alphabet, the number of syllables is output as two.
(2) When the first, second, and third characters are any vowel alphabet or “y” of the alphabet, the number of syllables is output as 3.
(3) In cases other than (1) and (2) above, the syllable number is output as 1.

If the input word contains any vowel alphabet or alphabet “y” and the word length is greater than 3, the input word is scanned from the beginning to the end. If a feature word string exists in the input word, the feature word string is replaced with a corresponding extended word string to obtain a replacement word. Calculate the number of vowel strings in the replacement word. One or a plurality of “a, e, i, o, u, or y” combined with one vowel string (that is, a plurality of vowel strings in a replacement word are separated by a consonant alphabet excluding y) Regardless of the number of “a, e, i, o, u, or y” to be combined, it is only 1 unless one of the consonant alphabets excluding y exists in the middle.
Two vowel strings). When the last character of the replacement word is “e”, the last character of the replacement word is deleted. When the first two characters of the replacement word are “mc”, the number of syllables is obtained by adding 1 to the number of vowel strings. If the first two characters of the replacement word are not “mc”, the number of syllables is equal to the number of vowel strings.

In order to obtain the readability coefficient and the reading difficulty coefficient, the readability computing server 3 calculates the total number of sentences, the total number of words, and the total syllable number, the calculation formula of the readability coefficient (Flesch Readability), and the reading difficulty coefficient (Flesch-Kincaid). Level readability derivation formula including formula
Assign to.

The readability feedback terminal 4 is used for fetching the calculation results of the readability coefficient and the reading difficulty coefficient into the target document and feeding it back to the user via the display device.

The target document acquisition method by the target document acquisition terminal may be any of the following, but is not limited thereto. Method 1: A search engine is used to search based on keywords, and a document that satisfies the requirements is searched from the index. Method 2: The user inputs a document. Method 3: An electronic document is formed by means such as scanning a paper document. Regardless of whether it is a document input by the user, an electronic document scanned by a scanning device, or a document searched through a network, the readability analysis is performed by the method, and the result of the readability analysis Can be directly displayed via the display terminal, and can be fed back intuitively to the user.

In the first embodiment shown in FIG. 2, the target document is acquired by the second acquisition method (character input method by the user) employed in the present invention. The target document shown in the figure is Irish writer James Joyce, a master of the flow of consciousness (
These are the first two paragraphs excerpted from a famous work by James Joyce called Finnegans Wake. This work is widely recognized in the West as the most ambiguous and unclear novel in English-speaking countries. This is because the author uses a lot of self-made words, uncommon words, strange words, ancient slang, mixed words, long words, etc. These words are either not or rarely recorded in the English dictionary, so the readability cannot be calculated by the usual method of examining the syllables in the dictionary. The calculation result of readability can be obtained by the standardization processing method and preset conversion conditions according to the present invention. Further, the readability formula and the target document can be simultaneously displayed on the display window in the readability evaluation system.

FIG. 3 is an interface diagram for simultaneously displaying the readability calculation result and the original document in the second embodiment. FIG. 4 is another interface diagram for simultaneously displaying the readability calculation result and the original document. The two figures used in the present invention are Twitter (Twitter) (
This is a result fed back after searching the keyword “climate” in the registered trademark. The readability index is captured in each tweet or miniblog of the search result and fed back to the user. Because Twitter is an instant messenger social tool, there are many internet terms and abbreviations as the characteristics of the language of tweets, Internet addresses are mixed in tweets, and many words are self-made or spoken, so many vocabularies are not in the English dictionary. However, there is little recording, and the normal method of examining syllables in the dictionary cannot calculate readability. If the method according to the present invention is used to search for tweets related to the keyword “climate”, the readability computing server calculates the readability of each tweet, and calculates the readability coefficient and reading difficulty coefficient that are calculated. Each tweet is displayed at the same time. For example, a tweet having the title “Pure Climate Skeptic” in FIG. 3 has a readability coefficient of 21.8 and a reading difficulty coefficient of 34.8.

Note that when retrieving the target document, the search engine for searching for keywords is
List all search sites, network platforms, and network databases that can be searched in English. Common ones are Twitter, Google (Google), Baidu (
Baidu), Sanlurin search (360 Search), and road search (youdao).

The rationale of the present invention is derived from two methods for calculating syllables in computational linguistics, a method based on data driving (machine learning) and a method based on rules. Data driven methods are derived from statistics on training data and are highly dependent on the size and type of training data. A model established based on training data can only predict target data similar to the level of training data. If the target data is significantly different from the model data, the prediction accuracy of the model is greatly reduced. The rules-based method derives from expert knowledge and is based on a long-standing, highly conceptualized and unified comprehensive experience of linguistic phenomena in linguists. The rules of the present invention are 3
Established based on a large English dictionary. The present invention is highly versatile and adaptable to various texts, is simple to execute applications, does not need to be trained on various types of models, is highly scalable and fast. Can be applied to various types and sizes of text.

The method according to the present invention is determined by trial and error based on a large amount of text. After testing the three major English dictionaries, the accuracy rate for calculating the number of syllables is
CMU Pronouncing Dictionary is 92%, Oxford Modern English-English Dictionary is 90%, Webster Dictionary is 90%.

In this description, “one embodiment”, “some embodiments”, “exemplary”,
Descriptions of reference terms such as “specific examples” or “several illustrations” include those specific characteristics, structures, materials, or features described in conjunction with the embodiments or examples in at least one embodiment or Included in the illustration. In the present specification, exemplary expressions for the above terms do not necessarily indicate the same embodiments or the same examples. Also, the particular characteristics, structures, materials, or features described may be combined in any suitable form in any one or more embodiments or examples.

Claims (8)

A readability evaluation method based on an English syllable calculation method, which is realized in a readability feedback platform including a target document acquisition terminal, a standardization processing server, a readability computing server, and a readability feedback terminal.
The target document is acquired by a predetermined acquisition method, the target document is divided into sentences based on the punctuation marks, the total number of sentences is obtained, and word division processing is performed on each divided sentence by a regular expression to obtain the total number of words. Step S1 performed by the target document acquisition terminal,
Step 2 performed by the standardization processing server that performs standardization processing according to a preset standardization method using each word in the target document obtained in step S1 as an input word, and obtains a plurality of standardized words;
A standardization processing server that calculates all standardized words obtained in step S2 as new input words, calculates the number of syllables according to preset conversion conditions, and sets the sum of the number of syllables of each input word as the total syllable Step 3 to perform,
Preset readability derivation formula for total sentence count, total word count, and total syllable count
Substituting for, readability coefficient (Flesch Readability) and reading difficulty coefficient (Flesch-Kin)
caid Level) is performed by the readability computing server,
A step 5 performed by the readability feedback terminal, which simultaneously displays the calculation result of the readability coefficient and the reading difficulty coefficient and the obtained target document and feeds back to the user;
The predetermined acquisition method is:
Search for text snippets or documents that meet the requirements from the index using all search engines that support English search, and index 1,
Any one of method 2 and other methods in which a user inputs a document;
When Method 1 is used in step S1, the calculation result of the readability coefficient and the reading difficulty coefficient is calculated below the target document or text snippet by using the target document or text snippet searched by the search engine and the keyword. Or display it at the same time above the title,
When method 2 is used in step S1, the target document input by the user is obtained by detecting the window of the readability feedback platform, and when the readability selection button is triggered, the target document is analyzed by the readability feedback platform. The readability coefficient and the reading difficulty coefficient are calculated, and the calculation results are displayed above the title of the window in the readability feedback platform. The preset standardization method in step S2 includes conversion of a suffix character and addition of a prefix at the beginning,
In the conversion of the suffix characters, the input word is divided into general suffix characters and conversion required suffix characters. The conversion required suffix characters are 5 suffixes, 3 suffixes, 4 suffix characters, 5 suffix characters and 7 suffix characters. It contains two categories and does not convert general suffix characters.
= 2, 3, 4, 5, 7) convert the last n characters to the corresponding suffix conversion key,
In adding the prefix at the beginning, the input word is divided into a general prefix character and a conversion required prefix character, and the conversion required prefix character is prefix 2.
It includes four categories of characters, prefix 3 characters, prefix 4 characters, prefix 5 characters, the general prefix characters are not converted, and the prefix m
The readability evaluation based on the English syllable calculation method according to claim 1, wherein the first m characters in the characters (m = 2, 3, 4, 5) are converted into corresponding prefix conversion keys. Method. The preset conversion conditions in step S3 are:
If all of the input words are consonant alphabets that do not contain “y”, make the number of syllable outputs equal to the word length of the input word,
If the input word contains any vowel alphabet or the alphabet “y” and the word length is less than 3, output the number of syllables as 1,
If the input word contains any vowel alphabet or the letter “y” and the word length is equal to 3,
(1) When the first and third characters are both “y” of an arbitrary vowel alphabet or alphabet and the second is a consonant alphabet, the number of syllables is output as 2,
(2) If the first, second and third characters are any vowel alphabet or “y” of the alphabet, the number of syllables is output as 3,
(3) In cases other than the above (1) and (2), the process is performed under any condition that the syllable number is output as 1,
If the input word contains any vowel alphabet or “y” of the alphabet and the word length is greater than 3, the input word is scanned from the beginning to the end, and if the input word has a characteristic word string, the characteristic word string Is replaced with the corresponding extended word string to obtain a replacement word, and the number of vowel strings in the replacement word is calculated,
One or a plurality of combined “a, e, i, o, u, or y” is converted into one vowel string (ie, a plurality of vowel strings in a replacement word are separated by a consonant alphabet excluding y). A character string, regardless of the quantity of “a, e, i, o, u or y” combined, it is only one vowel string as long as there is no consonant alphabet in the middle except y) When the last character of the replacement word is “e”, the last one character of the replacement word is deleted, and when the first two characters of the replacement word are “mc”, the number of syllables is expressed as a vowel string. 2. The English syllable according to claim 1, wherein if the first two characters of the replacement word are not “mc”, the number of syllables is equal to the number of vowel strings. Readability evaluation method based on calculation method. The first character is an arbitrary alphabet from alphabet a to alphabet z, and the combination of the second character and the third character is “ia”, “i”.
The feature word string is defined as being one of “o”, “iu”, “ae”, “eo”, and “ii”, is composed of four characters, the first character, the second character The character and the fourth character are the same as the first character, the second character, and the third character of the corresponding feature word string, respectively, and the third character is the same as the first character of the feature word string 4. The method of evaluating readability based on an English syllable calculation method according to claim 3, wherein an extended word string is defined as a thing. A target document acquisition terminal, a standardization processing server, a readability computing server, and a readability feedback terminal,
The target document acquisition terminal acquires a target document by a predetermined acquisition method, transmits the target document to the standardization processing server, performs a sentence division process on the target document based on punctuation marks, and A pre-processing for performing word division processing on each divided sentence with a regular expression is performed, and the total number of sentences and the total number of words of the target document acquired after the target document is pre-processed are transmitted to the readable computing server. ,
The standardization processing server reads the target document, performs standardization processing on the target document according to the preset standardization method, acquires the total syllable number according to the preset conversion condition, and transmits the total syllable number to the readable computing server. Including a standardization processing module for obtaining a standardized word, and a scan conversion module for calculating the number of syllables according to a preset conversion condition for the standardized word, the number of syllables of each input word Is the total number of syllables,
The readability computing server calculates the total number of sentences, the total number of words, and the total number of syllables to obtain a readability coefficient and a reading difficulty coefficient.
Preset readability derivation formula including formula for calculating the reading difficulty factor (Flesch-Kincaid Level)
Assigned to
The readability feedback terminal is used to simultaneously display the calculation result of the readability coefficient and the reading difficulty coefficient and the acquired target document as an active window on the display and to feed back to the user. Readability evaluation system based on calculation method. In the standardization processing module, a preset standardization method including suffix character conversion and prefix prefix is set.
In the conversion of the suffix characters, the input word is divided into general suffix characters and conversion required suffix characters. The conversion required suffix characters are 5 suffixes, 3 suffixes, 4 suffix characters, 5 suffix characters and 7 suffix characters. It contains two categories and does not convert general suffix characters.
= 2, 3, 4, 5, 7) convert the last n characters to the corresponding suffix conversion key,
In adding the prefix at the beginning, the input word is divided into a general prefix character and a conversion required prefix character, and the conversion required prefix character is prefix 2.
It includes four categories of characters, prefix 3 characters, prefix 4 characters, prefix 5 characters, the general prefix characters are not converted, and the prefix m
6. The readability evaluation based on the English syllable calculation method according to claim 5, wherein the first m characters in the characters (m = 2, 3, 4, 5) are converted into corresponding prefix conversion keys. system. Preset conversion conditions are set in the scan conversion module, and the preset conversion conditions are:
If all of the input words are consonant alphabets that do not contain “y”, make the number of syllable outputs equal to the word length of the input word,
If the input word contains any vowel alphabet or the alphabet “y” and the word length is less than 3, output the number of syllables as 1,
If the input word contains any vowel alphabet or the letter “y” and the word length is equal to 3,
(1) When the first and third characters are both “y” of an arbitrary vowel alphabet or alphabet and the second is a consonant alphabet, the number of syllables is output as 2,
(2) If the first, second and third characters are any vowel alphabet or “y” of the alphabet, the number of syllables is output as 3,
(3) In cases other than the above (1) and (2), the process is performed under any condition that the syllable number is output as 1,
The input word contains any vowel alphabet or the letter “y” and the word length is 3
If it is larger, the input word is scanned from the beginning to the end, and if there is a feature word string in the input word, the feature word string is replaced with the corresponding extended word string to obtain a replacement word, and the number of vowel strings in the replacement word Calculate
One or a plurality of combined “a, e, i, o, u, or y” is converted into one vowel string (ie, a plurality of vowel strings in a replacement word are separated by a consonant alphabet excluding y). A character string, regardless of the quantity of “a, e, i, o, u or y” combined, it is only one vowel string as long as there is no consonant alphabet in the middle except y) When the last character of the replacement word is “e”, the last one character of the replacement word is deleted, and when the first two characters of the replacement word are “mc”, the number of syllables is expressed as a vowel string. 6. The English syllable according to claim 5, wherein if the first two characters of the replacement word are not "mc", the number of syllables is equal to the number of vowel strings. Readability evaluation system based on calculation method. The first character is an arbitrary alphabet from alphabet a to alphabet z, and the combination of the second character and the third character is “ia”, “
A feature word string is defined as being one of io, iu, ae, eo, and ii, and is composed of four characters, the first character, the second character The character and the fourth character are the same as the first character, the second character, and the third character of the corresponding feature word string, respectively, and the third character is the same as the first character of the feature word string 8. The readability evaluation system based on the English syllable calculation method according to claim 7, wherein an extended word string is defined as a thing.