CN110232180A - A kind of automatic proposition method and system towards classic poetry assessment - Google Patents
A kind of automatic proposition method and system towards classic poetry assessment Download PDFInfo
- Publication number
- CN110232180A CN110232180A CN201910490907.3A CN201910490907A CN110232180A CN 110232180 A CN110232180 A CN 110232180A CN 201910490907 A CN201910490907 A CN 201910490907A CN 110232180 A CN110232180 A CN 110232180A
- Authority
- CN
- China
- Prior art keywords
- word
- poem
- verse
- candidate
- sentence
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/10—Text processing
- G06F40/166—Editing, e.g. inserting or deleting
Abstract
A kind of automatic proposition method and system towards classic poetry assessment.The present invention provides a kind of automatic Examination Paper Generating Systems towards classic poetry assessment, including poem library, label multiple-choice question proposition module, system selection topic proposition module, upper and lower sentence multiple-choice question proposition module, verse word filling topic proposition module, poem sequence proposition module, wherein: label multiple-choice question proposition module, for generating label multiple-choice question；Poem system selection inscribes proposition module, for generating system selection topic；Upper and lower sentence multiple-choice question proposition module, for generating sentence multiple-choice question up and down；Poem sequence proposition module, for generating sentence multiple-choice question up and down；Verse word filling inscribes proposition module, for generating word filling topic.In the present invention, the poem type of writing and semantic feature can be farthest excavated, the classic poetry exam pool of high quality is automatically generated, for efficiently, scientifically realizing the assessment of classic poetry attainment.
Description
Technical field
The present invention relates to field of computer technology, and in particular to it is a kind of towards classic poetry assessment automatic proposition method and be
System.
Background technique
With being constantly progressive for computer technology and hardware computing capability, artificial intelligence achieves many technological break-throughs, such as
AlphaGo can surpass go world champion by calculating, but in creative or artistry field, artificial intelligence still can not win
Appoint related work, such as classical Chinese poetry is a kind of art of language, literary achievements and artistic value with higher.It is ancient
Poem is provided simultaneously with regularity and abstractness, and the level and oblique tone rule of the different style of a verse, poem, etc.s has regulation, each also rhyme to be needed to match, sternly
The regulations of lattice is so that classic poetry has the aesthetic feeling on pronunciation and rhythm, simultaneously because Chinese culture is extensive, Chinese character meaning is abundant, it is right
There is also a variety of possibilities for the understanding of one poem meaning.
In Chinese language examination or poem class program or game, there is the proposition of poem class, common mode has:
(1) multiple-choice question: such as providing poem, selects content, emotion expressed by poem etc., also can choose the body of poem
System, such as five speech ancient poetry, poem with five characters in one line, seven-word poem；It can also provide sentence (lower sentence), select lower sentence (upper sentence).
(2) poem sequence is inscribed: being given four simple sentences, is selected four simple sentences and correctly put in order.
(3) poem gap-filling questions.The poem of a given hiatus or sentence, needs to fill out the word or sentence of gaps and omissions.
In either case, carrying out proposition based on classic poetry data is all a significant challenge.Firstly, this is to proposition
The artistic appreciation and proposition experience of people has higher requirements；Secondly, artificial proposition time cost is higher, need to undergo proposition, audit
Etc. multiple steps；Third, the quality of exam pool lack quantization approach, be easy to appear difficulty can not be measured, topic type is single, assessment object
The problems such as single.
Summary of the invention
In view of the deficiencies of the prior art, the present invention provides a kind of automatic proposition method and system towards classic poetry assessment,
It can be realized the automation of proposition.
To achieve the above object, the present invention is achieved by the following technical programs.
A kind of automatic proposition method towards classic poetry assessment, this method comprises: S1. determines topic type；S2. stich is selected
The text of word is as stem；Make correct option and candidate answers.
Preferably, the step S2 includes:
S21. if topic type is label multiple-choice question, stich word is selected, by poem mark corresponding in poem library
Label are used as correct option；
S22. it is chosen in vocabulary with the long identical word of correct option word as candidate word, calculates candidate word and correct option
Term vector cosine similarity, choose maximum three candidate words of cosine similarity as candidate answers.
Preferably, the generation method of the term vector are as follows:
Extensive news, encyclopaedia corpus are acquired, participle corpus is obtained after participle；
Utilize the neural network model of meaning of a word distribution hypothesis construction centre word prediction context；
Participle corpus is inputted into neural network model, the vector of all words indicates in output participle corpus.
Preferably, the step S2 includes:
S21. if topic type is system selection topic, the poem of all simple sentence sentences a length of 5 or 7 is chosen as stem；
S22. using the system of the poem as correct option, other systems are selected at random as candidate answers.
Preferably, the step S2 includes:
S21. if topic type is upper and lower sentence multiple-choice question, the upper sentence or lower sentence for selecting poem are as stem, then under corresponding to
Sentence or upper sentence are correct option；
S22. all verses in poem library are assessed；
S23. choosing assessment highest three poems of score is candidate answers.
Preferably, the step of assessment, includes:
Judge whether the verse in poem library is equal with the length of correct option and scores；Preferably, with correct option
Equal length meter 1 divides；
Judging the verse in poem library, whether present position is consistent with correct option present position in poem and scores；It is preferred that
, present position unanimously counts 1 point；
Judge whether the rhyme of the verse and correct option in poem library is consistent and scores；Preferably, rhyme unanimously counts 2
Point；
Calculate the repetitive rate in verse and the correct option in poem library with word and score；
Add up above-mentioned score, obtains the assessment score of the verse in poem library.
Preferably, the calculation formula of the repetitive rate with word are as follows:
R=NR/NC
Wherein NRFor number of characters shared in candidate verse and in correct option, NCFor total number of characters of candidate verse.
Preferably, the step S2 includes:
S21. if topic type is poem sequence topic, stich is selected, is used as stem after upsetting the sequence of sentence, and be language
Sentence label；
S22. the correct label sequence of statement sequence is correct option, generates other label sequences at random and answers as candidate
Case.
Preferably, the step S2 includes:
S21. if topic type is that verse word filling inscribes proposition, a poem is selected；
S22. the corresponding similar character list of word in verse is obtained；
S23. the word in verse is replaced with the similar character in similar character list, if there are similar characters in binary word list
The word group formed with the preceding word of the word in verse or rear word, then similar character is candidate word；
S24. if candidate number of words is more than or equal to 3, " _ _ " will use to replace the verse of word in verse as stem,
Using the word in verse as correct option, 3 candidate words are chosen as candidate answers.
Preferably, the step S22 are as follows:
The cosine similarity of the word vector of all words in the word vector and word table of the word in verse is calculated, it is small to exclude similarity
In the word of first threshold, similar character list is obtained.
Preferably, the step S24 further include:
If candidate number of words is less than 3, repeatedly step S22- step S23.
Preferably, the binary word list is that frequency of occurrence is greater than the column that the binary word of second threshold forms in poem library
Table.
Preferably, word vector training method are as follows:
Character segmentation is carried out to the poem in poem library；
Hypothesis, the neural network model of construction centre word prediction context are distributed using the meaning of a word；
The vector that training obtains all Chinese characters indicates that each word can be indicated by real vector.
It is a kind of towards classic poetry assessment automatic Examination Paper Generating System, the system include poem library, label multiple-choice question proposition module,
System selection inscribes proposition module, upper and lower sentence multiple-choice question proposition module, verse word filling topic proposition module, poem sequence proposition module,
Wherein:
Label multiple-choice question proposition module, for selecting stich, using its text as stem；By label corresponding to poem
As correct option；Obtain all candidate words identical with correct option word length in vocabulary；Calculate candidate word and correct option
The cosine similarity of term vector chooses maximum three words of similarity as candidate answers, to generate label multiple-choice question；
Poem system selection inscribes proposition module, for selecting all simple sentence sentences a length of 5 of any one head or 7 poems, just by it
Text is used as stem；Using the system of the poem as correct option；Other systems are selected at random as candidate answers, to generate poem
System selection topic；
Upper and lower sentence multiple-choice question proposition module, for selecting a poem, the upper sentence or lower sentence for selecting the poem are as stem, then
Corresponding lower sentence or upper sentence are correct option；All verses in poem library are assessed；Choose assessment score highest three
Poem is candidate answers, to generate sentence multiple-choice question up and down；
Poem sequence proposition module upsets sequence and is used as stem, and simple sentence mark for selecting four simple sentences of stich
Number；Label sequence according to the correct sequence composition of poem is correct option, generates the label sequence conduct different from correct option
Candidate answers, to generate poem sequence topic；
Verse word filling inscribes proposition module, for obtaining the corresponding similar character of word in verse for a given verse
List replaces the word in verse with the similar character in similar character list, if there are similar characters and verse in binary word list
In word preceding word or rear word composition word group, then similar character be candidate word；It, will if candidate number of words is more than or equal to 3
The verse of the word in " _ _ " replacement verse is used to choose 3 candidate word conducts using the word in verse as correct option as stem
Candidate answers, to generate verse word filling topic.
In the present invention, the poem type of writing and semantic feature can be farthest excavated, the ancient poetry of high quality is automatically generated
Word exam pool efficiently, scientifically realizes the assessment of classic poetry attainment.
Detailed description of the invention
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technical description to be briefly described, it should be apparent that, the accompanying drawings in the following description is the present invention
Some embodiments for those of ordinary skill in the art without creative efforts, can also basis
These attached drawings obtain other attached drawings.
Fig. 1 is label multiple-choice question proposition method flow diagram according to an embodiment of the invention；
Fig. 2 is system selection topic proposition method flow diagram according to an embodiment of the invention；
Fig. 3 is sentence multiple-choice question proposition method flow diagram up and down according to an embodiment of the invention；
Fig. 4 is gap-filling questions proposition method flow diagram according to an embodiment of the invention；
Fig. 5 is sequence topic proposition method flow diagram according to an embodiment of the invention；
Fig. 6 is the structural schematic diagram of proposition system according to an embodiment of the invention.
Specific embodiment
In order to make the object, technical scheme and advantages of the embodiment of the invention clearer, below in conjunction with the embodiment of the present invention
In attached drawing, technical solution in the embodiment of the present invention checked, completely described, it is clear that described embodiment is
A part of the embodiment of the present invention, instead of all the embodiments.Based on the embodiments of the present invention, those of ordinary skill in the art
Every other embodiment obtained without creative labor, shall fall within the protection scope of the present invention.
The present invention proposes a kind of automatic proposition method towards classic poetry, comprising steps of S1. determines topic type；S2. one is selected
The text of first poem is as stem；Make correct option and candidate answers.
Embodiment 1
As shown in Figure 1, stich word is selected, by the poem if determining topic type is label multiple-choice question in step sl
Word label corresponding in poem library is as correct option.
Poem can be it is artificial selected, can also be random or by being centainly intended to selection in poem library by computer.Label
The keyword for referring to stich can be the subject matter of stich, can be the emotion of stich expression, can be stich description
Things, it might even be possible to be stich relevant season, etc..
Later, it is chosen in vocabulary with the long identical word of correct option word as candidate word, calculates candidate word and correctly answer
The cosine similarity of the term vector of case chooses maximum three candidate words of cosine similarity as candidate answers.
Vocabulary is built by being segmented to the poem in poem library, and duplicate removal.
Term vector is obtained by training, training method are as follows: acquire extensive news, encyclopaedia corpus, carry out to it
Participle；Utilize the neural network model of meaning of a word distribution hypothesis (meaning of a word is determined by context) construction centre word prediction context；It will
It segments corpus and inputs neural network model, the vector of all words indicates in output participle corpus, it is preferred that each word can lead to
The real vector for crossing one 300 dimension is indicated.
Calculate the formula of cosine similarity are as follows:
Wherein n represents the dimension (taking 300 herein) of word (word) vector, xiIndicate numerical value of the correct item x in i-th dimension, yiIt indicates
Numerical value of word (word) y in i-th dimension.
How lower mask body introduction generates topic.For example, selected poem: " taking red for green, it is numerous and confused to think, wan and sallow fragmented to recall
Monarch.Do not believe that unpacking to test takes women's red-skirt than carrying out long lower tear."
First using its text as stem.Then, one of the label of the poem in poem library is emotion, and corresponding value is
" miss ", then the correct option of this poem label is " miss ".
Candidate answers are generated below.All (i.e. word is long is equal to 2) identical as the correct option word length is searched in vocabulary
Candidate word.As a result have " when cherishing ", " patriotic " etc..
According to the good term vector of pre-training, candidate word and correct option are converted into term vector.Calculate the word of candidate word
The cosine similarity of the term vector of vector and the correct option after descending ranking, takes ranking 1 to the word of ranking 3 ----" when cherishing, love
State, departure " is used as candidate answers.Therefore the topic ultimately generated is as follows:
Select the label of following poem: " taking red for green, it is numerous and confused to think, wan and sallow fragmented to recall monarch.Do not believe than carrying out long lower tear, unpacks
It tests and takes women's red-skirt."
A. the patriotic D. of C. is parted when miss B. cherishs
Embodiment 2
As shown in Fig. 2, it is long to choose first all simple sentence sentences if determining topic type is system selection topic in step sl
For 5 or 7 poem as stem.Using the system of the poem as correct option, other systems are selected at random as candidate
Answer.
The long chinese character quantity for referring to a verse in the present invention and including of sentence, does not include punctuate；Long simple sentence is then one
The chinese character quantity that the interior punctuate of connection or the simple sentence unit of space segmentation are included.
The following examples are specifically described.For example, " sense meets 12 for selection from the classic poetry of simple sentence sentence a length of 5 or 7
It is first " poem, using its text as stem.The system of " sense meet 12 first " is " five speech ancient poetry " in poem library, then system is being just
True answer is " five speech ancient poetry ".From system candidate item, (system candidate item includes five-character quatrain, poem with five characters in one line, five speech ancient poetry, five speeches
Music Bureau, seven-word poem, seven-character octave, seven-character "old style" verse, seven speech Music Bureau) in randomly choose other three kinds of poem systems, such as " five speech rule
Poem, seven-character octave, seven speech Music Bureau " are used as candidate answers.This 4 random alignments are generated into answer choice.Therefore it ultimately generates
Topic is as follows:
Select the system of following poem: " Lan Yechun is luxuriant, osmanthus China Qiu Jiaojie.This joyful business is the happy festival time from you.Who knows
The woods person of dwelling, hearsay seat are mutually pleased.Vegetation has conscience, what asks beauty to roll over? "
A. five speech ancient poetry B. poem with five characters in one line C. seven-word poem D. seven says Music Bureau
Embodiment 3
As shown in figure 3, if determining topic type is upper and lower sentence multiple-choice question in step sl, select poem upper sentence or
For lower sentence as stem, corresponding lower sentence or upper sentence are correct option；All verses in poem library are assessed later, selection is commented
Estimating highest three poems of score is candidate answers.
When assessment, judge whether each verse in poem library is equal with the length of correct option and scores；Preferably, with
The equal length meter 1 of correct option divides；
Judging each verse in poem library, whether present position is consistent with correct option present position in poem and scores；
Preferably, present position unanimously counts 1 point；
Judge whether the rhyme of each verse in poem library and correct option is consistent and scores；Preferably, rhyme is consistent
2 points of meter；
Calculate the repetitive rate in each verse and the correct option in poem library with word and score；The repetitive rate with word
Calculation formula are as follows: R=NR/NC
Wherein NRFor number of characters shared in each verse in poem library and in correct option, NCIt is every in poem library
Total number of characters of a verse.
Add up above-mentioned score, obtains the assessment score of the verse in poem library.
The following examples are specifically described.For example, poem of selection " flower splashes tear when sense, hates other bird soul-stirring ", selects " when sense
Flower splashes tear " be used as stem, then " hating other bird soul-stirring " is correct option.
All verses in poem library are assessed.Because of " hating other bird soul-stirring " simple sentence a length of 5, so all simple sentences are long not
Assessment score for 5 verse is 0, and the scoring of the verse of sentence a length of 5 is 1.The second of a poem is hoped in the spring because " hating other bird soul-stirring " is
Join the second simple sentence, so, all not remain unchanged in the assessment score of the verse of the second the second simple sentences, remaining is assessed originally
Add 1 on the basis of score.Rhythm portion of " heart " word of cause " hating other bird soul-stirring " in par rhythm is " invading ", so all end words do not exist
The assessment score for the verse for invading portion remains unchanged, remaining adds 2 on the basis of original score.Then calculate each verse with
" hating other bird conscientious " uses word repetitive rate, adds on the basis of the assessment score of each verse and uses word repetitive rate.Finally calculate
" there is the Wagtail Wagtail heart in former times, with boat without disloyalty, bird goes late in the unseen world " three highest scorings are obtained, as candidate answers.Therefore final production
Topic are as follows:
Select the lower sentence of following verse: " flower splashes tear when sense, _ _ _ _ _ _ _ _ _ _."
A. hating other bird soul-stirring B. former times to have Wagtail Wagtail heart C. with boat, bird goes late in the unseen world without disloyalty D.
Embodiment 4
As shown in figure 4, selecting stich if the topic type determined in step S1 is poem sequence topic, upsetting sentence
It is used as stem after sequence, and is each statement label；The correct label sequence of statement sequence is correct option, generates other at random
Label sequence is as candidate answers.
The following examples are specifically described.Hope Yue Yishi for example, given, selection " Mount Tai husband how the green unfinished good fortune in Shandong
The elegant yin-yang of clock mind cuts confused dawn " four simple sentences, they are upset after sequence as stem.It is such as " neat to upset the later simple sentence of sequence
How the unfinished yin-yang of Lu Qing cuts the dawn dusk elegant Mount Tai husband of good fortune clock mind " " green unfinished 1 yin-yang in Shandong cuts 2 good fortune of dawn dusk to flag sequence number
The elegant 3 Mount Tai husbands of clock mind how 4 ", according to former sentence, the correct option should be 4132.The all sequences of 1,2,3,4 compositions are found out, at random
Three sequences other than the correct option such as " 4312,4123,2413 " are selected to form distracter.Therefore the topic face ultimately generated are as follows:
The correct sequence of " green unfinished 1 yin-yang in Shandong cut the 2 elegant 3 Mount Tai husbands of good fortune clocks mind of dawn dusk how 4 " are as follows:
A.4132B.4312C.4123D.2413
Embodiment 5
As shown in figure 5, selecting a poem if the topic type determined in step S1 is that verse word filling inscribes proposition.Verse is filled out
Word topic is the verse of a given hiatus, and the word missed is selected in option.
The word traversed in verse in order stops traversing if the word has 3 or more candidate answers, which is made
For correct option.Topic is produced in this way.It is of course also possible to continue to traverse, multiple groups answer choice is obtained, then selects one again
Answer of the group as topic.
For a word, the cosine similarity of the word vector of all words in the word vector and word table of the word is calculated, excludes phase
It is less than the word of first threshold (such as 0.3) like degree, to obtain the similar character list of the word.Wherein, word table is by poem
Poem in library segmented, is built after duplicate removal.
Then, the word in verse is replaced with the similar character in similar character list, if there are phases in binary word list
Like the word group that the preceding word of the word in word and verse or rear word form, then similar character is candidate word；
Choose all candidate words, if candidate number of words is more than or equal to 3, the word in " _ _ " replacement verse will be used
Verse chooses 3 candidate words as candidate answers using the word in verse as correct option as stem.
If candidate number of words traverses next word, then repeats the above steps less than 3.
Binary word group refers to by two letters or two syllables, or two unitary combinations being made of two words, such as " two
Member ", " first word ", " word group ".Binary word list is that frequency of occurrence is greater than the list that the binary word of second threshold forms in poem library,
Preferred second threshold is 5.
Word vector is obtained by training, training method are as follows:
Character segmentation is carried out to the poem in poem library；
Hypothesis, the neural network model of construction centre word prediction context are distributed using the meaning of a word；
The vector that training obtains all Chinese characters indicates that each word can be indicated by real vector, preferably 300
Dimension.
Concrete example explanation below.For example, given " the luxuriant osmanthus China Qiu Jiaojie of Lan Yechun " one, it is assumed that carried out to " leaf " word
Aforesaid operations." leaf " word is first converted into word vector, calculates the cosine of the word vector of all words in the word vector and word table of " leaf "
Similarity, selection are greater than 0.3 all words with " leaf " word cosine similarity；By these words by the sequence of similarity from high to low according to
Secondary replacement " leaf ", and checking whether can be with the preceding word of leaf or rear word composition binary phrase.When such as with " flower " replacement " leaf ", discovery
" orchid " appears in binary word Groups List of the frequency greater than 5, then " flower " is used as one of candidate answers；It is replaced with " face "
When " leaf ", it is found that " Lan Yan " and " Yan Chun " not in binary word Groups List, then excludes " face " word, similarly retaining " tree, branch " is
Candidate answers.Candidate answers quantity reaches 3, stops calculating.Therefore the topic ultimately generated is as follows:
The word that the selection place of filling a vacancy should insert: " the orchid _ _ spring is luxuriant, osmanthus China Qiu Jiaojie."
A. D., C. tree of leaf B. flower
Embodiment 6
Poem knowledge mapping is exactly a kind of poem library, it includes more than the 240000 first classic poetry data from the pre-Qin days to Ming and Qing, is covered
More than lid 20 relationship, every head poem have author, good friend, date of birth, poem dynasty, term, theme, emotion, style, place
Equal labels.
The poem knowledge frame that poem knowledge mapping can use natural language processing technique and ancient literature expert formulates
Large-scale knowledge excavation, extraction and the presentation that classic poetry theme, emotion, style etc. are realized.Specific generation step is such as
Under:
1, for first poem each in poem database, title, author and content information are obtained.
2, establish ancient poetry system include six speeches, seven Gus, four-line poem with seven characters to a line, seven-syllable, seven rows, the poetry in the style of Li Sao, four speeches, pentasyllabic ancient-style poetry, pentasyllabic quatrain, an eight-line poem with five characters,
Five rows, miscellaneous totally ten two kinds of Gu.
3, establishing ancient poetry label includes subject matter, function, style, emotion and annalistic six class, wherein subject matter include love and marriage,
Frontier fortress's war, row labour of staying long in a strange place, dinner party of making friends, folklore, scenery with hills and waters rural area, current events are political, the singing of history is recalled antiquity, nine class of a poem of mystical excursion, wind
Lattice include energetic, quiet, wealthy remote, dreary, pure and fresh, soft and mild six classes, and emotion includes that compassion is indignant, depressed, respects, is impassioned, deserted and lonely, spacious
It reaches, is straightforward, miss, banter, is leisurely and comfortable, pleasant ten one kind.
4, people tag, the good friend that such as makes friends, the frequency of making friends, poem of making friends are established.
5, to the above attribute, carry out artificial+machine and combine mark, wherein system information, people tag progress machine are automatic
It extracts, poem label information is extracted using artificial mark+machine learning mode.
6, for extraction as a result, ancient poetry knowledge mapping is built, with triple form storing data, such as " Jiang Xue-
(relation: author)-LIU Zong-yuan ", " Jiang Xue-(relation: system)-pentasyllabic quatrain ", " Jiang Xue-(relation: style)-quiet
It is quiet " etc..
1230 first classic poetries are chosen from above-mentioned poem library, using the method in embodiment 1-5, are carried out examination question and are generated behaviour
Make, symbiosis is at 7631 examination questions.Receive the information such as examination question quantity and the topic type ratio that user determines, satisfaction can be provided and accordingly wanted
One or more sets objective item papers asked.
Embodiment 7
The embodiment of the present invention 7 also proposes a kind of automatic Examination Paper Generating System towards classic poetry assessment, as shown in fig. 6, the system
It is filled out including poem library, label multiple-choice question proposition module, system selection topic proposition module, upper and lower sentence multiple-choice question proposition module, verse
Proposition module that word inscribes proposition module, poem sorts, in which:
Label multiple-choice question proposition module selectes stich, using its text as stem for generating label multiple-choice question；It will
Label corresponding to poem is as correct option；Obtain all candidate words identical with correct option word length in vocabulary；It calculates and waits
The cosine similarity of the term vector of word and correct option is selected, chooses maximum three words of similarity as candidate answers.Vocabulary
Generation method is as described above.
Poem system selection inscribes proposition module, for generating system selection topic, selectes all simple sentence sentences a length of 5 of any one head
Or 7 poems, using its text as stem；Using the system of the poem as correct option；Other systems are selected at random to answer as candidate
Case；Specific topic generation method is as described above.
Upper and lower sentence multiple-choice question proposition module selects a poem for generating sentence multiple-choice question up and down, select the poem upper sentence or
Sentence is as stem under person, then corresponding lower sentence or upper sentence are correct option；All verses in poem library are assessed；Selection is commented
Estimating highest three poems of score is candidate answers；Specific appraisal procedure is as described above.
Poem sequence proposition module selects four simple sentences of stich, upsets sequence and make for generating sentence multiple-choice question up and down
For stem, and simple sentence label；Label sequence according to the correct sequence composition of poem is correct option, is generated different from correct option
Label sequence as candidate answers；
Verse word filling inscribes proposition module, for generating word filling topic, for a given verse, obtains the word pair in verse
The similar character list answered replaces the word in verse with the similar character in similar character list, if there are phases in binary word list
Like the word group that the preceding word of the word in word and verse or rear word form, then similar character is candidate word；If candidate number of words is greater than etc.
In 3, then the verse of the word in " _ _ " replacement verse will be used as stem, using the word in verse as correct option, selection 3
Candidate word is as candidate answers.Word table, the generation method of binary word list are as described above.
Poem library used in above-mentioned knowledge mapping can be used in poem library, also can according to need and voluntarily acquires, organizes.
The above examples are only used to illustrate the technical scheme of the present invention, rather than its limitations；Although with reference to the foregoing embodiments
Invention is explained in detail, those skilled in the art should understand that: it still can be to aforementioned each implementation
Technical solution documented by example is modified or equivalent replacement of some of the technical features；And these are modified or replace
It changes, departs from the spirit and scope of the technical scheme of various embodiments of the present invention the essence of corresponding technical solution.
Claims (14)
1. a kind of automatic proposition method towards classic poetry assessment, which is characterized in that this method comprises:
S1. topic type is determined；
S2. the text of stich word is selected as stem；Make correct option and candidate answers.
2. the method according to claim 1, wherein the step S2 includes:
S21. if topic type is label multiple-choice question, stich word is selected, poem label corresponding in poem library is made
For correct option；
S22. it is chosen in vocabulary with the long identical word of correct option word as candidate word, calculates the word of candidate word and correct option
The cosine similarity of vector chooses maximum three candidate words of cosine similarity as candidate answers.
3. the method according to claim 1, wherein the generation method of the term vector are as follows:
Extensive news, encyclopaedia corpus are acquired, participle corpus is obtained after participle；
Utilize the neural network model of meaning of a word distribution hypothesis construction centre word prediction context；
Participle corpus is inputted into neural network model, the vector of all words indicates in output participle corpus.
4. the method according to claim 1, wherein the step S2 includes:
S21. if topic type is system selection topic, the poem of all simple sentence sentences a length of 5 or 7 is chosen as stem；
S22. using the system of the poem as correct option, other systems are selected at random as candidate answers.
5. the method according to claim 1, wherein the step S2 includes:
S21. if topic type is upper and lower sentence multiple-choice question, the upper sentence or lower sentence for selecting poem as stem, then corresponding time sentence or
Upper sentence is correct option；
S22. all verses in poem library are assessed；
S23. choosing assessment highest three poems of score is candidate answers.
6. according to the method described in claim 5, it is characterized in that, the step of assessment include:
Judge whether the verse in poem library is equal with the length of correct option and scores；Preferably, with the length of correct option
Equal 1 point of meter；
Judging the verse in poem library, whether present position is consistent with correct option present position in poem and scores；Preferably,
Unanimously count 1 point in present position；
Judge whether the rhyme of the verse and correct option in poem library is consistent and scores；Preferably, rhyme unanimously counts 2 points；
Calculate the repetitive rate in verse and the correct option in poem library with word and score；
Add up above-mentioned score, obtains the assessment score of the verse in poem library.
7. according to the method described in claim 6, it is characterized in that, the calculation formula of the repetitive rate with word are as follows:
R=NR/NC
Wherein NRFor number of characters shared in candidate verse and in correct option, NCFor total number of characters of candidate verse.
8. the method according to claim 1, wherein the step S2 includes:
S21. if topic type is poem sequence topic, stich is selected, is used as stem after upsetting the sequence of sentence, and be sentence mark
Number；
S22. the correct label sequence of statement sequence is correct option, generates other label sequences at random as candidate answers.
9. proposition method according to claim 1, which is characterized in that the step S2 includes:
S21. if topic type is that verse word filling inscribes proposition, a poem is selected；
S22. the corresponding similar character list of word in verse is obtained；
S23. the word in verse is replaced with the similar character in similar character list, if there are similar characters and poem in binary word list
The word group of the preceding word of word in sentence or rear word composition, then similar character is candidate word；
S24. if candidate number of words is more than or equal to 3, " _ _ " will use to replace the verse of word in verse as stem, by poem
Word in sentence chooses 3 candidate words as candidate answers as correct option.
10. according to the method described in claim 9, it is characterized in that, the step S22 are as follows:
The cosine similarity of the word vector of the word in verse and the word vector of all words in word table is calculated, excludes similarity less than the
The word of one threshold value obtains similar character list.
11. according to the method described in claim 9, it is characterized in that, the step S24 further include:
If candidate number of words is less than 3, repeatedly step S22- step S23.
12. according to the method described in claim 9, it is characterized in that, the binary word list is that frequency of occurrence is big in poem library
In the list that the binary word of second threshold forms.
13. proposition method according to claim 10, wherein word vector training method are as follows:
Character segmentation is carried out to the poem in poem library；
Hypothesis, the neural network model of construction centre word prediction context are distributed using the meaning of a word；
The vector that training obtains all Chinese characters indicates that each word can be indicated by real vector.
14. a kind of automatic Examination Paper Generating System towards classic poetry assessment, which is characterized in that including poem library, label multiple-choice question proposition
Module, system selection topic proposition module, upper and lower sentence multiple-choice question proposition module, verse word filling topic proposition module, poem sequence proposition
Module, wherein
Label multiple-choice question proposition module, for selecting stich, using its text as stem；Using label corresponding to poem as
Correct option；Obtain all candidate words identical with correct option word length in vocabulary；Calculate the word of candidate word and correct option to
The cosine similarity of amount chooses maximum three words of similarity as candidate answers, to generate label multiple-choice question；
Poem system selection topic proposition module makees its text for selecting all simple sentence sentences a length of 5 of any one head or 7 poems
For stem；Using the system of the poem as correct option；Other systems are selected at random as candidate answers, to generate poem system
Multiple-choice question；
Upper and lower sentence multiple-choice question proposition module, for selecting a poem, the upper sentence or lower sentence for selecting the poem are then corresponded to as stem
Lower sentence or upper sentence are correct option；All verses in poem library are assessed；Choose assessment highest three poems of score
For candidate answers, to generate sentence multiple-choice question up and down；
Poem sequence proposition module upsets sequence and is used as stem, and simple sentence label for selecting four simple sentences of stich；It presses
Label sequence according to the correct sequence composition of poem is correct option, generates the label sequence different from correct option and answers as candidate
Case, to generate poem sequence topic；
Verse word filling inscribes proposition module, for obtaining the corresponding similar character list of word in verse for a given verse,
The word in verse is replaced with the similar character in similar character list, if there are the words in similar character and verse in binary word list
Preceding word or rear word composition word group, then similar character be candidate word；If candidate number of words is more than or equal to 3, will use " _ _ "
The verse of word in replacement verse is chosen 3 candidate words and is answered as candidate as stem using the word in verse as correct option
Case, to generate verse word filling topic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910490907.3A CN110232180B (en) | 2019-06-06 | 2019-06-06 | Automatic proposition method and system for ancient poetry evaluation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910490907.3A CN110232180B (en) | 2019-06-06 | 2019-06-06 | Automatic proposition method and system for ancient poetry evaluation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110232180A true CN110232180A (en) | 2019-09-13 |
| CN110232180B CN110232180B (en) | 2020-11-03 |
Family
ID=67858502
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910490907.3A Active CN110232180B (en) | 2019-06-06 | 2019-06-06 | Automatic proposition method and system for ancient poetry evaluation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110232180B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111783428A (en) * | 2020-07-07 | 2020-10-16 | 杭州叙简科技股份有限公司 | Emergency management type objective question automatic generation system based on deep learning |
| CN112101040A (en) * | 2020-08-20 | 2020-12-18 | 淮阴工学院 | Ancient poetry semantic retrieval method based on knowledge graph |
| CN112256829A (en) * | 2020-10-21 | 2021-01-22 | 东南大学 | Ancient poem generation method capable of controlling emotion and style |
| CN113239689A (en) * | 2021-07-07 | 2021-08-10 | 北京语言大学 | Selection question interference item automatic generation method and device for confusing word investigation |
| CN113254591A (en) * | 2021-06-16 | 2021-08-13 | 北京世纪好未来教育科技有限公司 | Topic generation method, device, equipment and storage medium |
| CN113361238A (en) * | 2021-05-21 | 2021-09-07 | 北京语言大学 | Method and device for automatically proposing question by recombining question types with language blocks |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100138217A1 (en) * | 2008-11-28 | 2010-06-03 | Institute For Information Industry | Method for constructing chinese dictionary and apparatus and storage media using the same |
| CN102385596A (en) * | 2010-09-03 | 2012-03-21 | 腾讯科技（深圳）有限公司 | Verse searching method and device |
| CN108073571A (en) * | 2018-01-12 | 2018-05-25 | 中译语通科技股份有限公司 | A kind of multi-language text method for evaluating quality and system, intelligent text processing system |
| CN109784165A (en) * | 2018-12-12 | 2019-05-21 | 平安科技（深圳）有限公司 | Generation method, device, terminal and the storage medium of poem content |
-
2019
- 2019-06-06 CN CN201910490907.3A patent/CN110232180B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100138217A1 (en) * | 2008-11-28 | 2010-06-03 | Institute For Information Industry | Method for constructing chinese dictionary and apparatus and storage media using the same |
| CN102385596A (en) * | 2010-09-03 | 2012-03-21 | 腾讯科技（深圳）有限公司 | Verse searching method and device |
| CN108073571A (en) * | 2018-01-12 | 2018-05-25 | 中译语通科技股份有限公司 | A kind of multi-language text method for evaluating quality and system, intelligent text processing system |
| CN109784165A (en) * | 2018-12-12 | 2019-05-21 | 平安科技（深圳）有限公司 | Generation method, device, terminal and the storage medium of poem content |
Non-Patent Citations (4)
| Title |
|---|
| SHUO LUO: ""诗词大会II-分级挑战"", 《诗词大会II》 * |
| 来斯惟: ""基于神经网络的词和文档语义向量表示方法研究"", 《中国科学院大学博士学位论文》 * |
| 组卷网: ""诗歌鉴赏"", 《HTTP://ZUJUAN.XKW.COM/10Q8037740.HTML》 * |
| 胡韧奋等: ""汉语词汇测试自动命题研究"", 《中文信息学报》 * |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111783428A (en) * | 2020-07-07 | 2020-10-16 | 杭州叙简科技股份有限公司 | Emergency management type objective question automatic generation system based on deep learning |
| CN111783428B (en) * | 2020-07-07 | 2024-01-23 | 杭州叙简科技股份有限公司 | Emergency management objective question automatic generation system based on deep learning |
| CN112101040A (en) * | 2020-08-20 | 2020-12-18 | 淮阴工学院 | Ancient poetry semantic retrieval method based on knowledge graph |
| CN112101040B (en) * | 2020-08-20 | 2024-03-29 | 淮阴工学院 | Ancient poetry semantic retrieval method based on knowledge graph |
| CN112256829A (en) * | 2020-10-21 | 2021-01-22 | 东南大学 | Ancient poem generation method capable of controlling emotion and style |
| CN113361238A (en) * | 2021-05-21 | 2021-09-07 | 北京语言大学 | Method and device for automatically proposing question by recombining question types with language blocks |
| CN113361238B (en) * | 2021-05-21 | 2022-02-11 | 北京语言大学 | Method and device for automatically proposing question by recombining question types with language blocks |
| CN113254591A (en) * | 2021-06-16 | 2021-08-13 | 北京世纪好未来教育科技有限公司 | Topic generation method, device, equipment and storage medium |
| CN113239689A (en) * | 2021-07-07 | 2021-08-10 | 北京语言大学 | Selection question interference item automatic generation method and device for confusing word investigation |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110232180B (en) | 2020-11-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110232180A (en) | A kind of automatic proposition method and system towards classic poetry assessment | |
| Rossi-Landi | Linguistics and economics | |
| CN104391980B (en) | The method and apparatus for generating song | |
| Tarlinskaja | Shakespeare and the versification of English drama, 1561-1642 | |
| Carr | The formation of the Hebrew Bible: A new reconstruction | |
| Mair et al. | The Sanskrit origins of recent style prosody | |
| Bostoen et al. | Linguistic innovation, political centralization and economic integration in the Kongo kingdom: Reconstructing the spread of prefix reduction | |
| Rusk | Critics and Commentators: The Book of Poems as Classic and Literature | |
| CN112183109B (en) | MASS-based poetry sentence generation information steganography method | |
| Kohler | Communicative Functions and Linguistic Forms in Speech Interaction: Volume 156 | |
| Salvatore | The Islamicate Adab Tradition vs. the Islamic Shari ‘a, from Pre-Colonial to Colonial | |
| CN113268581B (en) | Topic generation method and device | |
| Fusi | A multilanguage, modular framework for metrical analysis: IT patterns and theorical issues | |
| Erofeeva et al. | The Study Of Ancient Chronicles Within The Professional Philologist Educator’s Training System | |
| Duran | Non-standard forms of Swahili in west-central Kenya | |
| Giger et al. | Music in the Mirror: Reflections on the History of Music Theory and Literature for the 21st Century | |
| Beauchamp | The style of Nerval’s “Aurélia” | |
| CN112328095B (en) | Four-purpose phonetic and shape code Chinese character input method and input platform without using number keys | |
| Meyer | Generating Free Verse Poetry with Transformer Networks | |
| Herdan | The structuralistic approach to Chinese grammar and vocabulary: two essays | |
| Kohler | The transmission of meaning by prosodic phrasing | |
| Ismahani et al. | Analyzing Sentence Structures in English Idioms: a Simplified Syntactic Analysis with Tree Diagrams | |
| Baayen et al. | Max-Planck-Institute for Psycholinguistics: Annual Report Nr. 14 1993 | |
| Frye | Myth as the Matrix of Literature | |
| RAMİL | Literature and Sociology: A Siamese Twins’ Journey from Entwined to Estranged |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |