CN108536299A - Human-computer interaction result generation method based on matrix and system - Google Patents
Human-computer interaction result generation method based on matrix and system Download PDFInfo
- Publication number
- CN108536299A CN108536299A CN201810326047.5A CN201810326047A CN108536299A CN 108536299 A CN108536299 A CN 108536299A CN 201810326047 A CN201810326047 A CN 201810326047A CN 108536299 A CN108536299 A CN 108536299A
- Authority
- CN
- China
- Prior art keywords
- answer
- coefficient
- result
- matrix
- candidate
- 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
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
Abstract
The present invention relates to a kind of human-computer interaction result generation method and system based on matrix, matrix by being stored with answer records the distribution relation of problem and result, clearly reflect the dependence between each problem and each result, a large amount of computational space is occupied without carrying out Model Reconstruction, operation efficiency is greatly speeded up, while also improving the accuracy of finally obtained result.And it was proved that, the present invention goes to carry out human-computer interaction using 8422 results and 1860 problems, and can result user be returned in 0.1 second.
Description
Technical field
The present invention relates to human-computer interaction technique fields, more particularly to a kind of human-computer interaction result generation side based on matrix
Method and system.
Background technology
It is a process output and input on human-computer interaction process nature, people is referred to by man-machine interface to computer input
It enables, computer is after treatment presented to the user output result.The form output and input between people and computer is a variety of
Various, therefore interactive form is also diversified.One of which human-computer interaction mode is：By machine constantly to user into
Row is putd question to, and the answer each time according to user to enquirement, and finally determines the answer thought in the minds of user, to realization pair
The reading of user's heart answer, it can be understood as be one kind " thought reading " function of machine；Thereby increase the interest of human-computer interaction
Taste.
Currently, being usually all the man-machine friendship of reading core type realized using decision-tree model between machine and user in the prior art
Mutually so that machine can obtain the answer in the minds of user according to the answer of user.And decision-tree model be divided into dynamic modeling and
Static modelling, wherein dynamic modeling refers to that model creates completion in reading heart interactive process, that is, machine is proposed to user
The selection of each problem depend on the problem of user proposes the answer of problem thereon namely machine only in the problem
It can just be carried out when being required.The advantages of dynamic modeling is：It initiatively goes to create model, different answers of the user to same problem
It can all be derived with the different processes of problem proposition different as a result, it is possible to ensure the diversity of form.And static modelling refers to
To be model created completions before reading heart interaction, that is, before read the heart and interacting, created a series of
For model for machine choice and use, the answer that machine system can be according to user to problem constantly adjusts applicable models, multiple
It is redirected among model and finds next most appropriate problem, to continue to put question to user.The advantages of static modelling is：Due to being pre-
It first generates as a result, event can be so that the practical calculation amount read in heart interactive process be little, the arithmetic speed of machine system is fast.But
Realize that reading the heart interacts in the way of dynamic modeling and static modelling, however it remains following very important defect：
1) for dynamic modeling, since machine system according to the answer each time of user when generating corresponding problem, all
Data can once be calculated, when concurrent access amount is bigger, this operation consumption will become ignore, and finally lead
The operational efficiency of cause system substantially reduces；Simultaneously because needing to build corresponding decision tree to each user, can cause in machine
Deposit the steep increasing of occupancy.
2) for static modelling, since between starting service, all results all have determined, then heart interaction is read
The diversity of process is subject to limitation, and the accuracy of the result eventually led to is not high, is easy limited.
Also, both dynamic modeling and static modelling realize that the dependence of contextual issues all needs artificially to mark problem
Between relationship, when data volume is big, both realize read hearts interaction mode all do not have feasibility, can not reasonably realize
Problem duplicate removal and the problem of solving the problems, such as dependence.
Invention content
To solve the disadvantage that the above-mentioned prior art and deficiency, a present invention wherein purpose are to provide a kind of people based on matrix
Machine interaction result generation method, by the distribution relation based on matrix problem of implementation and answer, according to what is described to each answer
Distribution reflects the dependence between each problem, to improve the accuracy of finally obtained result, and need not carry out model weight
It builds and occupies a large amount of computational space, be conducive to greatly improve operation efficiency.It is a further object of the present invention to provide one kind to include
The human-computer interaction result based on matrix of the above-mentioned human-computer interaction result generation method based on matrix generates system.
The first purpose to realize the present invention, present invention firstly provides a kind of, and the human-computer interaction result based on matrix generates
Method comprising following steps：
S1：It has detected whether that question and answer open command inputs, has been to then follow the steps S2, otherwise keeps standby mode；
S2：The problem of initialization currently has proposed sum N；
S3：The problem of calling prestores set, results set and answer matrix, obtain a problem and are sent out from problem set
It puts question to；Wherein, described problem set is made of n different problems, and the results set is made of m different results；Institute
It is m * n matrix to state answer matrix, and the elements A ij in answer matrix is answer coefficient of i-th of result for j-th of problem；
Alternatively, the answer matrix is n × m matrixes, and the answering for i-th problem that be j-th result of the elements A ij in answer matrix
Case coefficient；
S4：User is obtained to the answer information currently putd question to, and information and answer matrix are answered according to current, in result set
Qualified candidate result is obtained in conjunction, generates candidate result set；
S5：The problem of are carried out by assignment operation, is currently had proposed for Issue Totals sum N=N+1；And judge currently
Whether sum of asking a question is greater than or equal to preset problematic amount threshold value Nmax, is to then follow the steps S9；Otherwise, step is executed
S6；
S6：According to candidate result set and answer matrix, a problem is obtained from problem set and continues to send out enquirement；
S7：User is obtained to the answer information currently putd question to, information and answer matrix are answered according to current, in candidate result
Qualified candidate result is obtained in set, updates candidate result set；
S8：Whether the number of the candidate result in candidate result set in judgment step S7 is less than or equal to fruiting quantities
Threshold value T is to then follow the steps S10, otherwise return to step S5；
S9：According to all answer information and answer matrix of user, each candidate knot of candidate result set is calculated
Fruit is to the matching factor asked a question；All matching factors are ranked up according to descending sequence, and obtain matching system
Numerical digit generates new candidate result set in first T of T candidate result；
S10：A candidate result is chosen from candidate result set, and is exported.
The present invention records the distribution relation of problem and result by being stored with the matrix of answer as a result, clearly reflects
Dependence between each problem and each result occupies a large amount of computational space, greatly greatly without carrying out Model Reconstruction
Fast operation efficiency, while also improving the accuracy of finally obtained result.It was proved that the present invention utilizes 8422 results
It goes to carry out human-computer interaction with 1860 problems, result user can be returned in 0.1 second.
Further, the value of answer coefficient is -1 or 0 or 1；When the value of answer coefficient is -1, indicate that answer coefficient is corresponding
As a result it is negative acknowledge for the answer of corresponding problem；When the value of answer coefficient is 0, the corresponding result of answer coefficient is indicated
Answer for corresponding problem is uncertain answers；When the value of answer coefficient is 1, the corresponding result of answer coefficient is indicated
Answer for corresponding problem is affirmative acknowledgement (ACK).By limiting herein, different answer information are indicated using -1,0 and 1, into one
Step simplifies the record to answer information, also reduces the occupied memory headroom of answer matrix, is conducive to the further of operation efficiency
It improves.
Further, in the step S3, while calling prestores problem set, the initial entropy set that prestores also is called；
The establishment step of the initial entropy set includes：According to the results set and answer matrix to prestore, asking of prestoring is calculated
Entropy of each problem to the comentropy of results set of set is inscribed, and records the correspondence of each problem and its entropy；To own
Entropy it is ascending or it is descending be arranged in order, form initial entropy set；
And in the step S3, the problem of being obtained from problem set and sending out enquirement, is for one with maximum entropy
Problem.
By limiting herein, be conducive to the screening degree for improving the problem of choosing for the first time to result, that is, by puing question to for the first time
Be with maximum entropy the problem of, so that it may to improve to the influence power of distribution of results, for example, there are two problem, 100 knots
100 results can be categorized into 80 results for having negative acknowledge to the first problem and affirmed back by fruit, first problem
20 answered are as a result, and 100 results can be categorized into have negative acknowledge to the Second Problem 60 by Second Problem
As a result have more as a result, then it is found that the classification of 100 results of first problem pair has more influence power with 40 of affirmative acknowledgement (ACK)
Big discrimination namely first problem can more rapidly realize the screening to result, the information entropy with bigger, therefore
First problem can be chosen preferentially to put question to.It is possible thereby to further increase the formation efficiency of result.
Further, the step S4 includes the following steps：
User is obtained to the answer information currently putd question to, and coefficient is answered according to answering information processing and obtaining user；
Judge that user answers whether coefficient is nonzero value, corresponds to current problem if it is, being selected in answer matrix
And the qualified answer coefficient consistent with user's answer coefficient, and selected in answer matrix and correspond to current problem and answer coefficient
For 0 answer coefficient undetermined；Then it is obtained in results set corresponding with each qualified answer coefficient and each answer coefficient undetermined
Candidate result generates candidate result set；If it is not, then obtaining all as a result, generating candidate result collection in results set
It closes.
By being further limited to step S4, be conducive to the screening for further improving candidate result so that in database
Preserve to certain problems do not have answer or answer it is uncertain or do not preserve some results of corresponding answer temporarily can be first
It remains, avoids screening out without exception and causing to ignore that may be present accurately as a result, being conducive to further increase result generation
Accuracy rate.
Further, the step S6 includes the following steps：
According to candidate result set and answer matrix, letter of each problem to candidate result set of problem set is calculated
The entropy of entropy is ceased, and records the correspondence of each problem and its entropy；
By each problem of problem set it is ascending to all entropy of candidate result set or it is descending be arranged in order,
Form candidate entropy set；
The problem corresponding with maximum entropy in candidate entropy set from selection in problem set, and continue to send out enquirement.
The problem of by further limiting to step S6, being conducive to improve each proposition to the influence power of distribution of results,
It can more rapidly realize the screening to result, the formation efficiency of result can be further increased.
Further, the step S7 includes the following steps：
User is obtained to the answer information currently putd question to, and coefficient is answered according to answering information processing and obtaining user；
Judge that user answers whether coefficient is nonzero value, corresponds to current problem if it is, being selected in answer matrix
And the qualified answer coefficient consistent with user's answer coefficient, and selected in answer matrix and correspond to current problem and answer coefficient
For 0 answer coefficient undetermined；Then it is obtained in candidate result set and each qualified answer coefficient and each answer coefficient pair undetermined
The candidate result answered generates candidate result set；If it is not, then not updating candidate result set.
By being further limited to step S7, be conducive to the screening for further improving candidate result so that in database
Preserve to certain problems do not have answer or answer it is uncertain or do not preserve some results of corresponding answer temporarily can be first
It remains, avoids screening out without exception and causing to ignore that may be present accurately as a result, being conducive to further increase result generation
Accuracy rate.
Further, the step S9 includes the following steps：
According to the enquirement sequence asked a question, answers coefficient to all users and be ranked up, generate user's coefficient sets U, U
={ U1..., UI..., UNmax, wherein UII-th user answer coefficient in expression user's coefficient sets, 1≤I≤Nmax, and
I is integer, and Nmax is problematic amount threshold value；And according to the enquirement sequence asked a question, own respectively to each candidate result
Qualified answer coefficient and all undetermined coefficients are ranked up, and generate each candidate result respectively has selected coefficient sets Ia, Ia=
{Ia1..., IaI..., IaNmax, wherein a indicates a-th of candidate result in candidate result set, IaIndicate candidate result collection
A-th of candidate result in conjunction has selected coefficient sets, IaICoefficient sets I has been selected in expressionaIn i-th answer coefficient, 1≤I
≤ Nmax, I are integer, and Nmax is problematic amount threshold value；And user in user's coefficient sets answer coefficient respectively with it is each
The answer coefficient in coefficient sets is selected to map one by one；
According to user coefficient sets U, calculating has respectively selected coefficient sets IaIn with mapping relations UIValue it is identical
IaINumber, respectively obtain the matching factor of each candidate result；
All matching factors are ranked up according to descending sequence, and obtains matching factor and is located at preceding T of T
Candidate result generates new candidate result set.
By being further limited to step S9, realizes when the quantity of remaining candidate result is excessive, tied using candidate
The answer coefficient of fruit and user to the consistent quantity of the answer coefficient of the enquirement proposed number, to select finally to need
It is choosing as a result, to ensure that the properer user of result chosen thinks to match as a result, also simplifying result and user answer
The calculating process of degree further increases the formation efficiency of result；And by alternative T candidate result, be conducive to the one of output
When a result inaccuracy, it is also an option that other results export, the alternative and accuracy of result output are improved.
Further, the step S10 includes step：Judgement jumps to whether the previous step of this step is step S8, is
A candidate result is then randomly selected from candidate result set, and is exported；Otherwise, being chosen from candidate result set has maximum
One candidate result of matching factor, and export.
By being further limited to step S10, be conducive to the delivery efficiency for further increasing correct result.
Further, in the step S5, problematic amount threshold value Nmax is 10~20；In the step S8, fruiting quantities threshold
Value T is 3.It by the restriction to problematic amount threshold value, avoids when not obtaining required amount of result also, because puing question to number excessive
And the situation for causing user to answer fatigue occurs, to further increase user experience and game excitement.And by fruiting quantities
The restriction of threshold value is realized when result needs to be investigated, and it is maximum several as a result, wrong in a result output to retain possibility
Afterwards, can also export another as a result, on the one hand can be preferably bonded it is that user is thought as a result, also enhance on the one hand with
Interaction between user.
Another object to realize the present invention, the present invention also provides can realize the above-mentioned human-computer interaction knot based on matrix
A kind of human-computer interaction result based on matrix of fruit generation method generates system, which includes readable storage medium storing program for executing, processor
And output equipment；The readable storage medium storing program for executing is stored with some instructions；The processor is loaded and is held according to some instructions
The row human-computer interaction result generation method described in any one of the above embodiments based on matrix；The output equipment output is handled by processor
Obtained result.Since the present invention is based on the human-computer interaction result of matrix generation systems can realize the people the present invention is based on matrix
Machine interaction result generation method, therefore the present invention is based on the human-computer interaction results of matrix to generate system also with the present invention is based on matrixes
Human-computer interaction result generation method caused by advantageous effects, details are not described herein.
In order to better understand and implement, the invention will now be described in detail with reference to the accompanying drawings.
Description of the drawings
Fig. 1 is that the present invention is based on the method flow diagrams of the human-computer interaction result generation method of matrix.
Specific implementation mode
For solve to utilize in the prior art dynamic modeling and static modelling in the way of realize read the heart interact it is existing scarce
Fall into --- realize that the dependence of contextual issues all needs the relationship between artificially mark problem not have when data volume is big
Feasibility, can not reasonably problem of implementation duplicate removal and the dependence between solving the problems, such as, and operational efficiency is low, accuracy is not high
Problem；The present invention has developed a kind of human-computer interaction result generation method and system based on matrix, and specific technical solution is such as
Shown in lower：
Referring to Fig. 1, the human-computer interaction result generation method the present invention is based on matrix includes the following steps：
S1：It has detected whether that question and answer open command inputs, has been to then follow the steps S2, otherwise keeps standby mode；That is, such as
Fruit embodies the function of the step using instantiated structure, it is assumed that assigned for processing equipment can realize the present invention based on
The function of the human-computer interaction result generation method of matrix, then at this point, an input interface of processing equipment, which is instruction, monitors interface,
The monitoring to instruction can be realized by input equipment by monitoring the input equipment that interface signal is connect with the instruction --- when
When user is by input equipment input signal, whether processor can be question and answer open command according to input-signal judging, if so,
S2 is thened follow the steps, if it is not, then keeping standby mode；
S2：The problem of initialization currently has proposed sum N；In step S2, the problem sum N=0 after initialization；
S3：The problem of calling prestores set, results set and answer matrix, obtain a problem and are sent out from problem set
It puts question to；
Wherein, described problem set is made of n different problems, and the results set is made of m different results；
The answer matrix is m * n matrix, and the elements A ij in answer matrix is answer system of i-th of result for j-th of problem
Number；Alternatively, the answer matrix be n × m matrixes, and the elements A ij in answer matrix be j-th of result for i-th of problem
Answer coefficient；In the present embodiment, preferably to reduce the occupied memory headroom of answer matrix, operation effect is further increased
The value of rate, answer coefficient is -1 or 0 or 1；When the value of answer coefficient is -1, indicate the corresponding result of answer coefficient for corresponding
The problem of answer be negative acknowledge,；When the value of answer coefficient is 0, indicate the corresponding result of answer coefficient for corresponding
The answer of problem is uncertain answers；When the value of answer coefficient is 1, indicate the corresponding result of answer coefficient for corresponding
The answer of problem is affirmative acknowledgement (ACK)；Wherein, negative acknowledge refer to user answer information be "no" or " no " or " non-" or
" NO " or "No", it is " not knowing " or " not knowing about " or " unclear " or " can that uncertainty, which is answered and refers to that user answers information,
Can be " or " may not be ", it is "Yes" or " to " that affirmative acknowledgement (ACK), which refers to that user answers information,；In the present embodiment, for side
Just to answering the parsing of information and improving analyzing efficiency, the answer of each problem only provides "Yes", "no" and " not really to the user
It is fixed " these three options；
For convenience of the understanding to problem set, results set and answer matrix, it is by name set of results set below
Example illustrates, that is, the problems in problem set is relevant with the name in name set, the performance of human-computer interaction at this time
Form is the form that name is guessed in enquirement, that is, the name that user is currently thought is guessd out by proposing multiple problems to user, it can
To be considered as a kind of reading heart.See table 1, the relationship between each answer coefficient in problem, name (i.e. result) and answer matrix is shown
Expectation；
Relationship between 1 problem of table, name (i.e. result), each answer coefficient three illustrates table
As a result, by table 1 it is found that Liu someone be not to be born interiorly, be singer, be performer, it is not known that whether enterpriser,
It has been got married that, and not dead；And Zhang people is not to be born, and is singer, is performer, is not enterpriser interiorly, it is not known that whether
It has been got married that, and dead.It { is born whether interiorly, if for song at this point, the problem of being prestored according to these information collection is combined into
Hand, if be performer, if be enterpriser, if got married, if dead }, the quantity n=6 of problem at this time, name set
For { Liu someone, Zhang people }, the quantity of result is m=2 at this time, and answer matrix isAnswer matrix at this time
For 2 × 6 matrixes, that is, in answer matrix, the same name is corresponded to all answer coefficients in a line, and in same row
All answer coefficients correspond to the same problem.Furthermore it is also possible to which the form of transition matrix, becomes 6 × 2 matrixes, such as
That is, in 6 × 2 answer matrix, the same problem is corresponded to all answer coefficients in a line, and all in same row answer
Case coefficient corresponds to the same name.
It is an example of the understanding to answer matrix, problem set and results set for convenience above, and it is of the invention
Be not limited to above application, such as can also by results set with sight spot gather or animation collections or collection of music or science in terms of
The form performance of guess set, problem is also required to adaptively change at this time, and user is obtained can be putd question to by limited number of time
The result thought in the heart such as sight spot or animation title or musical designation or celestial body title etc.；Also, problematic amount and fruiting quantities
Also it does not limit, e.g., can there is 100 problems, 20 results, it is possible to have the problem of greater number or smaller amounts and knot
Fruit, still, problematic amount are greater than or are equal to fruiting quantities.In the present embodiment, it is using answer matrix as m * n matrix hereinafter
Example illustrate, that is, all answer coefficients in same a line in answer matrix correspond to it is same as a result, and in same row
All answer coefficients correspond to the same problem.
S4：User is obtained to the answer information currently putd question to, and information and answer matrix are answered according to current, in result set
Qualified candidate result is obtained in conjunction, generates candidate result set；In the present embodiment, the step S4 includes the following steps：
S41：User is obtained to the answer information currently putd question to, and coefficient is answered according to answering information processing and obtaining user；
That is, if the answer information of user is "Yes", the user obtained according to "Yes" this answer information processing answers coefficient and is
1, if the answer information of user is "no", coefficient is answered according to the user that "no" this answer information processing obtains
It is -1, if it is " uncertain " that user, which answers information, is answered according to the user that " uncertain " this answer information processing obtains
Coefficient is 0；
S42：Judge that user answers whether coefficient is nonzero value, corresponds to currently if it is, being selected in answer matrix
Problem and the qualified answer coefficient consistent with user's answer coefficient, and selected in answer matrix and correspond to current problem and answer
The answer coefficient undetermined that coefficient is 0；Then it is obtained in results set and each qualified answer coefficient and each answer coefficient pair undetermined
The candidate result answered generates candidate result set；If it is not, then obtaining all as a result, generating candidate knot in results set
Fruit set；That is, when it is 1 (nonzero value) or -1 (nonzero value) that user, which answers coefficient, it is assumed that the answer coefficient of user is at this time
1, then the answer coefficient for current problem being corresponded in answer matrix and value is 1 and value is 0 is all recorded, wherein it is 1 to be worth
Answer coefficient is qualified answer coefficient, and the answer coefficient that value is 0 is answer coefficient undetermined；Further according to the answer coefficient point of record
It is corresponding as a result, these results are arranged and are preserved in a manner of set that its is not chosen, as candidate result set；And
When user's answer coefficient is 0, then it represents that the answer user of current problem can not determine, so can all retain all results
Come, the collection that all results are formed at this time is combined into candidate result set；Wherein, it should be noted that user answer coefficient be 0 this
The probability that situation occurs is smaller.
Pass through step S42 as a result, so that it is uncertain or temporarily do not preserve phase not have answer or answer to certain problems
It answers some results of answer that can first remain, avoids screening out without exception and causing to ignore that may be present accurately as a result, having
Conducive to further increase result generation accuracy rate.
S5：The problem of are carried out by assignment operation, is currently had proposed for Issue Totals sum N=N+1；And judge currently
Whether sum of asking a question is greater than or equal to preset problematic amount threshold value Nmax, is to then follow the steps S9；Otherwise, step is executed
S6；In the present embodiment, to avoid the situation for causing user to answer fatigue because puing question to number excessive from occurring, and user's body is improved
It tests and plays excitement, it is preferable that described problem amount threshold Nmax is 10~20；And user is being avoided to occur to answer the same of fatigue
When, the accuracy rate of result output is further increased, it is highly preferred that described problem amount threshold Nmax is 15；
S6：According to candidate result set and answer matrix, a problem is obtained from problem set and continues to send out enquirement；
S7：User is obtained to the answer information currently putd question to, information and answer matrix are answered according to current, in candidate result
Qualified candidate result is obtained in set, updates candidate result set；In the present embodiment, the step S7 includes following step
Suddenly：
S71：User is obtained to the answer information currently putd question to, and coefficient is answered according to answering information processing and obtaining user；
In step S71, if the answer information of user is "Yes", answered according to the user that "Yes" this answer information processing obtains
Coefficient is 1, if the answer information of user is "no", is answered according to the user that "no" this answer information processing obtains
Coefficient is -1, if it is " uncertain ", the user obtained according to " uncertain " this answer information processing that user, which answers information,
It is 0 to answer coefficient；
S72：Judge that user answers whether coefficient is nonzero value, corresponds to currently if it is, being selected in answer matrix
Problem and the qualified answer coefficient consistent with user's answer coefficient, and selected in answer matrix and correspond to current problem and answer
The answer coefficient undetermined that coefficient is 0；Then it is obtained in candidate result set and each qualified answer coefficient and each answer system undetermined
The corresponding candidate result of number, generates candidate result set；If it is not, then not updating candidate result set；That is, being returned in user
When to answer coefficient be 1 (nonzero value) or -1 (nonzero value), it is assumed that the answer coefficient of user is 1 at this time, then will be corresponding in answer matrix
It the answer coefficient that current problem and value are 1 and value is 0 all to record, wherein it is qualified answer system to be worth for 1 answer coefficient
Number, and the answer coefficient that value is 0 is answer coefficient undetermined；It is selected in candidate result set respectively further according to the answer coefficient of record
It takes its corresponding as a result, these results are arranged and are preserved again in a manner of set, that is, realizes candidate result set
Update, can further delete former candidate result set at this time, reduce the occupancy to memory；Alternatively, by candidate result set
The result for not meeting active user's answer coefficient is deleted, can also realize the update of candidate result set.And it is answered in user and is
When number is 0, then it represents that the answer user of current problem can not determine, so can all remain all candidate results, at this time
Candidate result set need not just be updated；Wherein, it should be noted that it is the probability that 0 this case occurs that user, which answers coefficient,
It is smaller.
Pass through step S72 as a result, so that it is uncertain or temporarily do not preserve phase not have answer or answer to certain problems
It answers some results of answer that can first remain, avoids screening out without exception and causing to ignore that may be present accurately as a result, having
Conducive to further increase result generation accuracy rate.
S8：Whether the number of the candidate result in candidate result set in judgment step S7 is less than or equal to fruiting quantities
Threshold value T is to then follow the steps S10, otherwise return to step S5；In the present embodiment, fruiting quantities threshold value T is preferably 3, to real
When present result needs to be investigated, reservation possibility is maximum several as a result, after a result output is wrong, can also export
On the one hand another is as a result, can be preferably bonded that user is thought as a result, also enhancing the friendship between user on the one hand
Mutually.
S9：According to all answer information and answer matrix of user, each candidate knot of candidate result set is calculated
Fruit is to the matching factor asked a question；All matching factors are ranked up according to descending sequence, and obtain matching system
Numerical digit generates new candidate result set in first T of T candidate result；In the present embodiment, the step S9 includes following step
Suddenly：
S91：According to the enquirement sequence asked a question, answers coefficient to all users and be ranked up, generate user's coefficient set
Close U, U={ U1..., UI..., UNmax, wherein UII-th user answer coefficient in expression user's coefficient sets, 1≤I≤
Nmax, and I is integer, Nmax is problematic amount threshold value；And according to the enquirement sequence asked a question, respectively to each candidate result
All qualified answer coefficients and all undetermined coefficients be ranked up, generate each candidate result respectively has selected coefficient sets
Ia, Ia={ Ia1..., IaI..., IaNmax, wherein a indicates a-th of candidate result in candidate result set, IaIndicate candidate
A-th of candidate result in results set has selected coefficient sets, IaICoefficient sets I has been selected in expressionaIn i-th answer system
Number, 1≤I≤Nmax, I are integer, and Nmax is problematic amount threshold value；And user in user's coefficient sets answer coefficient respectively with
Each answer coefficient selected in coefficient sets maps one by one；
S92：According to user coefficient sets U, calculating has respectively selected coefficient sets IaIn with mapping relations UIValue it is identical
IaINumber, respectively obtain the matching factor of each candidate result；That is, from first candidate result in candidate result set
Start to calculate matching factor, be finished until the matching factor of all candidate results in candidate result set all calculates；Wherein, often
The calculating process of the matching factor of a candidate result is：Assuming that Nmax=15, a=1 are sentenced one by one then from I=1 to I=15
It is disconnected --- I11Whether U is equal to1, I12Whether U is equal to2... ... I114Whether U is equal to14, I115Whether U is equal to15, after judgement,
If wherein have 14 pairs it is equal, matching factor is 14；If there is 8 pairs equal, then matching factor is 8；
S93：All matching factors are ranked up according to descending sequence, and obtains matching factor and is located at first T
T candidate result, generate new candidate result set；In the present embodiment, fruiting quantities threshold value T is preferably 3, then the step
S93 can be understood as：It is determined from all matching factors and is worth maximum 3 matching factors, to select this 3 matching factors
One-to-one 3 candidate results respectively, this 3 candidate results are preserved in a manner of gathering, as new candidate result collection
It closes, former candidate result set can be deleted at this time；Alternatively, being deleted other than this 3 candidate results in former candidate result set
Other candidate results, without in addition generating new candidate result set.
S10：A candidate result is chosen from candidate result set, and is exported.In the present embodiment, final defeated in order to improve
The accuracy of the candidate result gone out, step S10 is carried out it is perfect, improve after the step S10 include step：Judgement jumps to
Whether the previous step of this step is step S8, is, a candidate result is randomly selected from candidate result set, and export；It is no
Then, a candidate result with maximum matching factor is chosen from candidate result set, and is exported.
To improve the accuracy of the result exported, also wrapped after exporting candidate result as a kind of more preferably technical solution
Include following steps：
S11：User is received to currently exporting the feedback information of result；In the present embodiment, feedback information refers to output
As a result after, user can confirm the result, confirm that correct or confirmation is incorrect, to generate corresponding feedback information；
S12：Judge whether current output result is correct according to feedback information；If it is, keeping current operating conditions；Such as
Fruit is not that one in remaining candidate result output is then chosen from the candidate result set in step S10, and return to step
S11。
To improve screening degree of the problem of the choosing for the first time to result, the influence power to distribution of results is improved, result is accelerated
(generation can be understood as obtaining) efficiency is generated, as a kind of more preferably technical solution, in the step S3, is sent one's regards in advance calling
While topic set, the initial entropy set that prestores also is called；That is, not only preserving problem set, results set in database
With answer matrix, initial entropy set is also preserved.Wherein, the establishment step of the initial entropy set includes：According to prestoring
Results set and answer matrix, the entropy of each problem that the problem of prestoring is gathered to the comentropy of results set is calculated,
And record the correspondence of each problem and its entropy；By all entropy it is ascending or it is descending be arranged in order, formed initial
Entropy set.Then in the step S3, the problem of being obtained from problem set and sending out enquirement has maximum entropy for one
Problem.
Hereinafter, illustratively to the computational methods of the entropy of the comentropy of problem in the present embodiment：
Due to the essence of entropy be in a system confusion degree, the entropy of comentropy is bigger, show variable not really
It is qualitative bigger, then in the present embodiment, more candidate result can be weeded out as much as possible to realize to put question to every time so that
Result on leaf node is shorter from the path of root node, and preferably combine answer matrix, greatly reduce computational complexity and
Operation time applies selection of the comentropy realization to each asked questions.First, comentropy entropy calculation formula isAssuming that problem is respectively q1..., qi..., qn；Then need to calculate each problem for results set
Entropy, it is assumed that results set have 100 as a result, in problem qiUnder, there are 10 results to problem qiAnswer coefficient be 1,60
A result is to problem qiAnswer coefficient be -1,30 results to problem qiAnswer coefficient be 0, then problem qiLetter
Ceasing entropy is：
H(qi[100 × the log of)=- 10/2(10/100)] -60/ [100 × log2(60/100)] -30/ [100 × log2
(30/100)]=1.295；
As a result, according to above problem qiComentropy calculation, calculate the entropy of comentropy of all the problems.
Similarly, to improve screening degree of the problem of the choosing each time to result, the influence power to distribution of results is improved, is accelerated
As a result generation (generation can be understood as obtaining) efficiency, as a kind of more preferably technical solution, the step S6 includes following
Step：
S61：According to candidate result set and answer matrix, each problem of problem set is calculated to candidate result set
Comentropy entropy, and record the correspondence of each problem and its entropy；In step S61, in the calculation of entropy
The computational methods for stating entropy are identical, herein without repeating；
S62：By each problem of problem set it is ascending to all entropy of candidate result set or it is descending successively
Arrangement forms candidate entropy set；
S63：The problem corresponding with maximum entropy in candidate entropy set from selection in problem set, and continue to send out and carry
It asks.
In addition, the present invention also provides a kind of bases that can realize the above-mentioned human-computer interaction result generation method based on matrix
System is generated in the human-computer interaction result of matrix, which includes that readable storage medium storing program for executing, processor, input equipment and output are set
It is standby.
The readable storage medium storing program for executing is stored with some instructions；The processor loads according to some instructions and executes institute
The human-computer interaction result generation method based on matrix stated；The input equipment is for inputting question and answer open command and user feedback
Information is to processor；The output equipment exports the result handled by processor.
In addition, carrying the beauty with shape for convenience, it is preferable that the present invention is based on the generations of the human-computer interaction result of matrix
System further includes shell；The storage medium, processor and output equipment are installed in the shell, wherein output equipment
The input terminal of output end and input equipment exposes to shell.And further, shell can be manufactured in a manner of robot,
It can also be manufactured in other shapes.
The operation principle that the human-computer interaction result based on matrix of the present invention generates system can be from the present invention based on matrix
Human-computer interaction result generation method learn that therefore this will not be repeated here.
Although the present invention does not refer to specific system structure, to support this system normal work, the power supply needed
Module, processor peripheral circuit etc. are all necessary parts, and these must be partly prior art part, therefore without repeating.
Compared with the existing technology, the present invention is based on the human-computer interaction result generation methods and system of matrix, by being stored with
The distribution relation of matrix the record problem and result of answer, clearly reflects the dependence between each problem and each result, from
A large amount of computational space is occupied without carrying out Model Reconstruction, greatly speeds up operation efficiency, while also improving and finally obtaining
Result accuracy.And it was proved that, the present invention goes to carry out human-computer interaction using 8422 results and 1860 problems,
Result can be returned to user in 0.1 second.
Several embodiments of the invention above described embodiment only expresses, the description thereof is more specific and detailed, but simultaneously
It cannot therefore be construed as limiting the scope of the patent.It should be pointed out that coming for those of ordinary skill in the art
It says, without departing from the inventive concept of the premise, various modifications and improvements can be made, these belong to the protection of the present invention
Range.
Claims (10)
1. a kind of human-computer interaction result generation method based on matrix, it is characterised in that：Include the following steps：
S1：It has detected whether that question and answer open command inputs, has been to then follow the steps S2, otherwise keeps standby mode；
S2：The problem of initialization currently has proposed sum N；
S3：The problem of calling prestores set, results set and answer matrix, obtain a problem and are sent out from problem set and carry
It asks；Wherein, described problem set is made of n different problems, and the results set is made of m different results；It is described
Answer matrix is m * n matrix, and the elements A ij in answer matrix is answer coefficient of i-th of result for j-th of problem；Or
Person, the answer matrix is n × m matrixes, and the elements A ij in answer matrix is answer of j-th of result for i-th of problem
Coefficient；
S4：User is obtained to the answer information currently putd question to, and information and answer matrix are answered according to current, in results set
Qualified candidate result is obtained, candidate result set is generated；
S5：The problem of are carried out by assignment operation, is currently had proposed for Issue Totals sum N=N+1；And judge currently to have putd question to
Whether topic sum is greater than or equal to preset problematic amount threshold value Nmax, is to then follow the steps S9；Otherwise, step S6 is executed；
S6：According to candidate result set and answer matrix, a problem is obtained from problem set and continues to send out enquirement；
S7：User is obtained to the answer information currently putd question to, information and answer matrix are answered according to current, in candidate result set
It is middle to obtain qualified candidate result, update candidate result set；
S8：Whether the number of the candidate result in candidate result set in judgment step S7 is less than or equal to fruiting quantities threshold value
T is to then follow the steps S10, otherwise return to step S5；
S9：According to all answer information and answer matrix of user, each candidate result pair of candidate result set is calculated
The matching factor asked a question；All matching factors are ranked up according to descending sequence, and obtain matching factor position
In first T of T candidate result, new candidate result set is generated；
S10：A candidate result is chosen from candidate result set, and is exported.
2. the human-computer interaction result generation method according to claim 1 based on matrix, it is characterised in that：Answer coefficient
Value is -1 or 0 or 1；When the value of answer coefficient is -1, answer of the corresponding result of answer coefficient for corresponding problem is indicated
For negative acknowledge；When the value of answer coefficient is 0, indicate that the corresponding result of answer coefficient is not for the answer of corresponding problem
Certainty is answered；When the value of answer coefficient is 1, indicate that the corresponding result of answer coefficient is willing for the answer of corresponding problem
It is fixed to answer.
3. the human-computer interaction result generation method according to claim 1 based on matrix, it is characterised in that：The step S3
In, while calling prestores problem set, also call the initial entropy set that prestores；The establishment step of the initial entropy set
Including：According to the results set and answer matrix to prestore, each problem that the problem of prestoring is gathered is calculated to results set
The entropy of comentropy, and record the correspondence of each problem and its entropy；By all entropy it is ascending or it is descending successively
Arrangement, forms initial entropy set；
And in the step S3, the problem of being obtained from problem set and sending out enquirement, has the problem of maximum entropy for one.
4. the human-computer interaction result generation method according to claim 2 based on matrix, it is characterised in that：The step S4
Include the following steps：
User is obtained to the answer information currently putd question to, and coefficient is answered according to answering information processing and obtaining user；
Judge user answer coefficient whether be nonzero value, if it is, in answer matrix select corresponding to current problem and with
User answers coefficient consistent qualified answer coefficient, and the selected current problem and answer coefficient of corresponding to is 0 in answer matrix
Answer coefficient undetermined；Then candidate corresponding with each qualified answer coefficient and each answer coefficient undetermined is obtained in results set
As a result, generating candidate result set；If it is not, then obtaining all as a result, generating candidate result set in results set.
5. the human-computer interaction result generation method according to claim 4 based on matrix, it is characterised in that：The step S6
Include the following steps：
According to candidate result set and answer matrix, comentropy of each problem to candidate result set of problem set is calculated
Entropy, and record the correspondence of each problem and its entropy；
By each problem of problem set it is ascending to all entropy of candidate result set or it is descending be arranged in order, formed
Candidate entropy set；
The problem corresponding with maximum entropy in candidate entropy set from selection in problem set, and continue to send out enquirement.
6. the human-computer interaction result generation method according to claim 5 based on matrix, it is characterised in that：The step S7
Include the following steps：
User is obtained to the answer information currently putd question to, and coefficient is answered according to answering information processing and obtaining user；
Judge user answer coefficient whether be nonzero value, if it is, in answer matrix select corresponding to current problem and with
User answers coefficient consistent qualified answer coefficient, and the selected current problem and answer coefficient of corresponding to is 0 in answer matrix
Answer coefficient undetermined；Then it is obtained in candidate result set corresponding with each qualified answer coefficient and each answer coefficient undetermined
Candidate result generates candidate result set；If it is not, then not updating candidate result set.
7. the human-computer interaction result generation method according to claim 6 based on matrix, it is characterised in that：The step S9
Include the following steps：
According to the enquirement sequence asked a question, answers coefficient to all users and be ranked up, generate user's coefficient sets U, U=
{U1..., UI..., UNmax, wherein UIIndicate that the i-th user in user's coefficient sets answers coefficient, 1≤I≤Nmax, and I
For integer, Nmax is problematic amount threshold value；And according to the enquirement sequence asked a question, respectively to all conjunctions of each candidate result
Lattice answer coefficient and all undetermined coefficients are ranked up, and generate each candidate result respectively has selected coefficient sets Ia, Ia={ Ia1..., IaI..., IaNmax, wherein a indicates a-th of candidate result in candidate result set, IaIndicate candidate result set
In a-th of candidate result selected coefficient sets, IaICoefficient sets I has been selected in expressionaIn i-th answer coefficient, 1≤I≤
Nmax, I are integer, and Nmax is problematic amount threshold value；And the user in user's coefficient sets answers coefficient and has been selected respectively with each
Answer coefficient in coefficient sets maps one by one；
According to user coefficient sets U, calculating has respectively selected coefficient sets IaIn with mapping relations UIThe identical I of valueaI
Number, respectively obtains the matching factor of each candidate result；
All matching factors are ranked up according to descending sequence, and obtain matching factor positioned at first T of T candidate
As a result, generating new candidate result set.
8. the human-computer interaction result generation method according to claim 7 based on matrix, it is characterised in that：The step
S10 includes step：Judgement jumps to whether the previous step of this step is step S8, is selected at random from candidate result set
A candidate result is taken, and is exported；Otherwise, a candidate result with maximum matching factor is chosen from candidate result set, and
Output.
9. according to human-computer interaction result generation method of claim 1~8 any one of them based on matrix, it is characterised in that：
In the step S5, problematic amount threshold value Nmax is 10~20；In the step S8, fruiting quantities threshold value T is 3.
10. a kind of human-computer interaction result based on matrix generates system, it is characterised in that：Including readable storage medium storing program for executing, processor
And output equipment；The readable storage medium storing program for executing is stored with some instructions；The processor is loaded and is held according to some instructions
Human-computer interaction result generation method of row claim 1~9 any one of them based on matrix；Output equipment output by
The result that reason device is handled.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810326047.5A CN108536299B (en) | 2018-04-12 | 2018-04-12 | Matrix-based man-machine interaction result generation method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810326047.5A CN108536299B (en) | 2018-04-12 | 2018-04-12 | Matrix-based man-machine interaction result generation method and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108536299A true CN108536299A (en) | 2018-09-14 |
| CN108536299B CN108536299B (en) | 2021-04-27 |
Family
ID=63479949
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810326047.5A Active CN108536299B (en) | 2018-04-12 | 2018-04-12 | Matrix-based man-machine interaction result generation method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108536299B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178199A (en) * | 2019-12-20 | 2020-05-19 | 福建天泉教育科技有限公司 | Interactive question and answer method and terminal based on real-time image recognition |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005230290A (en) * | 2004-02-19 | 2005-09-02 | Sega Corp | Quiz game device |
| US20070111187A1 (en) * | 2005-10-24 | 2007-05-17 | Stu Rasmussen | Multiple-choice question database enhancement |
| CN102081631A (en) * | 2009-11-30 | 2011-06-01 | 国际商业机器公司 | Answer support system and method |
| US20140032540A1 (en) * | 1999-07-20 | 2014-01-30 | Google Inc. | Internet based system and apparatus for paying users to view content and receiving micropayments |
| CN105893391A (en) * | 2015-01-26 | 2016-08-24 | 阿里巴巴集团控股有限公司 | Intelligent answering method, apparatus and system, and electronic device |
| CN106844368A (en) * | 2015-12-03 | 2017-06-13 | 华为技术有限公司 | For interactive method, nerve network system and user equipment |
| CN107026940A (en) * | 2017-05-18 | 2017-08-08 | 北京神州泰岳软件股份有限公司 | A kind of method and apparatus for determining session feedback information |
| CN107818164A (en) * | 2017-11-02 | 2018-03-20 | 东北师范大学 | A kind of intelligent answer method and its system |
-
2018
- 2018-04-12 CN CN201810326047.5A patent/CN108536299B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20140032540A1 (en) * | 1999-07-20 | 2014-01-30 | Google Inc. | Internet based system and apparatus for paying users to view content and receiving micropayments |
| JP2005230290A (en) * | 2004-02-19 | 2005-09-02 | Sega Corp | Quiz game device |
| US20070111187A1 (en) * | 2005-10-24 | 2007-05-17 | Stu Rasmussen | Multiple-choice question database enhancement |
| CN102081631A (en) * | 2009-11-30 | 2011-06-01 | 国际商业机器公司 | Answer support system and method |
| CN105893391A (en) * | 2015-01-26 | 2016-08-24 | 阿里巴巴集团控股有限公司 | Intelligent answering method, apparatus and system, and electronic device |
| CN106844368A (en) * | 2015-12-03 | 2017-06-13 | 华为技术有限公司 | For interactive method, nerve network system and user equipment |
| CN107026940A (en) * | 2017-05-18 | 2017-08-08 | 北京神州泰岳软件股份有限公司 | A kind of method and apparatus for determining session feedback information |
| CN107818164A (en) * | 2017-11-02 | 2018-03-20 | 东北师范大学 | A kind of intelligent answer method and its system |
Non-Patent Citations (1)
| Title |
|---|
| 刘康 等: ""基于表示学习的知识库问答研究进展与展望"", 《自动化学报》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111178199A (en) * | 2019-12-20 | 2020-05-19 | 福建天泉教育科技有限公司 | Interactive question and answer method and terminal based on real-time image recognition |
| CN111178199B (en) * | 2019-12-20 | 2023-04-28 | 福建天泉教育科技有限公司 | Interactive question-answering method and terminal based on real-time image recognition |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108536299B (en) | 2021-04-27 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Wu et al. | A discrete particle swarm optimization for covering array generation | |
| CN107741976B (en) | Intelligent response method, device, medium and electronic equipment | |
| CN107516510A (en) | A kind of smart machine automated voice method of testing and device | |
| CN109165720A (en) | Neural network model compression method, device and computer equipment | |
| CN104965695B (en) | The method and apparatus of analog subscriber real-time operation | |
| CN110222164A (en) | A kind of Question-Answering Model training method, problem sentence processing method, device and storage medium | |
| CN107391549A (en) | News based on artificial intelligence recalls method, apparatus, equipment and storage medium | |
| CN109360550A (en) | Test method, device, equipment and the storage medium of voice interactive system | |
| CN107609130A (en) | A kind of method and server for selecting data query engine | |
| US5228115A (en) | Hybrid backtrack/lookahead search technique for constraint-satisfaction problems | |
| CN109299245A (en) | The method and apparatus that knowledge point is recalled | |
| CN109558310A (en) | Method for generating test case and device | |
| CN108280207A (en) | A method of the perfect Hash of construction | |
| CN109753436A (en) | Test method, device and the storage medium of intelligent customer service system | |
| CN108536299A (en) | Human-computer interaction result generation method based on matrix and system | |
| CN107562847A (en) | Information processing method and related product | |
| CN114547343A (en) | Method, apparatus, device, storage medium and program product for knowledgegraph completion | |
| CN109493862A (en) | Terminal, the determination method of voice server and computer readable storage medium | |
| CN109471951A (en) | Lyrics generation method, device, equipment and storage medium neural network based | |
| CN109977977A (en) | A kind of method and corresponding intrument identifying potential user | |
| CN111814414A (en) | Coverage rate convergence method and system based on genetic algorithm | |
| CN109656952A (en) | Inquiry processing method, device and electronic equipment | |
| CN115544033A (en) | Method, device, equipment and medium for updating weight checking vector library and data weight checking | |
| CN107433032A (en) | Chess game data processing method and device | |
| CN107169032A (en) | Database operation method and database server |
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 |