CN110221704A - A kind of input method, device and the device for input - Google Patents
A kind of input method, device and the device for input Download PDFInfo
- Publication number
- CN110221704A CN110221704A CN201810172598.0A CN201810172598A CN110221704A CN 110221704 A CN110221704 A CN 110221704A CN 201810172598 A CN201810172598 A CN 201810172598A CN 110221704 A CN110221704 A CN 110221704A
- Authority
- CN
- China
- Prior art keywords
- word
- group
- result
- score
- sequence
- 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.)
- Pending
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
- G06F3/02—Input arrangements using manually operated switches, e.g. using keyboards or dials
- G06F3/023—Arrangements for converting discrete items of information into a coded form, e.g. arrangements for interpreting keyboard generated codes as alphanumeric codes, operand codes or instruction codes
- G06F3/0233—Character input methods
- G06F3/0237—Character input methods using prediction or retrieval techniques
Abstract
The embodiment of the invention provides a kind of input method, device and for the device of input.Method therein comprises determining that corresponding group of word result of input string；Described group of word result is ranked up according to shot and long term memory network model；Group word result after sequence is shown.The embodiment of the present invention can make the group word result for showing user under the input scene of long entry more reasonable, be more in line with the actual demand of user.
Description
Technical field
The present invention relates to input method technique field more particularly to a kind of input method, device and for the device of input.
Background technique
For such as user of the language such as Chinese, Japanese, Korean, it is typically necessary through input method procedure and calculates
Machine interacts.For example, user can input string by a keyboard entry, dictionary and foundation are then inquired by input method procedure
The input string is converted to candidate item and the displaying of corresponding language, and then the candidate that user is selected by preset Standard Map rule
Xiang Shangping.
In practical applications, due to being limited by dictionary size, high frequency entry, and dictionary are usually only stored in dictionary
The length of each entry of middle storage is limited, such as each entry is no more than 5 words.For long entry, input method procedure
It is acquired by way of intelligent word, specifically, input method procedure is closed according to the n member in n-gram (the n member syntax) model
System carries out intelligent word, and common n-gram is generally included: the Bi-Gram of the binary and Tri-Gram of ternary.For example, " capital "
" Beijing " there is binary crelation " capital Beijing " therefore can be obtained by intelligent word.
Assuming that user is intended to input long entry " aircraft is sliding ", the input string that input method procedure obtains user is
" feijizhengzaihuaxing ", input method procedure carry out intelligent word for the input string using n-gram model, and
The group word result optimal to user's displaying group word path.However, input method procedure using n-gram model for the input string into
Row intelligent word, the obtained optimal group word result in group word path may be " aircraft Hua Xing "；This group of word result is not inconsistent simultaneously
Syntax rule is closed, the input demand of user is not met yet.
Summary of the invention
The embodiment of the present invention provides a kind of input method, device and the device for input, to solve to exist in the prior art
Under the input scene of long entry, the group word result that input method procedure is shown to user is possible and unreasonable, does not also meet user's
The problem of input demand.
To solve the above-mentioned problems, the embodiment of the invention discloses a kind of input methods, comprising:
Determine corresponding group of word result of input string；
Described group of word result is ranked up according to shot and long term memory network model；
Group word result after sequence is shown.
It is optionally, described that described group of word result is ranked up according to shot and long term memory network model, comprising:
Described group of word result is segmented, to obtain word sequence；
According to the shot and long term memory network model, corresponding first score of the word sequence is determined；
According to first score, corresponding group of word result of the word sequence is ranked up.
Optionally, described that corresponding first score of the word sequence is determined according to the shot and long term memory network model, packet
It includes:
According to the shot and long term memory network model, the corresponding prediction word of word in the word sequence and described is determined
Predict the probability value of word；
According to the probability value of the corresponding prediction word of word, institute's predicate and the prediction word in the word sequence, institute is determined
Corresponding first score of predicate sequence.
It is optionally, described that described group of word result is ranked up according to shot and long term memory network model, comprising:
According to the shot and long term memory network model, the first score of word sequence corresponding to described group of word result is determined；
According to multi-component grammar model, corresponding second score of described group of word result is determined；
According to first score and second score, described group of word result is ranked up.
It is optionally, described that described group of word result is ranked up according to shot and long term memory network model, comprising:
If corresponding group of word result of the input string meets predetermined order condition, according to shot and long term memory network model pair
Described group of word result is ranked up.
Optionally, judge whether corresponding group of word result of the input string meets predetermined order condition as follows:
The optimal group word result in group word path is determined from corresponding group of word result of the input string；
If there is binary crelation between adjacent word in the optimal group word result in described group of word path, it is determined that the input
Corresponding group of word result of going here and there is unsatisfactory for predetermined order condition；It is preset otherwise, it determines corresponding group of word result of the input string meets
Sort criteria.
Optionally, the method also includes:
If corresponding group of word result of the input string is unsatisfactory for predetermined order condition, show that described group of word path is optimal
Group word result.
On the other hand, the embodiment of the invention discloses a kind of input units, comprising:
Group word result determining module, for determining corresponding group of word result of input string；
Group word sort result module, for being ranked up according to shot and long term memory network model to described group of word result；
First display module, for being shown to the group word result after sequence.
Optionally, described group of word sort result module, comprising:
Submodule is segmented, for segmenting to described group of word result, to obtain word sequence；
First score computational submodule, for determining that the word sequence is corresponding according to the shot and long term memory network model
The first score；
First sorting sub-module, for arranging corresponding group of word result of the word sequence according to first score
Sequence.
Optionally, the first score computational submodule, comprising:
Predicting unit, for determining that the word in the word sequence is corresponding pre- according to the shot and long term memory network model
Survey the probability value of word and the prediction word；
Computing unit, for according to word, the institute's predicate corresponding prediction word and the prediction word in the word sequence
Probability value determines corresponding first score of the word sequence.
Optionally, described group of word sort result module, comprising:
First score computational submodule, for determining described group of word result institute according to the shot and long term memory network model
First score of corresponding word sequence；
Second score computational submodule, for determining that described group of word result corresponding second is obtained according to multi-component grammar model
Point；
Second sorting sub-module, for being carried out to described group of word result according to first score and second score
Sequence.
Optionally, described group of word sort result module, comprising:
Third sorting sub-module, if meeting predetermined order condition, basis for corresponding group of word result of the input string
Shot and long term memory network model is ranked up described group of word result.
Optionally, described device further include: condition judgment module, for judging corresponding group of word of the input string the result is that
It is no to meet predetermined order condition；The condition judgment module, comprising:
Optimal result determines submodule, for determining that group word path is optimal from corresponding group of word result of the input string
Group word result；
Judging submodule, if between word adjacent in the group word result optimal for described group of word path there is binary to close
System, it is determined that corresponding group of word result of the input string is unsatisfactory for predetermined order condition；Otherwise, it determines the input string is corresponding
Group word result meets predetermined order condition.
Optionally, described device further include:
Second display module is shown if being unsatisfactory for predetermined order condition for corresponding group of word result of the input string
The optimal group word result in described group of word path.
In another aspect, the embodiment of the invention discloses a kind of device for input, include memory and one or
The more than one program of person, one of them perhaps more than one program be stored in memory and be configured to by one or
It includes the instruction for performing the following operation that more than one processor, which executes the one or more programs:
Determine corresponding group of word result of input string；
Described group of word result is ranked up according to shot and long term memory network model；
Group word result after sequence is shown.
It is optionally, described that described group of word result is ranked up according to shot and long term memory network model, comprising:
Described group of word result is segmented, to obtain word sequence；
According to the shot and long term memory network model, corresponding first score of the word sequence is determined；
According to first score, corresponding group of word result of the word sequence is ranked up.
Optionally, described that corresponding first score of the word sequence is determined according to the shot and long term memory network model, packet
It includes:
According to the shot and long term memory network model, the corresponding prediction word of word in the word sequence and described is determined
Predict the probability value of word；
According to the probability value of the corresponding prediction word of word, institute's predicate and the prediction word in the word sequence, institute is determined
Corresponding first score of predicate sequence.
It is optionally, described that described group of word result is ranked up according to shot and long term memory network model, comprising:
According to the shot and long term memory network model, the first score of word sequence corresponding to described group of word result is determined；
According to multi-component grammar model, corresponding second score of described group of word result is determined；
According to first score and second score, described group of word result is ranked up.
It is optionally, described that described group of word result is ranked up according to shot and long term memory network model, comprising:
If corresponding group of word result of the input string meets predetermined order condition, according to shot and long term memory network model pair
Described group of word result is ranked up.
Optionally, described device is also configured to execute one or one by one or more than one processor
Procedure above includes the instruction for performing the following operation:
The optimal group word result in group word path is determined from corresponding group of word result of the input string；
If there is binary crelation between adjacent word in the optimal group word result in described group of word path, it is determined that the input
Corresponding group of word result of going here and there is unsatisfactory for predetermined order condition；It is preset otherwise, it determines corresponding group of word result of the input string meets
Sort criteria.
Optionally, the processor is also configured to execute one or one by one or more than one processor
A procedure above includes the instruction for performing the following operation:
If corresponding group of word result of the input string is unsatisfactory for predetermined order condition, show that described group of word path is optimal
Group word result.
In another aspect, be stored thereon with instruction the embodiment of the invention discloses a kind of machine readable media, when by one or
When multiple processors execute, so that device executes above-mentioned input method.
The embodiment of the present invention includes following advantages:
The embodiment of the present invention is after obtaining corresponding group of word result of input string, according to LSTM model to described group of word result
It is ranked up；Since LSTM model is a kind of time recurrent neural network, the dependence between remote word can be obtained,
For example, for input string " feijizhengzaihuaxing ", than between " aircraft " and " Hua Xing " between " aircraft " and " sliding "
With more dependence, therefore, according to LSTM model to a group word result " aircraft Hua Xing ", " aircraft is sliding ", " aircraft
Just in flower pattern " it is ranked up, a group word result " aircraft is sliding " can be ranked the first, input method procedure is according to the sequence knot
Fruit shows described group of word as a result, can be to user's displaying group word result " aircraft is sliding ", so that in the input field of long entry
Under scape, shows the group word result of user more reasonable, be more in line with the actual demand of user.
Detailed description of the invention
In order to illustrate the technical solution of the embodiments of the present invention more clearly, below by institute in the description to the embodiment of the present invention
Attached drawing to be used is needed to be briefly described, it should be apparent that, the accompanying drawings in the following description is only some implementations of the invention
Example, for those of ordinary skill in the art, without any creative labor, can also be according to these attached drawings
Obtain other attached drawings.
Fig. 1 is a kind of step flow chart of input method embodiment one of the invention；
Fig. 2 is a kind of step flow chart of input method embodiment two of the invention；
Fig. 3 is a kind of structural block diagram of input unit embodiment of the invention；
Fig. 4 is a kind of block diagram of device 800 for input of the invention；And
Fig. 5 is the structural schematic diagram of server in some embodiments of the present invention.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are some of the embodiments of the present invention, instead of all the embodiments.Based on this hair
Embodiment in bright, every other implementation obtained by those of ordinary skill in the art without making creative efforts
Example, shall fall within the protection scope of the present invention.
Embodiment of the method one
Referring to Fig.1, a kind of step flow chart of input method embodiment one of the invention is shown, can specifically include:
Step 101 determines corresponding group of word result of input string；
Step 102 is ranked up described group of word result according to shot and long term memory network model；
Step 103 is shown the group word result after sequence.
The embodiment of the present invention can be applied to the input method procedure of various input modes, for example, Pinyin Input, five it is defeated
Enter, English input, stroke input, handwriting input and voice input etc..User can be complete by any one of the above input mode
At the input of input string, that is, user can be inputted by physical keyboard, dummy keyboard, handwriting pad, touch screen etc..Its
In, input string, which can be, to be made of any one of number, symbol, phonetic, English alphabet etc. or several.For ease of description,
The embodiment of the present invention is illustrated using pinyin string as input string, and other types of input string is cross-referenced.
In a particular application, input method procedure can receive the input string of user, and then inquiring in dictionary whether there is
The corresponding entry of the input string then carries out intelligent word to input string according to n-gram model, obtains multiple groups of words if it does not exist
As a result, and corresponding group of word path of the multiple group of word result is ranked up, therefrom determine the optimal group word in group word path
As a result, and showing the optimal group word result in this group of word path to user.
For example, input method procedure carries out syllable to the input string and cuts for input string " feijizhengzaihuaxing "
Point, available following pinyin string " feiji ' zhengzai ' huaxing ", what it is due to dictionary storage is shorter basic entry,
There is no the corresponding long entries of the pinyin string, therefore input method procedure is needed according to n-gram model to input string
" feijizhengzaihuaxing " carries out intelligent word.Specifically, by inquiring dictionary, available syllable " feiji " is right
The entry answered includes: " aircraft ", " Fiji ", " fertile chicken " etc.；The corresponding entry of syllable " zhengzai " includes:；Syllable
" huaxing " corresponding entry includes: " sliding ", " Hua Xing ", " flower pattern " etc..N-gram model is closed according to the binary between entry
System or n-tuple relation, determine a group word path preferably organize word result include: " aircraft Hua Xing ", " aircraft is sliding ",
" aircraft is just in flower pattern " etc..
Due to n-gram model based on it is such a it is assumed that n-th of word appearance only it is related to the word of front n-1, and with
Other any words are all uncorrelated, in addition, for the n-gram model for being higher than ternary (such as quaternary, five yuan), not only training process
Huge corpus is needed, and Sparse is serious, time complexity is high, and accuracy but improves few, therefore is rarely employed.
And the Bi-Gram model of binary and the Tri-Gram model of ternary are more universal.Wherein, if the appearance of a word only relies only on
In the word that its front occurs, referred to as Bi-Gram.If the appearance of a word only relies upon two that its front occurs
Word, referred to as Tri-Gram.Therefore, n-gram model does not often consider that the dependence between remote word is closed during group word
System, the remote word refers to non-conterminous word in long entry, for example, " flying in long entry " aircraft Hua Xing "
Machine " and " Hua Xing " are remote word, and n-gram model is during group word, if n=2, it is not intended that " aircraft " and " China
It whether there is dependence between star ", therefore, n-gram model is ranked up above-mentioned group of word result according to group word path, obtains
The group word result optimal to group word path may be " aircraft Hua Xing ", and therefore, input method procedure is to user's displaying group word knot
Fruit " aircraft Hua Xing ", however, what user actually wanted to input is " aircraft is sliding ".As can be seen that input normal direction is used
The group word result of family displaying is simultaneously unreasonable, does not also meet the input demand of user.In addition, when user's entry to be inputted is longer,
Or user, when being intended to input a sentence, even the n-gram model of ternary, remaining on can not be obtained between remote word
Dependence, such as the institute of the masses " collection anxious ", n-gram model is it is not intended that with the presence or absence of dependence between " collection " and " urgency ".
In order to enable input method procedure can show group word that is more reasonable and meeting user demand as a result, originally to user
Inventive embodiments are after obtaining corresponding group of word result of input string, according to LSTM (Long Short-Term Memory, length
Phase memory network) model is ranked up described group of word result, and is shown to the group word result after sequence.
Since LSTM model is a kind of time recurrent neural networks model, the dependence that can be obtained between remote word is closed
System, for example, for input string " feijizhengzaihuaxing ", between " aircraft " and " sliding " than " aircraft " and " Hua Xing " it
Between have more dependence, therefore, according to LSTM model to above-mentioned group of word result " aircraft Hua Xing ", " aircraft is being slided
Row ", " aircraft is just in flower pattern " are ranked up, and group word result " aircraft is sliding " is ranked the first, and input method procedure is according to this
Ranking results show described group of word as a result, can be to user's displaying group word result " aircraft is sliding ", so that showing user
Group word result it is more reasonable, be more in line with the actual demand of user.
In an alternative embodiment of the invention, it is described according to shot and long term memory network model to described group of word result into
Row sequence, can specifically include:
If it is determined that corresponding group of word result of the input string meets predetermined order condition, then according to shot and long term memory network mould
Type is ranked up described group of word result.
In embodiments of the present invention, the time it takes, this hair are ranked up in order to save LSTM model to a group word result
Whether bright embodiment can also judge corresponding group of word result of the input string after obtaining corresponding group of word result of input string
Meet predetermined order condition, when determining that corresponding group of word result of the input string meets predetermined order condition, just according to recurrence
Neural network model is ranked up described group of word result.
In an alternative embodiment of the invention, corresponding group of word knot of the input string can be judged as follows
Whether fruit meets predetermined order condition:
Step S11, the optimal group word result in group word path is determined from corresponding group of word result of the input string；
If step S12, there is binary crelation between word adjacent in the optimal group word result in described group of word path, it is determined that
Corresponding group of word result of the input string is unsatisfactory for predetermined order condition；Otherwise, it determines corresponding group of word result of the input string
Meet predetermined order condition.
Specifically, input method procedure can carry out intelligent word to input string according to n-gram model, obtain multiple groups of words
As a result, and corresponding group of word path of the multiple group of word result is ranked up, therefrom determine the optimal group word in group word path
As a result.If there is binary crelation between adjacent word in the optimal group word result in this group of word path, illustrate this group of word path most
Excellent group word result is relatively reasonable, therefore, can not have to carry out corresponding group of word result of the input string according to LSTM model
Sequence.If there is the adjacent word without binary crelation in the optimal group word result in this group of word path, illustrate this group of word
The optimal group word result in path may be not reasonable, presets hence, it can be determined that corresponding group of word result of the input string meets
Sort criteria needs to be ranked up the corresponding group of word result of the input string according to LSTM model.
In a kind of application example of the invention, for input string " feijizhengzaihuaxing ", according to n-gram
Model carries out a group word, and is ranked up according to group word path to a group word result, and the sequence of available group of word result is as follows:
" aircraft Hua Xing ", " aircraft is sliding ", " aircraft is just in flower pattern " etc..Assuming that obtaining a group word path according to n-gram model
Optimal group word result is " aircraft Hua Xing ", then the group word result optimal to this group of word path segments, available
Word sequence " aircraft | | Hua Xing ", wherein there is binary crelation between adjacent word " aircraft " and " ", but it is adjacent
Do not have binary crelation between word " " and " Hua Xing ", can determine the optimal group word result in this group of word path and unreasonable,
Hence, it can be determined that described corresponding group of word result of input string " feijizhengzaihuaxing " meets predetermined order condition,
It needs to be ranked up corresponding group of word result of the input string " feijizhengzaihuaxing " according to LSTM model.
In another kind application example of the invention, for input string " woaizhongguocai ", it is assumed that according to n-gram
Model carries out a group word, and is ranked up according to group word path to a group word result, and the sequence of available group of word result is as follows:
" I likes Chinese dishes ", " I likes Chinese ", " I likes that China adopts " etc., wherein the optimal group word result in group word path is " in my love
State's dish ", then the group word result optimal to this group of word path segments, available word sequence " I likes | China | dish ", wherein
There is binary crelation between adjacent word " I likes " and " China ", and also there is binary between adjacent word " China " and " dish "
Relationship can determine that the optimal group word result in this group of word path is relatively reasonable, hence, it can be determined that the input string
" woaizhongguocai " corresponding group of word result is unsatisfactory for predetermined order condition, can not have to according to LSTM model to described
Corresponding group of word result of input string " woaizhongguocai " is ranked up, and directly can show this group of word path most to user
Excellent group word result " I likes Chinese dishes ".
In an alternative embodiment of the invention, the method can also include:
If corresponding group of word result of the input string is unsatisfactory for predetermined order condition, show that described group of word path is optimal
Group word result.
If corresponding group of word result of input string is unsatisfactory for predetermined order condition, namely the group word knot that current group word path is optimal
Fruit is relatively reasonable, therefore can not have to be ranked up corresponding group of word result of the input string according to LSTM model, Ke Yizhi
Connect the group word result for showing that described group of word path is optimal.
In an alternative embodiment of the invention, it is described according to shot and long term memory network model to described group of word result into
Row sequence, can specifically include:
Step S21, described group of word result is segmented, to obtain word sequence；
Optionally, in the available corresponding group of word result of the input string of the embodiment of the present invention organize word paths ordering before N
The group word result of position, wherein N is the natural number more than or equal to 1, and is arranged according to LSTM model this N number of group of word result
Sequence.In practical applications, the embodiment of the present invention is without restriction for the specific value of N, and the embodiment of the present invention is by taking N=4 as an example.
For example, input method procedure is according to n-gram model to input for input string " feijizhengzaihuaxing "
String " feijizhengzaihuaxing " carries out a group word, and is ranked up according to group word path to a group word result, available
4 group word results are as follows before group word paths ordering: " aircraft Hua Xing ", " aircraft is sliding ", " aircraft is just in flower pattern ",
" aircraft is just in Huaxing ".This 4 group word results are segmented, to obtain the corresponding word sequence of each group of word result.Specifically,
Available following word sequence: " aircraft | | Hua Xing ", " aircraft | | slide ", " aircraft | | flower pattern ", " aircraft |
| Huaxing ".
Step S22, according to the shot and long term memory network model, corresponding first score of the word sequence is determined；
Optionally, the embodiment of the present invention first can carry out duplicate removal processing to the word sequence before calculating the first score,
Again to the first score of word order column count after duplicate removal processing.For example, word sequence " go | where | have a meal " and " where | have a meal "
For identical group of word as a result, therefore, one of them can be deleted in word sequence.
Then, the word sequence after duplicate removal processing is sequentially input into LSTM model, LSTM model exports the word sequence pair
The first score answered.
In an alternative embodiment of the invention, described according to the shot and long term memory network model, determine institute's predicate
Corresponding first score of sequence, can specifically include:
Step S221, according to the shot and long term memory network model, the corresponding prediction word of word in the word sequence is determined,
And the probability value of the prediction word；
Step S222, according to the probability of the corresponding prediction word of word, institute's predicate and the prediction word in the word sequence
Value, determines corresponding first score of the word sequence.
LSTM model can predict the probability value that subsequent word occurs, and according to the probability value according to the word of front
Calculate the first score of entire word sequence.For example, for word sequence " aircraft | | slide ", can by " aircraft ", " ",
" sliding " sequentially inputs LSTM model, and LSTM model can determine the general of the corresponding prediction word of each word and the prediction word
Rate value, wherein the prediction word is next word that some word is likely to occur later, and such as " aircraft " corresponding prediction word can wrap
Include " ", " taking off ", " design " etc., and the probability that each prediction word is likely to occur is different.LSTM model calculates first " to fly
The corresponding prediction word of machine " is the probability value of " ", namely calculates the probability value of appearance " " after " aircraft ", and to this
Probability value takes the corresponding score of the available prediction word " " of logarithm.Then, then to calculate " " corresponding prediction word be " sliding
The probability value of row ", namely the probability value of appearance " sliding " after " " is calculated, and take logarithm available the probability value
Predict word " sliding " corresponding score.Finally, sum to the scores of all prediction words, the of available entire word sequence
One score, such as carrying out summation to " " and " sliding " corresponding score can be obtained entire word sequence " aircraft | | slide "
The first score.
Since LSTM model is during determining the first score of word sequence, can consider at a distance segment between according to
The relationship of relying, can be true during determining the first score of word sequence " aircraft | | Hua Xing " for example, in the examples described above
Determine the corresponding prediction word of each word, and therefore the probability value of prediction word when giving a mark to word " Hua Xing ", can combine it
The probability value namely LSTM model of the corresponding prediction word of preceding word " aircraft " can embody the pass of the dependence between remote word
System, so that ranking results are more accurate.
For example, LSTM model be calculated word sequence " aircraft | | Hua Xing ", " aircraft | | slide ", " aircraft | just
In | flower pattern ", " aircraft | | Huaxing " corresponding first score be respectively as follows: the 980,560,1300,1200, and first score
Score is smaller, indicates that corresponding group of word result is more excellent.
Specifically, above-mentioned word sequence is sequentially input LSTM model, LSTM model judges whether the word sequence of input meets
It is syntax rule, whether clear and coherent, and then give a mark to word sequence, obtain corresponding first score of the word sequence, and first
Point smaller, expression group word effect is more excellent.Wherein, it described first is scored at according to the corresponding prediction word of word each in the word sequence
Score obtains, and predicts that the score of word can take logarithm to obtain for the probability value to prediction word, and probability value is usually between 0~1,
Its bigger corresponding logarithm of probability value is smaller, therefore corresponding first score of word sequence is smaller.
Step S23, according to first score, corresponding group of word result of the word sequence is ranked up.
Finally, being ranked up according to the first score to above-mentioned group of word result, available following ranking results: " aircraft is just
Sliding ", " aircraft Hua Xing ", " aircraft is just in Huaxing ", " aircraft is just in flower pattern ".Input method procedure can will arrange as a result,
User is showed in primary group of word result " aircraft is sliding ", so that show the group word result of user more reasonable,
It is more in line with the actual demand of user.
In an alternative embodiment of the invention, it is described according to shot and long term memory network model to described group of word result into
Row sequence, can specifically include:
Step S31, according to the shot and long term memory network model, first of word sequence corresponding to described group of word result is determined
Score；
Step S32, according to multi-component grammar model, corresponding second score of described group of word result is determined；
Step S33, according to first score and second score, described group of word result is ranked up.
In embodiments of the present invention, when a group word result meets predetermined order condition, it can be used alone LSTM model pair
Described group of word result is ranked up, so that the group word result for coming front can embody the dependence between remote word,
The group word result for being presented to user is more reasonable.Optionally, the embodiment of the present invention can also be according to multi-component grammar model and described
Shot and long term memory network model carries out comprehensive marking to a group word result, to further increase group reasonability of word result and accurate
Property.
Specifically, in corresponding group of word result of the available input string organize word paths ordering top N group word as a result,
Wherein, N is the natural number more than or equal to 1, and is segmented to described group of word result, to obtain word sequence, according to LSTM mould
Type determines corresponding first score of the word sequence, has between corresponding with the word sequence group of word result of first score
There is corresponding relationship；And the second score of the N number of group of word result is determined according to n-gram model, then obtained according to described first
Point and second score, the N number of group of word result is ranked up.
Optionally, the embodiment of the present invention can calculate the N number of group of word according to first score and second score
As a result corresponding third score is ranked up the N number of group of word result according to third score.For example, can be by following public
Formula calculates the corresponding third score com_score of the N number of group of word result:
Com_score=w1 × lstm_score+w2 × ngram_score (1)
In formula (1), w1 is the corresponding weight of LSTM model, and w2 is the corresponding weight of n-gram model, lstm_
Score is corresponding first score of group word result determined according to LSTM model, and ngram_score is true according to n-gram model
Second score of fixed group word result.It is appreciated that in practical applications, those skilled in the art can be clever according to the actual situation
The specific value living that the w1 and w2 is arranged, the embodiment of the present invention are without restriction for the specific value of w1 and w2.
To sum up, the embodiment of the present invention is after obtaining corresponding group of word result of input string, according to LSTM model to described group
Word result is ranked up；Since LSTM model is a kind of time recurrent neural network, can obtain between remote word according to
The relationship of relying, for example, for input string " feijizhengzaihuaxing ", than " aircraft " and " China between " aircraft " and " sliding "
Dependence is had more between star ", therefore, according to LSTM model to a group word result " aircraft Hua Xing ", " aircraft is being slided
Row ", " aircraft is just in flower pattern " are ranked up, and a group word result " aircraft is sliding " can be ranked the first, input method procedure root
According to the ranking results, described group of word is shown as a result, can be to user's displaying group word result " aircraft is sliding ", so that in long word
Under the input scene of item, shows the group word result of user more reasonable, be more in line with the actual demand of user.
Embodiment of the method two
Referring to Fig. 2, a kind of step flow chart of input method embodiment two of the invention is shown, can specifically include:
Step 201, the input string for receiving user；
Wherein, the input string be specifically as follows Pinyin Input string, five input strings, English input string, stroke input string,
Handwriting input string and voice input string etc..
It whether there is the corresponding entry of the input string in step 202, inquiry dictionary；If it exists, 212 are thened follow the steps；
If it does not exist, 203 are thened follow the steps；
In embodiments of the present invention, the dictionary can specifically include system dictionary, user thesaurus, cell dictionary etc., this
Inventive embodiments are without restriction for the concrete type of the dictionary.In addition, the dictionary can store for user client
Local dictionary, alternatively, can also be deposited for storage online dictionary in the server, the embodiment of the present invention for the dictionary
Storage space is set without restriction.
Step 203 determines corresponding group of word result of the input string；
Specifically, intelligent word can be carried out to the input string according to n-gram model, with the determination input string pair
The group word answered is as a result, described group of word result can be one or more.
Step 204 judges whether corresponding group of word result of the input string meets predetermined order condition；If satisfied, then holding
Row step 205；If not satisfied, thening follow the steps 211；
Intelligent word is being carried out to the input string according to n-gram model, is determining corresponding group of word result of the input string
Later, described group of word result can be ranked up according to group word path, the optimal group word in the group word path to be ranked the first
As a result.If there is binary crelation between adjacent word in the optimal group word result in this group of word path, it is determined that the input string pair
The group word result answered is unsatisfactory for predetermined order condition；Otherwise, it determines corresponding group of word result of the input string meets predetermined order
Condition.
Step 205 obtains the group word result that word paths ordering top N is organized in described corresponding group of word result of input string；
Step 206 segments the group word result of the top N, to obtain word sequence；
Step 207, according to LSTM model, determine corresponding first score of the word sequence；
Step 208, according to n-gram model, determine corresponding second score of group word result of the top N；
Step 209, according to first score and second score, determine that the group word result of the top N is corresponding
Third score；
Specifically, the corresponding third score of group word result of the top N can be calculated according to formula (1).
Step 210, according to the third score, the group word result of the top N is ranked up and is shown；
Specifically, the group word result of the top N can be ranked up, and the row of displaying according to the third score
In the first group word result.
Step 211 shows and organizes the optimal group word result in word path according to what n-gram model determined；
Step 212 shows the corresponding entry of the input string.
To sum up, the embodiment of the present invention is after obtaining corresponding group of word result of input string, according to LSTM model to described group
Word result is ranked up；Since LSTM model is a kind of time recurrent neural network, can obtain between remote word according to
The relationship of relying, so that showing the group word result of user more reasonable under the input scene of long entry, being more in line with the reality of user
Border demand.
Using example one
Step A1, the input string of user is received；
For example, the input string of received user is " feijizhengzaihuaxing ".
Step A2, corresponding group of word result of the input string is determined；
Specifically, syllabification is carried out to input string " feijizhengzaihuaxing ", it is corresponding obtains the input string
Phonetic network carries out the building of WordNet according to n-gram model and the phonetic network, and obtains group word result and a group
Corresponding group of word path score of word result, and acquisition comes first 3 group word results and corresponding group of word path score has
Body is as follows, wherein the smaller expression group word path of score is more excellent:
1) aircraft (2613) Hua Xing
2) aircraft is sliding (2685)
3) aircraft is just in flower pattern (2839)
Step A3, the group word result for coming first 3 is segmented；
Specifically, above-mentioned group of word result " aircraft Hua Xing ", " aircraft is sliding ", " aircraft is just in flower pattern " are carried out
Participle, available following word sequence:
1) aircraft | | Hua Xing
2) aircraft | | it slides
3) aircraft | | flower pattern
Step A4, corresponding first score of the word sequence is determined according to LSTM model；
For example, determining that corresponding first score of above-mentioned word sequence is specific as follows:
Aircraft | | Hua Xing (980)
Aircraft | | slide (560)
Aircraft | | flower pattern (1300)
Step A5, according to the first score and the second score (group word path score), the corresponding third of calculating group word result is obtained
Point, it is ranked up according to third score；
Specifically, can be calculated according to formula (1) above-mentioned group of word result " aircraft Hua Xing ", " aircraft is sliding ",
" aircraft is just in flower pattern " corresponding third score, it is assumed that w1=0.65, w2=0.35, then according to the third score pair being calculated
Above-mentioned word segmentation result is resequenced, and obtained ranking results are specific as follows:
1) aircraft is sliding (1304)
2) aircraft (1552) Hua Xing
3) aircraft is just in flower pattern (1839)
The group word result that A6, displaying rank the first (aircraft is sliding).
Using example two
Step A1, the input string of user is received；
For example, the input string of the user obtained is " henhaobug ".
Step A2, corresponding group of word result of the input string is determined；
Specifically, syllabification is carried out to input string " henhaobug ", obtains the corresponding phonetic network of the input string, root
The building of WordNet is carried out according to n-gram model and the phonetic network, and it is corresponding with group word result to obtain group word result
Group word path score, and obtain come first 4 group word result and corresponding group of word path score it is specific as follows, wherein
The smaller expression group word path of score is more excellent:
1) fine bug (3500)
2) very well only (3620)
It 3) very well should not (3710)
4) very well inexpensive (3800)
In a kind of application example of the invention, it is assumed that " fine " corresponding unitary word frequency is 800, " bug " corresponding one
First word frequency is 2100, and " very well | bug " corresponding binary word frequency is 600, and three is added corresponding group of available " fine bug "
Word path score is 3500；Wherein, word frequency can refer to the number that a word occurs in corpus, and the embodiment of the present invention is for specific
Word frequency it is without restriction.Certainly, in practical applications, described group of word path score can also be calculated according to Bayesian formula
It arrives.
Step A3, the group word result for coming first 3 is segmented；
Specifically, above-mentioned group of word result is segmented, available following word sequence:
Very well | bug
Very well | but
Very well | it should not
Very well | it is inexpensive
Step A4, corresponding first score of the word sequence is determined according to LSTM model；
Specifically, above-mentioned word sequence is sequentially input LSTM model, LSTM model judges whether the word sequence of input meets
It is syntax rule, whether clear and coherent, and then give a mark to word sequence, obtain corresponding first score of the word sequence, and first
Point smaller, expression group word effect is more excellent.
For example, determining that corresponding first score of above-mentioned word sequence is specific as follows:
Very well | bug (1200)
Very well | but (500)
Very well | it should not (600)
Very well | inexpensive (650)
Step A5, according to the first score and the second score (group word path score), the corresponding third of calculating group word result is obtained
Point, it is ranked up according to third score；
Specifically, the corresponding third score of above-mentioned group of word result can be calculated according to formula (1), it is assumed that w1=1, w2=1,
It is then resequenced according to the third score being calculated to above-mentioned word segmentation result, obtained ranking results are specific as follows:
1) very well only (4120)
It 2) very well should not (4310)
3) very well inexpensive (4450)
3) fine bug (4700)
The group word result (very well only) that A6, displaying rank the first.
Installation practice
Referring to Fig. 3, a kind of structural block diagram of input unit embodiment of the invention is shown, can specifically include:
Group word result determining module 301, for determining corresponding group of word result of input string；
Group word sort result module 302, for being ranked up according to shot and long term memory network model to described group of word result；
First display module 303, for being shown to the group word result after sequence.
Optionally, described group of word sort result module 302, can specifically include:
Submodule is segmented, for segmenting to described group of word result, to obtain word sequence；
First score computational submodule, for determining that the word sequence is corresponding according to the shot and long term memory network model
The first score；
First sorting sub-module, for arranging corresponding group of word result of the word sequence according to first score
Sequence.
Optionally, the first score computational submodule, can specifically include:
Predicting unit, for determining that the word in the word sequence is corresponding pre- according to the shot and long term memory network model
Survey the probability value of word and the prediction word；
Computing unit, for according to word, the institute's predicate corresponding prediction word and the prediction word in the word sequence
Probability value determines corresponding first score of the word sequence.
Optionally, described group of word sort result module 302, can specifically include:
First score computational submodule, for determining described group of word result institute according to the shot and long term memory network model
First score of corresponding word sequence；
Second score computational submodule, for determining that described group of word result corresponding second is obtained according to multi-component grammar model
Point；
Second sorting sub-module, for being carried out to described group of word result according to first score and second score
Sequence.
Optionally, described group of word sort result module 302, can specifically include:
Third sorting sub-module, if meeting predetermined order condition, basis for corresponding group of word result of the input string
Shot and long term memory network model is ranked up described group of word result.
Optionally, described device can also include: condition judgment module, for judging corresponding group of word knot of the input string
Whether fruit meets predetermined order condition；The condition judgment module, can specifically include:
Optimal result determines submodule, for determining that group word path is optimal from corresponding group of word result of the input string
Group word result；
Judging submodule, if between word adjacent in the group word result optimal for described group of word path there is binary to close
System, it is determined that corresponding group of word result of the input string is unsatisfactory for predetermined order condition；Otherwise, it determines the input string is corresponding
Group word result meets predetermined order condition.
Optionally, described device can also include:
Second display module is shown if being unsatisfactory for predetermined order condition for corresponding group of word result of the input string
The optimal group word result in described group of word path.
For device embodiment, since it is basically similar to the method embodiment, related so being described relatively simple
Place illustrates referring to the part of embodiment of the method.
All the embodiments in this specification are described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiments, the same or similar parts between the embodiments can be referred to each other.
About the device in above-described embodiment, wherein modules execute the concrete mode of operation in related this method
Embodiment in be described in detail, no detailed explanation will be given here.
The embodiment of the invention also discloses a kind of devices for input, include memory and one or one
Above program, one of them perhaps more than one program be stored in memory and be configured to by one or one with
It includes the instruction for performing the following operation that upper processor, which executes the one or more programs:
Determine corresponding group of word result of input string；
Described group of word result is ranked up according to shot and long term memory network model；
Group word result after sequence is shown.
Optionally, described that described group of word result is ranked up according to shot and long term memory network model, it can specifically include:
Described group of word result is segmented, to obtain word sequence；
According to the shot and long term memory network model, corresponding first score of the word sequence is determined；
According to first score, corresponding group of word result of the word sequence is ranked up.
Optionally, described that corresponding first score of the word sequence is determined according to the shot and long term memory network model, tool
Body may include:
According to the shot and long term memory network model, the corresponding prediction word of word in the word sequence and described is determined
Predict the probability value of word；
According to the probability value of the corresponding prediction word of word, institute's predicate and the prediction word in the word sequence, institute is determined
Corresponding first score of predicate sequence.
Optionally, described that described group of word result is ranked up according to shot and long term memory network model, it can specifically include:
According to the shot and long term memory network model, the first score of word sequence corresponding to described group of word result is determined；
According to multi-component grammar model, corresponding second score of described group of word result is determined；
According to first score and second score, described group of word result is ranked up.
Optionally, described that described group of word result is ranked up according to shot and long term memory network model, it can specifically include:
If corresponding group of word result of the input string meets predetermined order condition, according to shot and long term memory network model pair
Described group of word result is ranked up.
Optionally, described device is also configured to execute one or one by one or more than one processor
Procedure above includes the instruction for performing the following operation:
The optimal group word result in group word path is determined from corresponding group of word result of the input string；
If there is binary crelation between adjacent word in the optimal group word result in described group of word path, it is determined that the input
Corresponding group of word result of going here and there is unsatisfactory for predetermined order condition；It is preset otherwise, it determines corresponding group of word result of the input string meets
Sort criteria.
Optionally, the processor is also configured to execute one or one by one or more than one processor
A procedure above includes the instruction for performing the following operation:
If corresponding group of word result of the input string is unsatisfactory for predetermined order condition, show that described group of word path is optimal
Group word result.
Fig. 4 is a kind of block diagram of device 800 for input shown according to an exemplary embodiment.For example, device 800
It can be mobile phone, computer, digital broadcasting terminal, messaging device, game console, tablet device, Medical Devices,
Body-building equipment, personal digital assistant etc..
Referring to Fig. 4, device 800 may include following one or more components: processing component 802, memory 804, power supply
Component 806, multimedia component 808, audio component 810, the interface 812 of input/output (I/O), sensor module 814, and
Communication component 816.
The integrated operation of the usual control device 800 of processing component 802, such as with display, telephone call, data communication, phase
Machine operation and record operate associated operation.Processing element 802 may include that one or more processors 820 refer to execute
It enables, to perform all or part of the steps of the methods described above.In addition, processing component 802 may include one or more modules, just
Interaction between processing component 802 and other assemblies.For example, processing component 802 may include multi-media module, it is more to facilitate
Interaction between media component 808 and processing component 802.
Memory 804 is configured as storing various types of data to support the operation in equipment 800.These data are shown
Example includes the instruction of any application or method for operating on device 800, contact data, and telephone book data disappears
Breath, picture, video etc..Memory 804 can be by any kind of volatibility or non-volatile memory device or their group
It closes and realizes, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM) is erasable to compile
Journey read-only memory (EPROM), programmable read only memory (PROM), read-only memory (ROM), magnetic memory, flash
Device, disk or CD.
Power supply module 806 provides electric power for the various assemblies of device 800.Power supply module 806 may include power management system
System, one or more power supplys and other with for device 800 generate, manage, and distribute the associated component of electric power.
Multimedia component 808 includes the screen of one output interface of offer between described device 800 and user.One
In a little embodiments, screen may include liquid crystal display (LCD) and touch panel (TP).If screen includes touch panel, screen
Curtain may be implemented as touch screen, to receive input signal from the user.Touch panel includes one or more touch sensings
Device is to sense the gesture on touch, slide, and touch panel.The touch sensor can not only sense touch or sliding action
Boundary, but also detect duration and pressure associated with the touch or slide operation.In some embodiments, more matchmakers
Body component 808 includes a front camera and/or rear camera.When equipment 800 is in operation mode, such as screening-mode or
When video mode, front camera and/or rear camera can receive external multi-medium data.Each front camera and
Rear camera can be a fixed optical lens system or have focusing and optical zoom capabilities.
Audio component 810 is configured as output and/or input audio signal.For example, audio component 810 includes a Mike
Wind (MIC), when device 800 is in operation mode, when such as call mode, recording mode, and voice recognition mode, microphone is matched
It is set to reception external audio signal.The received audio signal can be further stored in memory 804 or via communication set
Part 816 is sent.In some embodiments, audio component 810 further includes a loudspeaker, is used for output audio signal.
I/O interface 812 provides interface between processing component 802 and peripheral interface module, and above-mentioned peripheral interface module can
To be keyboard, click wheel, button etc..These buttons may include, but are not limited to: home button, volume button, start button and lock
Determine button.
Sensor module 814 includes one or more sensors, and the state for providing various aspects for device 800 is commented
Estimate.For example, sensor module 814 can detecte the state that opens/closes of equipment 800, and the relative positioning of component, for example, it is described
Component is the display and keypad of device 800, and sensor module 814 can be with 800 1 components of detection device 800 or device
Position change, the existence or non-existence that user contacts with device 800,800 orientation of device or acceleration/deceleration and device 800
Temperature change.Sensor module 814 may include proximity sensor, be configured to detect without any physical contact
Presence of nearby objects.Sensor module 814 can also include optical sensor, such as CMOS or ccd image sensor, at
As being used in application.In some embodiments, which can also include acceleration transducer, gyro sensors
Device, Magnetic Sensor, pressure sensor or temperature sensor.
Communication component 816 is configured to facilitate the communication of wired or wireless way between device 800 and other equipment.Device
800 can access the wireless network based on communication standard, such as WiFi, 2G or 3G or their combination.In an exemplary implementation
In example, communication component 816 receives broadcast singal or broadcast related information from external broadcasting management system via broadcast channel.
In one exemplary embodiment, the communication component 816 further includes near-field communication (NFC) module, to promote short range communication.Example
Such as, NFC module can be based on radio frequency identification (RFID) technology, Infrared Data Association (IrDA) technology, ultra wide band (UWB) technology,
Bluetooth (BT) technology and other technologies are realized.
In the exemplary embodiment, device 800 can be believed by one or more application specific integrated circuit (ASIC), number
Number processor (DSP), digital signal processing appts (DSPD), programmable logic device (PLD), field programmable gate array
(FPGA), controller, microcontroller, microprocessor or other electronic components are realized, for executing the above method.
In the exemplary embodiment, a kind of non-transitorycomputer readable storage medium including instruction, example are additionally provided
It such as include the memory 804 of instruction, above-metioned instruction can be executed by the processor 820 of device 800 to complete the above method.For example,
The non-transitorycomputer readable storage medium can be ROM, random access memory (RAM), CD-ROM, tape, floppy disk
With optical data storage devices etc..
A kind of non-transitorycomputer readable storage medium, when the instruction in the storage medium is by device (terminal or clothes
Be engaged in device) processor execute when, enable a device to execute input method shown in FIG. 1, which comprises determine input string
Corresponding group of word result；Described group of word result is ranked up according to shot and long term memory network model；To the group word knot after sequence
Fruit is shown.
Fig. 5 is the structural schematic diagram of server in the embodiment of the present invention.The server 1900 can be different because of configuration or performance
And generate bigger difference, may include one or more central processing units (central processing units,
CPU) 1922 (for example, one or more processors) and memory 1932, one or more storage application programs
1942 or data 1944 storage medium 1930 (such as one or more mass memory units).Wherein, memory 1932
It can be of short duration storage or persistent storage with storage medium 1930.Be stored in storage medium 1930 program may include one or
More than one module (diagram does not mark), each module may include to the series of instructions operation in server.Further
Ground, central processing unit 1922 can be set to communicate with storage medium 1930, and storage medium 1930 is executed on server 1900
In series of instructions operation.
Server 1900 can also include one or more power supplys 1926, one or more wired or wireless nets
Network interface 1950, one or more input/output interfaces 1958, one or more keyboards 1956, and/or, one or
More than one operating system 1941, such as Windows ServerTM, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM
Etc..
Those skilled in the art after considering the specification and implementing the invention disclosed here, will readily occur to of the invention its
Its embodiment.The present invention is directed to cover any variations, uses, or adaptations of the invention, these modifications, purposes or
Person's adaptive change follows general principle of the invention and including the undocumented common knowledge in the art of the disclosure
Or conventional techniques.The description and examples are only to be considered as illustrative, and true scope and spirit of the invention are by following
Claim is pointed out.
It should be understood that the present invention is not limited to the precise structure already described above and shown in the accompanying drawings, and
And various modifications and changes may be made without departing from the scope thereof.The scope of the present invention is limited only by the attached claims
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Above to a kind of input method provided by the present invention, a kind of input unit and a kind of device for input, into
It has gone and has been discussed in detail, used herein a specific example illustrates the principle and implementation of the invention, the above implementation
The explanation of example is merely used to help understand method and its core concept of the invention；Meanwhile for the general technology people of this field
Member, according to the thought of the present invention, there will be changes in the specific implementation manner and application range, in conclusion this explanation
Book content should not be construed as limiting the invention.
Claims (10)
1. a kind of input method, which is characterized in that the described method includes:
Determine corresponding group of word result of input string；
Described group of word result is ranked up according to shot and long term memory network model；
Group word result after sequence is shown.
2. the method according to claim 1, wherein it is described according to shot and long term memory network model to described group of word
As a result it is ranked up, comprising:
Described group of word result is segmented, to obtain word sequence；
According to the shot and long term memory network model, corresponding first score of the word sequence is determined；
According to first score, corresponding group of word result of the word sequence is ranked up.
3. according to the method described in claim 2, it is characterized in that, described according to the shot and long term memory network model, determination
Corresponding first score of the word sequence, comprising:
According to the shot and long term memory network model, the corresponding prediction word of word and the prediction in the word sequence are determined
The probability value of word；
According to the probability value of the corresponding prediction word of word, institute's predicate and the prediction word in the word sequence, institute's predicate is determined
Corresponding first score of sequence.
4. the method according to claim 1, wherein it is described according to shot and long term memory network model to described group of word
As a result it is ranked up, comprising:
According to the shot and long term memory network model, the first score of word sequence corresponding to described group of word result is determined；
According to multi-component grammar model, corresponding second score of described group of word result is determined；
According to first score and second score, described group of word result is ranked up.
5. the method according to claim 1, wherein it is described according to shot and long term memory network model to described group of word
As a result it is ranked up, comprising:
If corresponding group of word result of the input string meets predetermined order condition, according to shot and long term memory network model to described
Group word result is ranked up.
6. according to the method described in claim 5, it is characterized in that, judging corresponding group of word of the input string as follows
As a result whether meet predetermined order condition:
The optimal group word result in group word path is determined from corresponding group of word result of the input string；
If there is binary crelation between adjacent word in the optimal group word result in described group of word path, it is determined that the input string pair
The group word result answered is unsatisfactory for predetermined order condition；Otherwise, it determines corresponding group of word result of the input string meets predetermined order
Condition.
7. according to claim 1 to any method in 6, which is characterized in that the method also includes:
If corresponding group of word result of the input string is unsatisfactory for predetermined order condition, the optimal group word in described group of word path is shown
As a result.
8. a kind of input unit characterized by comprising
Group word result determining module, for determining corresponding group of word result of input string；
Group word sort result module, for being ranked up according to shot and long term memory network model to described group of word result；
First display module, for being shown to the group word result after sequence.
9. a kind of device for input, which is characterized in that it include memory and one or more than one program,
Perhaps more than one program is stored in memory and is configured to be executed by one or more than one processor for one of them
The one or more programs include the instruction for performing the following operation:
Determine corresponding group of word result of input string；
Described group of word result is ranked up according to shot and long term memory network model；
Group word result after sequence is shown.
10. a kind of machine readable media is stored thereon with instruction, when executed by one or more processors, so that device is held
Input method of the row as described in one or more in claim 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810172598.0A CN110221704A (en) | 2018-03-01 | 2018-03-01 | A kind of input method, device and the device for input |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810172598.0A CN110221704A (en) | 2018-03-01 | 2018-03-01 | A kind of input method, device and the device for input |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110221704A true CN110221704A (en) | 2019-09-10 |
Family
ID=67821960
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810172598.0A Pending CN110221704A (en) | 2018-03-01 | 2018-03-01 | A kind of input method, device and the device for input |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110221704A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112987940A (en) * | 2021-04-27 | 2021-06-18 | 广州智品网络科技有限公司 | Input method and device based on sample probability quantization and electronic equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101644961A (en) * | 2009-08-14 | 2010-02-10 | 北京搜狗科技发展有限公司 | Encoded string sequencing method, device and character input method and device |
| US20100125459A1 (en) * | 2008-11-18 | 2010-05-20 | Nuance Communications, Inc. | Stochastic phoneme and accent generation using accent class |
| CN106774970A (en) * | 2015-11-24 | 2017-05-31 | 北京搜狗科技发展有限公司 | The method and apparatus being ranked up to the candidate item of input method |
| US20170255278A1 (en) * | 2014-10-16 | 2017-09-07 | Touchtype Ltd. | Text prediction integration |
| CN107305438A (en) * | 2016-04-21 | 2017-10-31 | 北京搜狗科技发展有限公司 | The sort method and device of candidate item, the device sorted for candidate item |
| CN107621891A (en) * | 2017-09-28 | 2018-01-23 | 北京新美互通科技有限公司 | A kind of text entry method, device and electronic equipment |
-
2018
- 2018-03-01 CN CN201810172598.0A patent/CN110221704A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100125459A1 (en) * | 2008-11-18 | 2010-05-20 | Nuance Communications, Inc. | Stochastic phoneme and accent generation using accent class |
| CN101644961A (en) * | 2009-08-14 | 2010-02-10 | 北京搜狗科技发展有限公司 | Encoded string sequencing method, device and character input method and device |
| US20170255278A1 (en) * | 2014-10-16 | 2017-09-07 | Touchtype Ltd. | Text prediction integration |
| CN106774970A (en) * | 2015-11-24 | 2017-05-31 | 北京搜狗科技发展有限公司 | The method and apparatus being ranked up to the candidate item of input method |
| CN107305438A (en) * | 2016-04-21 | 2017-10-31 | 北京搜狗科技发展有限公司 | The sort method and device of candidate item, the device sorted for candidate item |
| CN107621891A (en) * | 2017-09-28 | 2018-01-23 | 北京新美互通科技有限公司 | A kind of text entry method, device and electronic equipment |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112987940A (en) * | 2021-04-27 | 2021-06-18 | 广州智品网络科技有限公司 | Input method and device based on sample probability quantization and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN106708282B (en) | A kind of recommended method and device, a kind of device for recommendation | |
| CN106774970A (en) | The method and apparatus being ranked up to the candidate item of input method | |
| CN107221330A (en) | Punctuate adding method and device, the device added for punctuate | |
| CN109243430A (en) | A kind of audio recognition method and device | |
| CN107305438A (en) | The sort method and device of candidate item, the device sorted for candidate item | |
| CN109961791A (en) | A kind of voice information processing method, device and electronic equipment | |
| CN109144285A (en) | A kind of input method and device | |
| CN107346182A (en) | A kind of method for building user thesaurus and the device for building user thesaurus | |
| CN111984749A (en) | Method and device for ordering interest points | |
| CN108874827A (en) | A kind of searching method and relevant apparatus | |
| CN110069624A (en) | Text handling method and device | |
| CN109101505A (en) | A kind of recommended method, recommendation apparatus and the device for recommendation | |
| CN110929176A (en) | Information recommendation method and device and electronic equipment | |
| CN110110207A (en) | A kind of information recommendation method, device and electronic equipment | |
| CN110110204A (en) | A kind of information recommendation method, device and the device for information recommendation | |
| CN108803890A (en) | A kind of input method, input unit and the device for input | |
| CN107422872A (en) | A kind of input method, device and the device for input | |
| CN109521888A (en) | A kind of input method, device and medium | |
| CN108628461A (en) | A kind of input method and device, a kind of method and apparatus of update dictionary | |
| CN110221704A (en) | A kind of input method, device and the device for input | |
| CN111639209A (en) | Book content searching method, terminal device and storage medium | |
| CN109901726A (en) | A kind of candidate word generation method, device and the device generated for candidate word | |
| CN110119461A (en) | A kind of processing method and processing device of query information | |
| CN110858099B (en) | Candidate word generation method and device | |
| CN109799916A (en) | A kind of candidate item association method and device |
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 |