CN108334601A - Song recommendations method, apparatus and storage medium based on label topic model - Google Patents
Song recommendations method, apparatus and storage medium based on label topic model Download PDFInfo
- Publication number
- CN108334601A CN108334601A CN201810097213.9A CN201810097213A CN108334601A CN 108334601 A CN108334601 A CN 108334601A CN 201810097213 A CN201810097213 A CN 201810097213A CN 108334601 A CN108334601 A CN 108334601A
- Authority
- CN
- China
- Prior art keywords
- song
- theme
- label
- probability distribution
- list
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/63—Querying
- G06F16/635—Filtering based on additional data, e.g. user or group profiles
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/60—Information retrieval; Database structures therefor; File system structures therefor of audio data
- G06F16/68—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
- G06F16/686—Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually using information manually generated, e.g. tags, keywords, comments, title or artist information, time, location or usage information, user ratings
Abstract
The embodiment of the invention discloses a kind of song recommendations method, apparatus and storage medium based on label topic model, this method includes:The single set of song is obtained, singly set includes that multiple songs are single to the song, and the song includes singly subject information and number of songs;The single tag set of the song is built according to the subject information, the tag set includes at least one theme label;Theme label in the tag set is distributed into the song in the song list;Obtain the new theme probability distribution of the song;The target topic label that song is assigned in the song list is determined according to the new theme probability distribution;Corresponding song recommendations list is generated according to the target topic label that song is assigned in the song list, and song recommendations are carried out based on the song recommendations list.This programme can improve the accuracy of song recommendations.
Description
Technical field
The present invention relates to multimedia technology fields, and in particular to a kind of song recommendations method based on label topic model,
Device and storage medium.
Background technology
With the fast development of network, daily life increasingly be unable to do without network, listens song by network, sees video
And see news etc. and become daily life custom, by taking music as an example, increase with the explosivity of music data
Long, user is increasingly difficult to select the music oneself liked in more music datas of comforming, therefore, actively recommends to user interested
Music, become for a kind of feasible and efficient scheme.
Suggested design primarily now is with collaborative filtering and LDA (Latent Dirichlet Allocation, theme
Model) based on, wherein collaborative filtering is the shadow listened an old song form for data source as input, be easy by flowing water with user
Ring, by collaborative filtering method be pushed out come song be often partial to hot topic;Although and traditional LDA models can be used
The theme distribution at family and the list of songs of theme, but LDA models are highly susceptible to the influence of language material training data and cause
There is deviation when training, causes recommendation results inaccurate.
Invention content
An embodiment of the present invention provides a kind of song recommendations method, apparatus and storage medium based on label topic model,
The accuracy of song recommendations can be greatly improved.
An embodiment of the present invention provides a kind of song recommendations methods based on label topic model, including:
The single set of song is obtained, singly set includes that multiple songs are single to the song, and the song includes singly subject information;
The single tag set of the song is built according to the subject information, the tag set includes at least one theme mark
Label;
Theme label in the tag set is distributed into the song in the song list;
The new theme probability distribution of the song is obtained, the new theme probability distribution includes that the song is currently assigned to
The probability distribution of each theme label;
The target topic label that song is assigned in the song list is determined according to the new theme probability distribution;
Corresponding song recommendations list is generated according to the target topic label that song is assigned in the song list, and is based on institute
It states song recommendations list and carries out song recommendations.
The embodiment of the present invention additionally provides a kind of song recommendations device based on label topic model, including:
First acquisition unit, for obtaining the single set of song, singly set includes that multiple songs are single to the song, and the song includes singly master
Inscribe information;
Construction unit, for building the single tag set of the song according to the subject information, the tag set includes
At least one theme label;
Allocation unit, for the theme label in the tag set to be distributed to the song in the song list;
Second acquisition unit, the new theme probability distribution for obtaining the song, the new theme probability distribution include
The song is currently assigned to the probability distribution of each theme label;
Determination unit, for determining the target topic that is assigned to of song in the song list according to the new theme probability distribution
Label;
Recommendation unit, for generating corresponding song recommendations according to the target topic label that song is assigned in the song list
List, and song recommendations are carried out based on the song recommendations list.
In addition, the embodiment of the present invention also provides a kind of storage medium, it is stored with processor-executable instruction, the processing
Device provides such as above-mentioned song recommendations method by executing described instruction.
The embodiment of the present invention obtains the single set of song first, and singly set includes that multiple songs are single to the song, and the song includes singly master
Information is inscribed, the single tag set of the song is then built according to the subject information, the tag set includes at least one theme
Theme label in the tag set is subsequently distributed to the song in the song list, then obtains the song by label
New theme probability distribution, the target topic mark that is assigned to of song in the song list is determined according to the new theme probability distribution
Label finally generate corresponding song recommendations list according to the target topic label that song is assigned in the song list, and are based on institute
It states song recommendations list and carries out song recommendations.I.e. the embodiment of the present invention is by using the single theme label of song, by unsupervised LDA
Model conversation is to have the topic model of supervision to be trained, and final song recommendations list is generated, to improve song recommendations
Accuracy.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly described, it should be apparent that, drawings in the following description are only some embodiments of the invention, for
For those skilled in the art, without creative efforts, it can also be obtained according to these attached drawings other attached
Figure.
Fig. 1 is the schematic diagram of a scenario of song recommendations method provided in an embodiment of the present invention;
Fig. 2 is the flow diagram of song recommendations method provided in an embodiment of the present invention;
Fig. 3 is the formation schematic diagram of tag set provided in an embodiment of the present invention;
Fig. 4 is theme song distribution schematic diagram provided in an embodiment of the present invention;
Fig. 5 is the probability graph model of label topic model provided in an embodiment of the present invention;
Fig. 6 is that training data provided in an embodiment of the present invention summarizes schematic diagram;
Fig. 7 is gibbs sampler process schematic provided in an embodiment of the present invention;
Fig. 8 is the structural schematic diagram of song recommendations device provided in an embodiment of the present invention;
Fig. 9 is another structural schematic diagram of song recommendations device provided in an embodiment of the present invention;
Figure 10 is the structural schematic diagram of server provided in an embodiment of the present invention.
Specific implementation mode
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 describes, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, the every other implementation that those skilled in the art are obtained without creative efforts
Example, shall fall within the protection scope of the present invention.
Referring to Fig. 1, which is the schematic diagram of a scenario of song recommendations method provided in an embodiment of the present invention, can be in the scene
Including song recommendations device, which can specifically be integrated in the network equipments such as server, and the server can
With the server cluster being made of several servers or a cloud computing service center.
As shown in Figure 1, may include server a, server b and terminal c in the scene, wherein terminal c can be intelligence
Mobile phone, personal computer etc..For example, server b first obtains the single set of song from server a, singly set includes multiple to the song
Song is single, and the song includes singly subject information.Then, the single tag set of the song, the mark are built according to the subject information
Label set includes at least one theme label.Theme label in the tag set is distributed into the song in the song list.
Subsequently, the new theme probability distribution of the song is obtained, the new theme probability distribution includes that the song is currently assigned to
The probability distribution of each theme label.Finally, determine that song is distributed in the song list according to the new theme probability distribution
The target topic label arrived generates corresponding song recommendations according to the target topic label that song is assigned in the song list and arranges
Table, and song recommendations are carried out based on the song recommendations list.Further, server b can arrange the song recommendations of generation
Table is sent to terminal c, recommends user's selection and listens to.In addition, can also include multiple client in the scene.
The embodiment of the present invention provides a kind of song recommendations method, apparatus and storage medium based on label topic model.
Wherein, the probability graph model of the label topic model is as shown in Figure 5.
Embodiment one,
In embodiments of the present invention, it will be described from the angle of song recommendations device, which specifically may be used
To integrate in the server.
A kind of song recommendations method based on label topic model, including:The single set of song is obtained, singly set includes the song
Multiple songs are single, and the song includes singly subject information;The single tag set of the song, the label are built according to the subject information
Set includes at least one theme label;Theme label in the tag set is distributed into the song in the song list;It obtains
It includes that the song is currently assigned to each master to take the new theme probability distribution of the song, the new theme probability distribution
Inscribe the probability distribution of label;The target topic mark that song is assigned in the song list is determined according to the new theme probability distribution
Label;Corresponding song recommendations list is generated according to the target topic label that song is assigned in the song list, and is based on the song
Bent recommendation list carries out song recommendations.
Refer to Fig. 2, Fig. 2 is the flow diagram of song recommendations method provided in an embodiment of the present invention, and this method can be with
Including:
S101 obtains the single set of song, and singly set includes that multiple songs are single to the song, and the song includes singly subject information.
It trains to come from music libraries it is understood that the song list set is server, server, which needs to choose, to be somebody's turn to do
Song in the single set of song, generates individual character recommendation list, recommends user and listens to and collect etc..
Further, before obtaining the single set of song, can also include:Obtain music libraries in all songs, and to its into
Row training, generates training data, which is the single set of song.
Preferably, which can be made of two parts, first, high-quality song is single, quantity is about 900,000 or so, two
It is artificial song list, the music data which is listened to and collected recently by user forms, and quantity is about 100,000,000 or so, tool
Body is as shown in Figure 6.
S102 builds the single tag set of the song according to the subject information, and the tag set includes at least one
Theme label.
In the present embodiment, each song is single to correspond to a tag set, and the wherein tag set includes one or more
A theme label.As shown in figure 3, extracting the structure such as " Guangdong language ", " classics ", " vicissitudes " and " deep feeling " theme labels from song list
Tag set.
Wherein, sing list in the equal corresponding label set of each song in a theme label, behind can pass through training
A most suitable theme label is chosen for each song.
Theme label in the tag set is distributed to the song in the song list by S103.
In specific implementation process, is sampled based on Gibbs (Gibbs Sample, gibbs), extracted from tag set
Theme label, each song being randomly assigned in the corresponding song list of the tag set, each sub-distribution, each song,
One and only one theme label.
Wherein, it is that will sing singly to regard document as to carry out the key of personalized recommendation using topic model, and structure sings mono- document mould
Then type builds the frequency matrix of document according to the mono- document model of the song.
Optionally, before distributing theme label for each song, each in the song list set can also be sung
It singly regards as with a document, then structure frequency matrix doc is arranged this frequency matrix doc corpus as input several
Statistic and Di Li Cray hyper parameters.
Wherein it is possible to following statistic is set, such as:
nm,z=t:It indicates in document m, the word one for being assigned to theme z shares t.
nz,t=k:Indicate the number that theme k is assigned in word t.
zm,n=k:It indicates in document m, word n is assigned to theme k.
Also, the value of above-mentioned several statistics is initialized as 0.
Preferably, it is based on topic model, it can be from Ge Dan- document models middle school's acquistion to two variable αs and β, variable α
Indicate that the corresponding Di Li Crays hyper parameter of theme, variable β indicate the corresponding Di Li Crays hyper parameter of song.
When being assigned to a theme z for the song t in m-th of song list, matrix is initialized, i.e.,:
zm,t=z
nm,z=nm,z+1
nz,t=nz,t+1
S104, obtains the new theme probability distribution of the song, and the new theme probability distribution includes that the song is current
It is assigned to the probability distribution of each theme label.
In specific implementation process, the new master that information generates the song can be distributed according to the theme label of remaining song
Probability distribution is inscribed, is as follows:
The single theme probability distribution of acquisition of information song and theme song probability distribution are distributed according to the theme label of remaining song.
According to the single theme probability distribution of the song and theme song probability distribution, the new theme probability point of the song is generated
Cloth.
Specifically, song list theme probability distribution θ and theme song probability distribution can be calculated according to following formula
Wherein, wherein nm,zIndicate the number of songs that theme z is assigned in singing list m, the nz,tIndicate song t quilts
It is assigned to the number of theme z.
It should be understood that described distribute the single theme probability distribution of acquisition of information song and master according to the theme label of remaining song
Song probability distribution is inscribed, may include:
Obtain current nm,zAnd nz,tAnd preset α and β;
According to the current n of acquisitionm,zAnd nz,tAnd preset α and β generates the single theme probability distribution of song and theme song is general
Rate is distributed.
Preferably, song list theme probability distribution θ and theme song probability distribution are being generatedIt later, can be according to the song list
Theme probability distribution θ and theme song probability distributionGenerate the new theme probability distribution of the song.
Specifically, the new theme probability distribution of the song can be calculated according to following formula:
Wherein, p (z | d, w) is the vector that a size is K dimensions, total number of topics that wherein K is.
It should be noted that before the new theme probability distribution for generating the song, can also include:
The theme label for distributing to the song before is removed, and updates nm,zAnd nz,t。
I.e.:
nm,z=nm,z-1
nz,t=nz,t-1
S105 determines the target topic label that song is assigned in the song list according to the new theme probability distribution.
It, can also be according to the new theme probability distribution, for one new theme of the song sample, newly in specific implementation process
The sampling of theme is limited in the song and corresponds to inside the single tag set of song.
Specifically, new theme can be generated according to following formula:
znew=label*numpy.random.multinomial (p (z | d, w))
Wherein, label is the 0-1 vectors that a size is k dimensions, and therefore, the result of sampling also will be in original first standard inspection
Sampling, the new theme z obtained after sampling in labelnew, while updating zm,n, nm,z, nz,t, i.e.,:
zm,t=znew
The above process is that a Gibbs (gibbs) samples, and above-mentioned behaviour has been carried out to all songs in all song lists
After work, an iteration is just completed, detailed process is as shown in Figure 7.
It constantly repeats the above process, generally use puzzlement degree perplexity weighs topic model, perplexity's
Calculation formula is:
Perplexity=e-loglikelihood/N
Wherein, N indicates all and sings the number of songs for including singly, and loglikelihood is maximum likelihood, calculation formula
For:
In conclusion new theme solution procedure is:It is given at random in song list when initialization and distributes theme per song, then
Count nm,zAnd nz,t, each round calculating p (z | d, w), that is, the theme distribution of current song is excluded, then according to the master of other songs
Topic distribution estimation current song belongs to the probability distribution of each theme, is subsequently the song sample one according to this probability distribution
A new theme.The theme that next song is constantly updated with same method, until finding song list theme probability distribution θ and theme
Song probability distributionMarkov chain convergence, stop iteration, export to be estimated parameter song list theme probability distribution θ and theme
Song probability distributionThe final theme per song also obtains simultaneously.
Therefore, by aforesaid operations, specific theme song distribution list can be obtained, it can under each theme label
To include a first or number of songs, it is of course also possible to be zero song.
S106 generates corresponding song recommendations list according to the target topic label that song is assigned in the song list, and
Song recommendations are carried out based on the song recommendations list.
In specific implementation process, which can specifically include:
User's theme probability distribution and theme song probability distribution are generated based on the theme label that the song is finally distributed;
Song recommendations are generated according to user's theme probability distribution, theme song probability distribution and default recommendation condition
List, and song recommendations are carried out based on the song recommendations list.
Wherein, which can be minority's song and long-tail song, and the long-tail song i.e. demand are small
With the song of sales volume difference.
For example, the song that suggested design now is recommended is typically all partial heat door, and meet the song of minority's taste, but
It cannot be recommended, so being also difficult to be found by user.And minority's song and long-tail song are can be found that and excavated by this programme
It is bent.The song for recommending out by this programme can be suitble to like the user of minority's school.
It can be seen from the above, the song recommendations method of the embodiment of the present invention, by first obtaining the single set of song, the song is singly gathered
Single including multiple songs, the song includes singly subject information, then builds the single tag set of the song according to the subject information,
The tag set includes at least one theme label, and the theme label in the tag set is subsequently distributed to the song
Song in list then obtains the new theme probability distribution of the song, and the song is determined according to the new theme probability distribution
The target topic label that song is assigned in list finally generates phase according to the target topic label that song is assigned in the song list
The song recommendations list answered, and song recommendations are carried out based on the song recommendations list.I.e. the embodiment of the present invention is by using song
Single theme label, is to have the topic model of supervision to be trained by unsupervised LDA model conversations, generates final song recommendations
List, to improve the accuracy of song recommendations.
Embodiment two,
The key that personalized recommendation is carried out using topic model is will to sing singly to regard document as, and the song sung in list is equivalent to list
Word, each song are singly usually constructed with specific style, such as different schools etc., these styles are exactly the single theme label of song,
Each style (theme label) has specific song below, as shown in Figure 4.
In order to better illustrate method described in above example, the present embodiment will be that document is illustrated song nonoculture
It is bright.
S201 extracts training data.
Wherein, which includes multiple documents, each document includes multiple words and one or more
A theme label.The theme label of each document is extracted, tag set is constituted, is distributed as label theme priori,
When distributing theme label, the theme label in corresponding label set is distributed for each document.
S202 builds frequency matrix.
It is understood that after structure frequency matrix, following statistic can also be set, and be initialized as 0.
nm,z=t:It indicates in document m, the word one for being assigned to theme z shares t.
nz,t=k:Indicate the number that theme k is assigned in word t.
zm,n=k:It indicates in document m, word n is assigned to theme k.
Then, variable α and variable β are set, and variable α is the parameter of the priori Dirichlet distributions of document-theme, variable
β is the parameter of the priori Dirichlet distributions of theme-song.
S203 distributes theme label.
Preferably, it initializes, a theme label, unlike LDA, label topic model is distributed for each word
When initialization, it is not randomly assigned a theme, using the theme label of document as priori data, the word below document instead of
It is randomly assigned in the document tag set, when distributing a theme label z for the word t in m documents, initializes square
Battle array zm,n, nm,z, nz,t, have:
zm,t=z
nm,z=nm,z+1
nz,t=nz,t+1
S204, iteration.
Specifically, iteration each time, redistributes theme, respectively in each text of iteration by gibbs samplings
Each word, it is necessary first to the theme for distributing to the word be removed, and change nm,zAnd nz,tSo that:
nm,z=nm,z-1
nz,t=nz,t-1
Then the distribution of new theme is sought according to following formula:
Wherein:
P (z | d, w) is the vector that a size is K dimensions, and wherein K is total theme label number, when obtain p (z | d, w) to
After amount, so that it may which, to be sampled according to this probability distribution, unlike LDA, label topic model can be by the sampling of data
It is limited in inside corresponding document tag set, i.e.,
znew=label*numpy.random.multinomial (p (z | d, w))
Wherein, label is the 0-1 vectors that a size is k dimensions, and therefore, the result of sampling also will be in original tally set
Sampling, the new theme z obtained after sampling in closingnew, while updating zm,n, nm,z, nz,t:
zm,t=znew
The above process is that a Gibbs is sampled, and after being carried out aforesaid operations to all words in all documents, is just completed
An iteration, process can specifically be indicated with 7 figures.Constantly repeat the process of S204, generally use puzzlement degree
Perplexity weighs topic model, and the calculation formula of perplexity is:
Perplexity=e-loglikelihood/N
Wherein, N indicates that the word number that all documents include, loglikelihood are maximum likelihoods, and calculation formula is:
Wherein θ andPlease refer to above-mentioned expression.
In conclusion sampling process is specially:It gives each word in document to distribute theme at random when initialization, then unites
Count nm,zAnd nz,t, each round calculating p (z | d, w), that is, the theme distribution of current word is excluded, then according to the theme of other words
Distribution estimation current word belongs to the probability distribution of each theme, is subsequently that the word samples one according to this probability distribution
New theme.The theme that next word is constantly updated with same method, until finding document subject matter probability distribution θ and subject word
Probability distributionMarkov chain convergence, stop iteration, export parameter document subject matter probability distribution θ to be estimated and subject word
Probability distributionThe theme of final each word also obtains simultaneously.
The present embodiment is by the way that the sampling of topic model to be limited in the tag set of corresponding document, unsupervised theme
Model, becoming has the topic model of supervision to train.Model can be each according to the label data of priori in initialization
A word distributes a relatively reasonable theme, and can gradually correct mistake in the training process so that finally obtained number
According to more accurate.
Embodiment three,
In order to preferably implement above method, the embodiment of the present invention also provides a kind of song based on label topic model and pushes away
Device is recommended, which can specifically be integrated in the equipment such as server, such as service server, and wherein noun contains
Adopted identical with above-mentioned song recommendations method, specific implementation details can be with the explanation in reference method embodiment.
For example, as shown in figure 8, the song recommendations device may include first acquisition unit 301, construction unit 302, distribution
Unit 303, second acquisition unit 304, determination unit 305 and recommendation unit 306 are as follows:
(1) first acquisition unit 301;
First acquisition unit 301, for obtaining the single set of song, singly set includes that multiple songs are single to the song, and the song singly wraps
Include subject information.
Wherein, which is combined into the training data extracted from server a, and the training data is by two part structures
At first, high-quality song is single, quantity is about 900,000 or so;Second is that artificial song list, quantity is probably 100,000,000 or so, the artificial song list
The music data listened to and collected recently by user forms, specific as shown in Figure 6.
(2) construction unit 302;
Construction unit 302 builds the single tag set of the song according to the subject information, and the tag set includes extremely
A few theme label.
In the present embodiment, each song is single to correspond to a tag set, and the wherein tag set includes one or more
A theme label.As shown in figure 3, extracting the structure such as " Guangdong language ", " classics ", " vicissitudes " and " deep feeling " theme labels from song list
Tag set.
Wherein, sing list in the equal corresponding label set of each song in a theme label, behind can pass through training
A most suitable theme label is chosen for each song.
(3) allocation unit 303;
Allocation unit 303, for the theme label in the tag set to be distributed to the song in the song list.
It is understood that the theme allocation unit 303, is specifically used for for each song distribution in each song list
Theme, wherein when carrying out multinomial distribution distribution theme label, the sampling of theme label is limited in the label of priori
Portion so that unsupervised topic model becomes the topic model for having supervision.
(4) second acquisition unit 304;
Second acquisition unit 304, the new theme probability distribution for obtaining the song, the new theme probability distribution packet
Include the probability distribution that the song is currently assigned to each theme label.
As shown in figure 9, the second acquisition unit 304 may include:
First generates subelement 3041, for distributing acquisition of information song single theme probability according to the theme label of remaining song
Distribution and theme song probability distribution, the residue song are the song in addition to the song in the song list.
Specifically, the first generation subelement 3041 is specifically, the first generation subelement can be according to following formula meter
Calculate song list theme probability distribution θ and theme song probability distribution
Wherein, wherein nm,zIndicate the number of songs that theme z is assigned in singing list m, the nz,tIndicate song t quilts
It is assigned to the number of theme z.
Second generates subelement 3042, for according to the single theme probability distribution of the song and theme song probability distribution, life
At the new theme probability distribution of the song.
Specifically, the second generation subelement 3042 can calculate new theme probability distribution according to following formula:
Optionally, which can be also used for:
Obtain current nm,zAnd nz,tAnd preset α and β.
According to the current n of acquisitionm,zAnd nz,tAnd preset α and β generates the single theme probability distribution of song and theme song is general
Rate is distributed.
For example, the first generation subelement 3041 can distribute information according to the current theme label of song, the single master of song is generated
Probability distribution and theme song probability distribution are inscribed, second generates subelement 3042, and acquisition should from the first generation subelement 3041
The single theme probability distribution of song and theme song probability distribution, generate the new theme distribution probability of the song.
(5) determination unit 305;
Determination unit 305, for determining the target that is assigned to of song in the song list according to the new theme probability distribution
Theme label.
It is understood that the determination unit 305 is sampled based on Gibbs (gibbs), respectively in each song list of iteration
Each song, each round all calculates current new theme distribution probability, then samples new theme label according to this probability.Directly
To discovery song list theme probability distribution θ and theme song probability distributionMarkov chain convergence, stop iteration, output it is desired
Parameter sings list theme probability distribution θ and theme song probability distributionThe final theme label per song also obtains simultaneously.
Specifically, as shown in figure 9, the determination unit 305 may include:
Subelement 3061 is recycled, it is corresponding main for being chosen from the tag set according to the new theme probability distribution
Label is inscribed, and returns to execution and the theme label of selection is distributed into corresponding song step in the song list, it is default until meeting
It is terminated when condition.
Determination subelement 3062, for the theme label using song in the song list is finally distributed to as the song
Target topic label.
(6) recommendation unit 306.
Recommendation unit 306, for generating corresponding song according to the target topic label that song is assigned in the song list
Recommendation list, and song recommendations are carried out based on the song recommendations list.
As shown in figure 9, the recommendation unit 306 may include:
Third generates subelement 3061, and the theme label for finally being distributed based on the song generates user's theme probability
Distribution and theme song probability distribution;
Recommend subelement 3062, is used for according to user's theme probability distribution, theme song probability distribution and presets
Recommendation condition generates song recommendations list, and carries out song recommendations based on the song recommendations list.
It can be together with reference to figure 9, for another structural schematic diagram of the song recommendations device.It can be specifically, the device can be with
Including:
Updating unit 307, for distributing information based on the current theme label of the song, to nm,zAnd nz,tIt carries out more
Newly.
Clearing cell 308 for removing the theme label for distributing to the song before, and updates nm,zAnd nz,t。
When it is implemented, above each unit can be realized as independent entity, arbitrary combination can also be carried out, is made
It is realized for same or several entities, the specific implementation of above each unit can be found in the embodiment of the method for front, herein not
It repeats again.
It can be seen from the above, the song recommendations device of the embodiment of the present invention, by first obtaining the single set of song, the song is singly gathered
Single including multiple songs, the song includes singly subject information, then builds the single tag set of the song according to the subject information,
The tag set includes at least one theme label, and the theme label in the tag set is subsequently distributed to the song
Song in list then obtains the new theme probability distribution of the song, and the song is determined according to the new theme probability distribution
The target topic label that song is assigned in list finally generates phase according to the target topic label that song is assigned in the song list
The song recommendations list answered, and song recommendations are carried out based on the song recommendations list.I.e. the embodiment of the present invention is by using song
Single theme label, is to have the topic model of supervision to be trained by unsupervised LDA model conversations, generates final song recommendations
List, to improve the accuracy of song recommendations.
Example IV,
Correspondingly, the embodiment of the present invention also provides a kind of server, as shown in Figure 10, which provides for the embodiment of the present invention
Server structural schematic diagram, specifically:
The server may include one or processor 401, one or more meters of more than one processing core
The components such as memory 402, power supply 403 and the input unit 404 of calculation machine readable storage medium storing program for executing.Those skilled in the art can manage
It solves, server architecture does not constitute the restriction to server shown in Figure 10, may include than illustrating more or fewer portions
Part either combines certain components or different components arrangement.Wherein:
Processor 401 is the control centre of the server, utilizes each of various interfaces and the entire server of connection
Part by running or execute the software program being stored in memory 402 and/or subelement, and calls and is stored in storage
Data in device 402, the various functions and processing data of execute server, to carry out integral monitoring to server.Optionally,
Processor 401 may include one or more processing cores;Preferably, processor 401 can integrate application processor and modulation /demodulation
Processor, wherein the main processing operation system of application processor, user interface and application program etc., modem processor master
Handle wireless communication.It is understood that above-mentioned modem processor can not also be integrated into processor 401.
Memory 402 can be used for storing software program and subelement, and processor 401 is stored in memory by operation
402 software program and subelement, to perform various functions application and data processing.Memory 402 can include mainly
Storing program area and storage data field, wherein storing program area can storage program area, the application journey needed at least one function
Sequence (such as sound-playing function, image player function etc.) etc.;Storage data field can be stored to be created according to using for server
Data etc..In addition, memory 402 may include high-speed random access memory, can also include nonvolatile memory, example
Such as at least one disk memory, flush memory device or other volatile solid-state parts.Correspondingly, memory 402 may be used also
To include Memory Controller, to provide access of the processor 401 to memory 402.
Server further includes the power supply 403 powered to all parts, it is preferred that power supply 403 can pass through power management system
System is logically contiguous with processor 401, to realize the work(such as management charging, electric discharge and power managed by power-supply management system
Energy.Power supply 403 can also include one or more direct current or AC power, recharging system, power failure monitor electricity
The random components such as road, power supply changeover device or inverter, power supply status indicator.
The server may also include input unit 404, which can be used for receiving the number or character letter of input
Breath, and generation keyboard related with user setting and function control, mouse, operating lever, optics or trace ball signal are defeated
Enter.
Although being not shown, server can also be including display unit etc., and details are not described herein.Specifically in the present embodiment,
Processor 401 in server can according to following instruction, by the process of one or more application program is corresponding can
It executes file to be loaded into memory 402, and the application program being stored in memory 402 is run by processor 401, from
And realize various functions, it is as follows:
The single set of song is taken, singly set includes that multiple songs are single to the song, and the song includes singly subject information;
The single tag set of the song is built according to the subject information, the tag set includes at least one theme mark
Label;
Theme label in the tag set is distributed into the song in the song list;
The new theme probability distribution of the song is obtained, the new theme probability distribution includes that the song is currently assigned to
The probability distribution of each theme label;
The target topic label that song is assigned in the song list is determined according to the new theme probability distribution;
Corresponding song recommendations list is generated according to the target topic label that song is assigned in the song list, and is based on institute
It states song recommendations list and carries out song recommendations.
The server may be implemented effective achieved by any song recommendations device that the embodiment of the present invention is provided
Effect refers to the embodiment of front, and details are not described herein.
The server of the embodiment of the present invention, by first obtaining the single set of song, singly set includes that multiple songs are single to the song, described
Single song includes subject information, then builds the single tag set of the song, the tag set according to according to the subject information
Including at least one theme label, the theme label in the tag set is subsequently distributed into the song in the song list,
Then the new theme probability distribution for obtaining the song determines that song is distributed in the song list according to the new theme probability distribution
The target topic label arrived finally generates corresponding song recommendations according to the target topic label that song is assigned in the song list
List, and song recommendations are carried out based on the song recommendations list.That is the embodiment of the present invention by using the single theme label of song,
It is to have the topic model of supervision to be trained by unsupervised LDA model conversations, final song recommendations list is generated, to carry
The high accuracy of song recommendations.
In the above-described embodiments, it all emphasizes particularly on different fields to the description of each embodiment, there is no the portion being described in detail in some embodiment
Point, the detailed description above with respect to song recommendations method is may refer to, details are not described herein again.
It should be noted that for song recommendations method of the present invention, this field common test personnel can manage
Solution realizes all or part of flow of the song recommendations method described in the embodiment of the present invention, is that can be controlled by computer program
Relevant hardware is made to complete, the computer program can be stored in a computer read/write memory medium, such as be stored in end
It in the memory at end, and is executed, may include in the process of implementation as the session is close by least one processor in the terminal
The flow of the embodiment of the generation method of key.Wherein, the storage medium can be magnetic disc, CD, read-only memory (ROM,
Read Only Memory), random access memory (RAM, Random Access Memory) etc..
For the song recommendations device described in the embodiment of the present invention, each function module can be integrated in a processing core
Can also be that modules physically exist alone in piece, can also two or more modules be integrated in a module.On
The form realization that hardware had both may be used in integrated module is stated, can also be realized in the form of software function module.The collection
If at module realized in the form of software function module and when sold or used as an independent product, can also be stored in
In one computer read/write memory medium, the storage medium is for example read-only memory, disk or CD etc..
The song recommendations method, apparatus based on label topic model is provided for the embodiments of the invention above to have carried out in detail
Thin to introduce, principle and implementation of the present invention are described for specific case used herein, and above example is said
The bright method and its core concept for being merely used to help understand the present invention;Meanwhile for those skilled in the art, according to this hair
Bright thought, there will be changes in the specific implementation manner and application range, in conclusion the content of the present specification should not manage
Solution is limitation of the present invention.
Claims (15)
1. a kind of song recommendations method based on label topic model, which is characterized in that including:
The single set of song is obtained, singly set includes that multiple songs are single to the song, and the song includes singly subject information and number of songs;
The single tag set of the song is built according to the subject information, the tag set includes at least one theme label;
Theme label in the tag set is distributed into the song in the song list;
Obtain the new theme probability distribution of the song, the new theme probability distribution include the song be currently assigned to it is each
The probability distribution of the theme label;
The target topic label that song is assigned in the song list is determined according to the new theme probability distribution;
Corresponding song recommendations list is generated according to the target topic label that song is assigned in the song list, and is based on the song
Bent recommendation list carries out song recommendations.
2. song recommendations method according to claim 1, which is characterized in that determine institute according to the new theme probability distribution
The target topic label that song is assigned in song list is stated, including:
Corresponding theme label is chosen from the tag set according to the new theme probability distribution, and return to execution to choose
Theme label distribute to corresponding song step in the song list, terminated when meeting preset condition;
The theme label of song in the song list will finally be distributed to as the target topic label of the song.
3. song recommendations method according to claim 1, which is characterized in that the new theme probability for obtaining the song
Distribution, including:
The single theme probability distribution of acquisition of information song and theme song probability distribution are distributed according to the theme label of remaining song, it is described
Remaining song is the song in addition to the song in the song list;
According to the single theme probability distribution of the song and theme song probability distribution, the new theme probability distribution of the song is generated.
4. song recommendations method according to claim 3, which is characterized in that the theme label according to remaining song point
Single theme probability distribution and theme song probability distribution are sung with acquisition of information, including:
Obtain current nm,zAnd nz,tAnd preset α and β;
According to the current n of acquisitionm,zAnd nz,tAnd preset α and β generates the single theme probability distribution of song and theme song probability point
Cloth;
Wherein, the nm,zIndicate the number of songs that theme label z is assigned in singing list m, the nz,tIndicate that song t is divided
It is fitted on the number of theme label z, the α indicates that the corresponding Di Li Crays hyper parameter of theme, the β indicate the corresponding Di Li of song
Cray hyper parameter.
5. according to claim 4 any one of them song recommendations method, which is characterized in that obtain the new of the song described
Before theme probability distribution, further include:
The theme label for distributing to the song before is removed, and updates nm,zAnd nz,t。
6. song recommendations method according to claim 4 or 5, which is characterized in that in the theme label by selection point
In song list described in dispensing after corresponding song, further include:
Information is distributed based on the current theme label of the song, to nm,zAnd nz,tIt is updated.
7. song recommendations method according to claim 1, which is characterized in that described to be assigned to according to song in the song list
Target topic label generate corresponding song recommendations list, and song recommendations are carried out based on the song recommendations list, including:
User's theme probability distribution and theme song probability distribution are generated based on the theme label that the song is finally distributed;
Song recommendations row are generated according to user's theme probability distribution, theme song probability distribution and default recommendation condition
Table, and song recommendations are carried out based on the song recommendations list.
8. a kind of song recommendations device based on label topic model, which is characterized in that including:
First acquisition unit, for obtaining the single set of song, singly set includes that multiple songs are single to the song, and the song includes singly that theme is believed
Breath;
Construction unit, for building the single tag set of the song according to the subject information, the tag set includes at least
One theme label;
Allocation unit, for the theme label in the tag set to be distributed to the song in the song list;
Second acquisition unit, the new theme probability distribution for obtaining the song, the new theme probability distribution include described
Song is currently assigned to the probability distribution of each theme label;
Determination unit, for determining the target topic mark that is assigned to of song in the song list according to the new theme probability distribution
Label;
Recommendation unit is arranged for generating corresponding song recommendations according to the target topic label that song is assigned in the song list
Table, and song recommendations are carried out based on the song recommendations list.
9. song recommendations device according to claim 8, which is characterized in that the determination unit includes:
Subelement is recycled, for choosing corresponding theme label from the tag set according to the new theme probability distribution,
And return to execution and the theme label of selection is distributed into corresponding song step in the song list, it is whole when meeting preset condition
Only;
Determination subelement, for the theme label using song in the song list is finally distributed to as the target topic of the song
Label.
10. song recommendations device according to claim 8, which is characterized in that the second acquisition unit includes:
First generates subelement, for distributing the single theme probability distribution of acquisition of information song and master according to the theme label of remaining song
Song probability distribution is inscribed, the residue song is the song in addition to the song in the song list;
Second generates subelement, for according to the single theme probability distribution of the song and theme song probability distribution, generating the song
Bent new theme probability distribution.
11. song recommendations device according to claim 10, which is characterized in that
Described first generates subelement, is specifically used for obtaining current nm,zAnd nz,tAnd preset α and β;It is current according to obtaining
Nm,zAnd nz,tAnd preset α and β generates the single theme probability distribution of song and theme song probability distribution;
Wherein, the nm,zIndicate the number of songs that theme label z is assigned in singing list m, the nz,tIndicate that song t is divided
It is fitted on the number of theme label z, the α indicates the Di Li Cray hyper parameters of the interest probabilities distribution of all users in song list, institute
It states β and indicates the corresponding Di Li Crays hyper parameter of song.
12. song recommendations device according to claim 10, which is characterized in that the song recommendations device further includes:
Clearing cell for removing the theme label for distributing to the song before, and updates nm,zAnd nz,t。
13. song recommendations device according to claim 11 or 12, which is characterized in that the song recommendations device also wraps
It includes:
Updating unit, for distributing information based on the current theme label of the song, to nm,zAnd nz,tIt is updated.
14. song recommendations device according to claim 8, which is characterized in that the recommendation unit includes:
Third generates subelement, and the theme label for finally being distributed based on the song generates user's theme probability distribution and master
Inscribe song probability distribution;
Recommend subelement, for according to user's theme probability distribution, theme song probability distribution and default recommendation condition
Song recommendations list is generated, and song recommendations are carried out based on the song recommendations list.
15. a kind of storage medium, is stored with processor-executable instruction, which is provided such as by executing described instruction
Song recommendations method described in any one of claim 1-7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810097213.9A CN108334601B (en) | 2018-01-31 | 2018-01-31 | Song recommendation method and device based on tag topic model and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810097213.9A CN108334601B (en) | 2018-01-31 | 2018-01-31 | Song recommendation method and device based on tag topic model and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108334601A true CN108334601A (en) | 2018-07-27 |
CN108334601B CN108334601B (en) | 2021-03-16 |
Family
ID=62927598
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810097213.9A Active CN108334601B (en) | 2018-01-31 | 2018-01-31 | Song recommendation method and device based on tag topic model and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108334601B (en) |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109213853A (en) * | 2018-08-16 | 2019-01-15 | 昆明理工大学 | A kind of Chinese community's question and answer cross-module state search method based on CCA algorithm |
CN109933678A (en) * | 2019-03-07 | 2019-06-25 | 合肥工业大学 | Art work recommended method, device, readable medium and electronic equipment |
CN110378488A (en) * | 2019-07-22 | 2019-10-25 | 深圳前海微众银行股份有限公司 | Federal training method, device, training terminal and the storage medium of client variation |
CN111090771A (en) * | 2019-10-31 | 2020-05-01 | 腾讯音乐娱乐科技(深圳)有限公司 | Song searching method and device and computer storage medium |
CN111611429A (en) * | 2019-02-25 | 2020-09-01 | 北京嘀嘀无限科技发展有限公司 | Data annotation method and device, electronic equipment and computer readable storage medium |
CN111753049A (en) * | 2020-06-15 | 2020-10-09 | 广东美的厨房电器制造有限公司 | Menu recommendation method and device, household appliance and storage medium |
CN112533030A (en) * | 2019-09-19 | 2021-03-19 | 聚好看科技股份有限公司 | Display method, display equipment and server of singing interface |
CN112836082A (en) * | 2021-02-08 | 2021-05-25 | 咪咕音乐有限公司 | Method and device for generating song list, electronic equipment and storage medium |
CN113220931A (en) * | 2021-03-24 | 2021-08-06 | 西安交通大学 | Multi-label song menu recommendation method, system, equipment and storage medium |
CN113449147A (en) * | 2021-07-06 | 2021-09-28 | 乐视云计算有限公司 | Video recommendation method and device based on theme |
CN110378488B (en) * | 2019-07-22 | 2024-04-26 | 深圳前海微众银行股份有限公司 | Client-side change federal training method, device, training terminal and storage medium |
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100332567A1 (en) * | 2009-06-26 | 2010-12-30 | Ramin Samadani | Media Playlist Generation |
CN102654859A (en) * | 2011-03-01 | 2012-09-05 | 北京彩云在线技术开发有限公司 | Method and system for recommending songs |
CN103970802A (en) * | 2013-02-05 | 2014-08-06 | 百度在线网络技术(北京)有限公司 | Song recommending method and device |
CN105718566A (en) * | 2016-01-20 | 2016-06-29 | 中山大学 | Intelligent music recommendation system |
US20160314133A1 (en) * | 2013-02-06 | 2016-10-27 | Facebook, Inc. | Pattern labeling |
CN106339507A (en) * | 2016-10-31 | 2017-01-18 | 腾讯科技(深圳)有限公司 | Method and device for pushing streaming media message |
CN106649686A (en) * | 2016-12-16 | 2017-05-10 | 天翼爱音乐文化科技有限公司 | User interest group dividing method and system based on multilayer potential features |
CN107368584A (en) * | 2017-07-21 | 2017-11-21 | 山东大学 | A kind of individualized video recommends method and system |
-
2018
- 2018-01-31 CN CN201810097213.9A patent/CN108334601B/en active Active
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100332567A1 (en) * | 2009-06-26 | 2010-12-30 | Ramin Samadani | Media Playlist Generation |
CN102654859A (en) * | 2011-03-01 | 2012-09-05 | 北京彩云在线技术开发有限公司 | Method and system for recommending songs |
CN103970802A (en) * | 2013-02-05 | 2014-08-06 | 百度在线网络技术(北京)有限公司 | Song recommending method and device |
US20160314133A1 (en) * | 2013-02-06 | 2016-10-27 | Facebook, Inc. | Pattern labeling |
CN105718566A (en) * | 2016-01-20 | 2016-06-29 | 中山大学 | Intelligent music recommendation system |
CN106339507A (en) * | 2016-10-31 | 2017-01-18 | 腾讯科技(深圳)有限公司 | Method and device for pushing streaming media message |
CN106649686A (en) * | 2016-12-16 | 2017-05-10 | 天翼爱音乐文化科技有限公司 | User interest group dividing method and system based on multilayer potential features |
CN107368584A (en) * | 2017-07-21 | 2017-11-21 | 山东大学 | A kind of individualized video recommends method and system |
Non-Patent Citations (3)
Title |
---|
SHOHEI KINOSHITA等: ""LDA-based music recommendation with CF-based similar user selection"", 《2015 IEEE 4TH GLOBAL CONFERENCE ON CONSUMER ELECTRONICS》 * |
曹磊: ""基于移动上下文的音乐推荐系统"", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
陶秋雨等: ""基于概率图模型的音乐推荐方法"", 《电子设计工程》 * |
Cited By (18)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109213853B (en) * | 2018-08-16 | 2022-04-12 | 昆明理工大学 | CCA algorithm-based Chinese community question-answer cross-modal retrieval method |
CN109213853A (en) * | 2018-08-16 | 2019-01-15 | 昆明理工大学 | A kind of Chinese community's question and answer cross-module state search method based on CCA algorithm |
CN111611429A (en) * | 2019-02-25 | 2020-09-01 | 北京嘀嘀无限科技发展有限公司 | Data annotation method and device, electronic equipment and computer readable storage medium |
CN111611429B (en) * | 2019-02-25 | 2023-05-12 | 北京嘀嘀无限科技发展有限公司 | Data labeling method, device, electronic equipment and computer readable storage medium |
CN109933678A (en) * | 2019-03-07 | 2019-06-25 | 合肥工业大学 | Art work recommended method, device, readable medium and electronic equipment |
CN109933678B (en) * | 2019-03-07 | 2021-04-06 | 合肥工业大学 | Artwork recommendation method and device, readable medium and electronic equipment |
CN110378488A (en) * | 2019-07-22 | 2019-10-25 | 深圳前海微众银行股份有限公司 | Federal training method, device, training terminal and the storage medium of client variation |
CN110378488B (en) * | 2019-07-22 | 2024-04-26 | 深圳前海微众银行股份有限公司 | Client-side change federal training method, device, training terminal and storage medium |
CN112533030B (en) * | 2019-09-19 | 2022-05-17 | 聚好看科技股份有限公司 | Display method, display equipment and server of singing interface |
CN112533030A (en) * | 2019-09-19 | 2021-03-19 | 聚好看科技股份有限公司 | Display method, display equipment and server of singing interface |
CN111090771B (en) * | 2019-10-31 | 2023-08-25 | 腾讯音乐娱乐科技(深圳)有限公司 | Song searching method, device and computer storage medium |
CN111090771A (en) * | 2019-10-31 | 2020-05-01 | 腾讯音乐娱乐科技(深圳)有限公司 | Song searching method and device and computer storage medium |
CN111753049A (en) * | 2020-06-15 | 2020-10-09 | 广东美的厨房电器制造有限公司 | Menu recommendation method and device, household appliance and storage medium |
CN111753049B (en) * | 2020-06-15 | 2024-04-16 | 广东美的厨房电器制造有限公司 | Menu recommendation method and device, household electrical appliance and storage medium |
CN112836082A (en) * | 2021-02-08 | 2021-05-25 | 咪咕音乐有限公司 | Method and device for generating song list, electronic equipment and storage medium |
CN113220931A (en) * | 2021-03-24 | 2021-08-06 | 西安交通大学 | Multi-label song menu recommendation method, system, equipment and storage medium |
CN113220931B (en) * | 2021-03-24 | 2023-01-03 | 西安交通大学 | Multi-label song menu recommendation method, system, equipment and storage medium |
CN113449147A (en) * | 2021-07-06 | 2021-09-28 | 乐视云计算有限公司 | Video recommendation method and device based on theme |
Also Published As
Publication number | Publication date |
---|---|
CN108334601B (en) | 2021-03-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108334601A (en) | Song recommendations method, apparatus and storage medium based on label topic model | |
Melcer et al. | Games research today: Analyzing the academic landscape 2000-2014 | |
CN111885399B (en) | Content distribution method, device, electronic equipment and storage medium | |
CN107066621A (en) | A kind of search method of similar video, device and storage medium | |
CN105138670B (en) | Audio file label generating method and system | |
CN103869998B (en) | A kind of method and device being ranked up to candidate item caused by input method | |
CN106294830A (en) | The recommendation method and device of multimedia resource | |
EP3940555A2 (en) | Method and apparatus of processing information, method and apparatus of recommending information, electronic device, and storage medium | |
CN110209869A (en) | A kind of audio file recommended method, device and storage medium | |
CN103970866B (en) | Microblog users interest based on microblogging text finds method and system | |
CN111723295B (en) | Content distribution method, device and storage medium | |
CN102890698A (en) | Method for automatically describing microblogging topic tag | |
CN106168980A (en) | Multimedia resource recommends sort method and device | |
Li et al. | A method of purchase prediction based on user behavior log | |
CN105740342A (en) | Social relation topic model based social network friend recommendation method | |
CN109240637A (en) | Processing method, device, equipment and the storage medium of volume adjustment | |
CN109961041A (en) | A kind of video frequency identifying method, device and storage medium | |
CN108334528A (en) | A kind of information recommendation method and device | |
CN109509010A (en) | A kind of method for processing multimedia information, terminal and storage medium | |
Almars et al. | Modelling user attitudes using hierarchical sentiment-topic model | |
CN110852047A (en) | Text score method, device and computer storage medium | |
CN105930532B (en) | A kind of method and apparatus from multimedia resource to user that recommending | |
CN110209875A (en) | User content portrait determines method, access object recommendation method and relevant apparatus | |
CN110377778A (en) | Figure sort method, device and electronic equipment based on title figure correlation | |
CN109241410A (en) | A kind of article recommended 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 | ||
GR01 | Patent grant | ||
GR01 | Patent grant |