CN109408665A - A kind of information recommendation method and device, storage medium - Google Patents
A kind of information recommendation method and device, storage medium Download PDFInfo
- Publication number
- CN109408665A CN109408665A CN201811642134.8A CN201811642134A CN109408665A CN 109408665 A CN109408665 A CN 109408665A CN 201811642134 A CN201811642134 A CN 201811642134A CN 109408665 A CN109408665 A CN 109408665A
- Authority
- CN
- China
- Prior art keywords
- song
- data
- default
- list
- full dose
- 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
Landscapes
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The embodiment of the invention discloses a kind of information recommendation method and devices, storage medium, this method comprises: obtaining the target user in default measurement period for the historical operating data of full dose history selection song;Corresponding relationship and the historical operating data based on preset operation data and operation score value obtain the history evaluation data of every history selection song in the full dose history selection song;The history evaluation data like degree to song for characterizing the target user;Time attenuation processing is carried out to the history evaluation data, obtains the historical statistics evaluation data of every history selection song;Data are evaluated according to the historical statistics, full dose history selection song is ranked up, obtain the ranking results of the full dose history selection song;It is single according to the ranking results and default full dose song, recommend target song.
Description
Technical field
The present invention relates to the information processing technology more particularly to a kind of information recommendation methods and device, storage medium.
Background technique
With the development of terminal and Internet technology, user generally listens to music by music application, and music application is logical
Cross some music recommendation algorithms, recommend to meet the music of its personal preference for user, music recommendation algorithm used at present, be by
Multiple users having similar tastes and interests constitute a group, utilize the interested information of users all in group and uninterested letter
Breath generates recommendation information, recommendation information is recommended to all users, obtains personal recommendation information in this way by affiliated group
Middle other users are affected, if there are abnormal user or other users interested information inaccuracy in other users,
The problem that the deviation of recommendation information and personal preference will be caused larger, that is to say, that the accuracy of recommendation information is lower.
Summary of the invention
It is a primary object of the present invention to propose a kind of information recommendation method and device, storage medium, recommendation can be improved
The accuracy of information.
The technical scheme of the present invention is realized as follows:
The embodiment of the invention provides a kind of information recommendation methods, which comprises
The target user in default measurement period is obtained for the historical operating data of full dose history selection song;
Corresponding relationship and the historical operating data based on preset operation data and operation score value, obtain described
Full dose history selects the history evaluation data of every history selection song in song;The history evaluation data are described for characterizing
Target user likes degree to song;
Time attenuation processing is carried out to the history evaluation data, obtains the historical statistics of every history selection song
Evaluate data;
Data are evaluated according to the historical statistics, full dose history selection song is ranked up, the full dose is obtained
The ranking results of history selection song;
It is single according to the ranking results and default full dose song, recommend target song.
In above scheme, the default measurement period includes at least one preset time period;The history evaluation data packet
Include the corresponding period evaluation data of each period at least one described preset time period;
It is described that time attenuation processing is carried out to the history evaluation data, obtain the history of every history selection song
Statistical appraisal data, comprising:
According to default attenuation coefficient and the corresponding relationship of time and the temporal information of each period characterization, obtain
To each period corresponding attenuation coefficient;
According to the attenuation coefficient, sum at least one described preset time period corresponding period evaluation data,
Obtain the historical statistics evaluation data.
It is described that single, recommendation target song is sung according to the ranking results and default full dose in above scheme, comprising:
According to the ranking results, determine from full dose history selection song with reference to song;
Song is referred to according to described, is determined from the default full dose song list single with reference to song;
According to described with reference to song list, recommend the target song.
It is described to be sung with reference to each in song with reference to the corresponding son reference of song with reference to song list including described in above scheme
It is single;
It is described single with reference to song according to described, recommend the target song, comprising:
According to the son with reference to the number of repetition of every song in song list, the target is determined from the reference song list
Song recommends the target song.
In above scheme, the number of repetition according to every song in the son reference song list is single from the reference song
In determine the target song, comprising:
Every historical statistics evaluation data with reference to song are ranked up, every number with reference to song is obtained
According to sequence serial number;
Number of repetition to the son with reference to every song in song list is ranked up, and obtains the son with reference to every head in song list
The number sequence serial number of song;
According to the number sort serial number, from it is described son with reference to song list in determine it is described it is every refer to song it is corresponding to
Recommend song, obtains at least one song to be recommended;
According to the number of the data sorting serial number and the song to be recommended sequence serial number, the song to be recommended is determined
Hobby score value;
According to the hobby score value, the target song is determined from least one described song to be recommended.
In above scheme, song is referred to according to described described, is determined from the default full dose song list single with reference to song
Later, the method also includes:
According to default song attributes, the song single with reference to each song in song list is quantified, the reference is obtained
Sing the reference attributive character that each song is single in list;
According to the single Rating Model of default song and it is described refer to attributive character, obtain comprising described with reference to song list each in song list
Appraisal result score data;The default corresponding relationship for singing single Rating Model characterization attributes feature and score data;
When being less than default scoring threshold value there are the first score data in the score data, by described with reference to song Dan Zhongyu
The corresponding song of first score data it is single from described with reference to being deleted in song list, obtain updated single with reference to song.
In above scheme, the basis it is default sing single Rating Model and it is described refer to attributive character, obtain comprising described
Before score data with reference to the single appraisal result of each song in song list, the method also includes:
Within default cycle of training, the sample operations data that target user is directed to full dose sample song are obtained;
According to the sample operations data, determine that positive sample song is single and negative sample is sung from the default full dose song list
It is single;
The single corresponding positive score data of positive sample song is obtained, the single and described positive score data of positive sample song is constituted
Positive sample;
The single corresponding negative score data of negative sample song is obtained, the single and described negative score data of negative sample song is constituted
Negative sample;
Using the positive sample and the negative sample, default initial list Rating Model of singing is trained, until after training
The initial accuracy rate for singing single Rating Model be not less than default accuracy threshold;
By the single Rating Model of initial song after the training, as the single Rating Model of the default song.
The embodiment of the invention provides a kind of information recommending apparatus, described device includes: acquiring unit, computing unit, declines
Subtract processing unit and recommendation unit;Wherein,
The acquiring unit, for obtaining the going through for full dose history selection song of the target user in default measurement period
History operation data;
The computing unit, for based on preset operation data and operation score value corresponding relationship and the history
Operation data obtains the history evaluation data of every history selection song in the full dose history selection song;The history is commented
Valence mumber likes degree to song according to for characterizing the target user;
The attenuation processing unit obtains every head for carrying out time attenuation processing to the history evaluation data
History selects the historical statistics of song to evaluate data;
The recommendation unit carries out full dose history selection song for evaluating data according to the historical statistics
Sequence obtains the ranking results of the full dose history selection song;And list is sung according to the ranking results and default full dose, it pushes away
Recommend target song.
In above scheme, the default measurement period includes at least one preset time period;The history evaluation data packet
Include the corresponding period evaluation data of each period at least one described preset time period;
The attenuation processing unit, specifically for according to presetting attenuation coefficient and the corresponding relationship of time and described every
The temporal information of a period characterization, obtains corresponding attenuation coefficient of each period;And according to the attenuation coefficient,
It sums at least one described preset time period corresponding period evaluation data, obtains the historical statistics evaluation data.
In above scheme, the recommendation unit is specifically used for according to the ranking results, selects to sing from the full dose history
It is determined in song with reference to song;And song is referred to according to described, it is determined from the default full dose song list single with reference to song;And
According to described with reference to song list, recommend the target song.
It is described to be sung with reference to each in song with reference to the corresponding son reference of song with reference to song list including described in above scheme
It is single;
The recommendation unit, specifically for the number of repetition according to the son with reference to every song in song list, from the ginseng
It examines in song list and determines the target song, recommend the target song.
In above scheme, the recommendation unit, specifically for evaluating data to every historical statistics with reference to song
It is ranked up, obtains every data sorting serial number with reference to song;And to the son with reference to the weight of every song in song list
Again number is ranked up, and obtains the son with reference to the number sequence serial number of every song in song list;And it is sorted according to the number
Serial number, it is described every with reference to the corresponding song to be recommended of song with reference to being determined in song list from the son, it obtains at least one and waits for
Recommend song;And it according to the number of the data sorting serial number and the song to be recommended sequence serial number, determines described to be recommended
The hobby score value of song;And according to the hobby score value, the target is determined from least one described song to be recommended
Song.
In above scheme, the recommendation unit is also used to described be sung with reference to song from the default full dose according to described
It determines, according to default song attributes, to sing single song amount of progress with reference to each in song list to described with reference to after song list in list
Change, obtains the reference attributive character single with reference to each song in song list;And according to the single Rating Model of default song and the reference
Attributive character obtains the score data comprising the appraisal result single with reference to each song in song list;The default song singly scores
The corresponding relationship of model characterization attributive character and score data;And when there are the first score datas to be less than in the score data
When presetting scoring threshold value, the corresponding song list of first score data with reference to described in song Dan Zhongyu is deleted from the reference song list
It removes, obtains updated single with reference to song.
In above scheme, described device further include: model training unit, in the single Rating Model of the default song of the basis
Attributive character is referred to described, before obtaining the score data comprising the appraisal result single with reference to each song in song list,
In default cycle of training, the sample operations data that target user is directed to full dose sample song are obtained;And according to the sample operations
Data determine that positive sample song is single and negative sample song is single from the default full dose song list;And it obtains the positive sample and sings single pair
The positive score data answered, the single and described positive score data of positive sample song constitute positive sample;And obtain the negative sample song list
Corresponding negative score data, the single and described negative score data of negative sample song constitute negative sample;And using the positive sample and
The negative sample is trained the default initial single Rating Model of song, until the single Rating Model of initial song after training is accurate
Rate is not less than default accuracy threshold;And it by the single Rating Model of initial song after the training, singly scores as the default song
Model.
The embodiment of the invention provides a kind of information recommending apparatus, described device includes: that processor, memory and communication are total
Line, the memory are communicated by the communication bus with the processor, and the memory stores the processor can
One or more program executed is executed as above by the processor when one or more of programs are performed
The step of stating any one information recommendation method.
The embodiment of the invention provides a kind of computer readable storage medium, the computer-readable recording medium storage has
Program causes at least one described processor to execute such as any of the above-described when described program is executed by least one processor
The step of information recommendation method.
The embodiment of the present invention provides a kind of information recommendation method and device, storage medium, which comprises obtains default
Target user in measurement period selects the historical operating data of song for full dose history;Based on preset operation data and behaviour
The corresponding relationship and the historical operating data for making score value obtain every history selection in the full dose history selection song
The history evaluation data of song;The history evaluation data like degree to song for characterizing the target user;To institute
It states history evaluation data and carries out time attenuation processing, obtain the historical statistics evaluation data of every history selection song;Root
Data are evaluated according to the historical statistics, full dose history selection song is ranked up, obtain the full dose history selection song
Bent ranking results;It is single according to the ranking results and default full dose song, recommend target song.Using above-mentioned technology realization side
Case carries out time attenuation processing to the history evaluation data in default measurement period, the history system after obtaining time attenuation processing
Meter evaluation data, and then the ranking results of full dose history selection song are obtained, recommend target song according to ranking results, due to going through
Commentary on historical events or historical records valence mumber is according to indicating that target user likes degree to song within a preset period of time, then to going through in default measurement period
Commentary on historical events or historical records valence mumber can obtain indicating the historical statistics evaluation data that user likes at present, Jin Erli according to time attenuation processing is carried out
With the ranking results of the historical statistics evaluation data of full dose historical song, obtained target song more meets target user and currently likes
It is good, improve the accuracy of recommendation information.
Detailed description of the invention
Fig. 1 is a kind of structural schematic diagram of information recommendation system provided in an embodiment of the present invention;
Fig. 2 is a kind of flow chart one of information recommendation method provided in an embodiment of the present invention;
Fig. 3 is a kind of flowchart 2 of information recommendation method provided in an embodiment of the present invention;
Fig. 4 is a kind of structural schematic diagram one of information recommending apparatus provided in an embodiment of the present invention;
Fig. 5 is a kind of structural schematic diagram two of information recommending apparatus provided in an embodiment 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.
In subsequent description, it is only using the suffix for indicating such as " module ", " component " or " unit " of element
Be conducive to explanation of the invention, itself there is no a specific meaning.Therefore, " module ", " component " or " unit " can mix
Ground uses.
As shown in Figure 1, a kind of structural schematic diagram of its information recommendation system of each embodiment to realize the present invention, this is
System 1 may include: server 10 and terminal 11;Wherein, server 10 can store the data file of application program, application program
User data etc., user data can also be broadcasted with it is synchronous etc.;Terminal 11 is when running application program, with server
10 carry out data interaction, can download the data file of the application program from server 10 by network, can also be to service
10 upload user of device operates the operation data of application program generation, and terminal 11 can be implemented in a variety of manners, for example, can be with
Being includes that the mobile terminals, and desktop computer such as mobile phone, tablet computer, laptop, palm PC etc. are solid
Determine terminal;Should program include music class application program.
It will be understood by those skilled in the art that information recommendation system structure shown in Fig. 1 is not constituted to recommendation information
The restriction of recommender system, information recommendation system may include components more more or fewer than diagram, or combine certain components, or
The different component layout of person.
It should be noted that the embodiment of the present invention can be realized based on information recommendation system shown in FIG. 1, information recommendation
Device can be server 10.
Embodiment one
The embodiment of the present invention provides a kind of information recommendation method, as shown in Fig. 2, this method comprises:
S201: the target user in default measurement period is obtained for the historical operating data of full dose history selection song.
Information recommending apparatus from terminal, obtains target user for full dose history selection song according to preset time period
Period operation data saves preset time period and period operation data correspondence, until obtaining in default measurement period
Historical operating data, default measurement period include at least one preset time period, and historical operating data includes that at least one is default
Period operation data in period in each period;For example, preset time period can be one day or a hour etc..
Illustratively, by taking music class application program as an example, when user uses music class application program at the terminal, terminal pair
The login account information (for example, User Identity number (ID, IDentity)) of user is recorded, and terminal is also to operand
According to being recorded, operation data is that user carries out the data generated after selection operation to the song in music class application program, eventually
End saves User ID and operation data correspondence, and the period operation data combination User ID in each period is sent
To server, server is saved according to period operation data of the User ID to each user;Wherein, for the selection of song
Operation includes: to thumb up operation, collection operation, comment operation, sharing operation, listen to operation, cut song operation, audition operation, purchase
Operate, be set as CRBT operation, down operation etc., operation data includes the song identity (for example, song ID) and selection of song
The operation species of operation.
S202: corresponding relationship and historical operating data based on preset operation data and operation score value obtain full dose
History selects the history evaluation data of every history selection song in song;History evaluation data are for characterizing target user's singing in antiphonal style
Bent likes degree.
Information recommending apparatus determines the historical operating data of each user in target user from historical operating data, according to pre-
If operation data and operation score value corresponding relationship, when corresponding to the period each in the historical operating data of each user
Section operation data carries out operation score value and calculates, and obtains corresponding period evaluation data of each period, and then obtain each user
The history evaluation data being made of corresponding period all periods evaluation data.
Illustratively, preset time period is one day, calculates corresponding period evaluation data of each period, is exactly to calculate to use
The family odd-numbered day, information recommending apparatus can be accustomed to according to routine use, for each operation species to the degree of liking of different songs
Be arranged it is corresponding like score value, which reflects that user expressed by the selection operation of each operation species likes degree,
Obtain the corresponding relationship of preset operation data and operation score value as shown in table 1 below:
Table 1
Operation data | Score value |
Cut song operation | -0.5 |
Label is liked operating | 2 |
Label does not like operation | -2 |
Down operation | 4 |
Comment operation | 1 |
Sharing operation | 2 |
Collection operation | 2 |
Audition operation | 0.5 |
Listen to operation | 4 |
It is set as CRBT operation | 3 |
Search operation | 1 |
Illustratively, information recommending apparatus is according to the selection operation in the corresponding relationship and historical operating data of table 1
Operation species calculate the history evaluation data of every history selection song in full dose history selection song, and according to following form
Recorded: " User ID-song ID- operation score value-date ", the date is each period pair at least one preset time period
The temporal information answered.
Illustratively, after server obtains the period operation data of each period, period operation data is utilized, is calculated
Each user evaluates data for the period of every history selection song, for example, the period operation data packet of history selection song A
Include: in one day, user B is searched for 5 times, shares 4 times, collects 1 time, is commented on 2 times, downloads 2 times and cut song 3 times, then according to table
1 corresponding relationship obtains formula (1), using formula (1) be calculated user B in one day for history selection song A when
Section evaluation data are 23.5:
5 × 1+4 × 2+1 × 2+2 × 1+2 × 4+3 × (- 0.5)=23.5 (1)
When this day, corresponding temporal information was on October 11st, 2018, then can select to sing to history according to following form
The period evaluation data of bent A are recorded: user B-song A-23.5-20181011.
S203: carrying out time attenuation processing to history evaluation data, and the historical statistics for obtaining every history selection song is commented
Valence mumber evidence.
The history evaluation data for each user that information recommending apparatus obtains are by the corresponding period evaluation of each period
Data composition, time attenuation processing is carried out to evaluation of all periods corresponding period data, it is corresponding to obtain each user
Data are evaluated in historical statistics.
It should be noted that since hobby of the user to song may be dynamic change, according only to a preset time
Period evaluation data in section tend not to accurately determine the current hobby of the user, thus information recommending apparatus is according to default
History evaluation data in measurement period, determine the hobby of user;For example, presetting with 90 days as a measurement period, root
Score value is liked for 90 odd-numbered days of every history selection song according to user, to determine that user was directed to every first history in 90 days
Selection song likes total score.
In some embodiments, information recommending apparatus is according to default attenuation coefficient and the corresponding relationship of time and each
The temporal information of period characterization, obtains corresponding attenuation coefficient of each period;It is default at least one according to attenuation coefficient
Period, corresponding period evaluation data were summed, and obtained historical statistics evaluation data.
Illustratively, since hobby of the user to song is dynamic change, thus user evaluates data at recent period
User is more accurately often reacted to the current hobby of song than the period of early stage evaluation data, that is to say, that period review number
According to the increase with the time interval with current time, gradually decay to the reflection accuracy that user currently likes, thus information
Recommendation apparatus is in the finish time of default measurement period, according to corresponding temporal information of each period and default measurement period pair
The time interval for the initial time answered determines corresponding attenuation coefficient of each period, according to attenuation coefficient, to all periods
Corresponding period evaluation data are weighted summation, obtain historical statistics evaluation data.
Illustratively, it presets attenuation coefficient and the corresponding relationship of time is expressed as formula (2):
N (t)=N0e-t/T (2)
Wherein, N0Indicate that the initial attenuation coefficient at the beginning of default measurement period, t indicate that each period is corresponding
Time interval at the beginning of temporal information and default measurement period, N (t) indicate corresponding attenuation coefficient of each period, e
Indicate that index, T indicate the finish time of default measurement period.
S204: evaluating data according to historical statistics, is ranked up to full dose history selection song, obtains the selection of full dose history
The ranking results of song.
Information recommending apparatus evaluates data according to the historical statistics of each user, according to the big of historical statistics evaluation data
It is small, ascending sort or descending sort are carried out to full dose history selection song, obtain ranking results.
S205: single, recommendation target song is sung according to ranking results and default full dose.
Information recommending apparatus is to each user, and based on ranking results, n refers to song, n from full dose history selection song
For the integer greater than 0, n is greater than going through for other songs in full dose history selection song with reference to the historical statistics evaluation data of song
History statistical appraisal data, n are that the favorite n of each user refers to song with reference to song, refer to song and sound further according to n
The default full dose of happy class application program sings list, determines target song.
It should be noted that a kind of information recommendation method as shown in Figure 3, the specific implementation of step S205 is as executed
S2051-S2053, comprising:
S2051: it according to ranking results, determines from full dose history selection song with reference to song;
Information recommending apparatus selection n from full dose history selection song refers to song.
S2052: it according to reference song, is determined from default full dose song list single with reference to song;
Information recommending apparatus refers to song to each in reference song, determines from default full dose song list comprising each ginseng
All song nonocultures are that each son with reference to song is single with reference to song, and then obtains by all sub- references by all song lists for examining song
Sing single group at reference song singly.
S2053: single, recommendation target song is sung according to reference.
Information recommending apparatus determines target song from the single song of reference song, and is recommended by being sent to terminal to user
Target song.
In some embodiments, single with reference to song includes with reference to each single with reference to song with reference to the corresponding son of song in song;Letter
Number of repetition of the recommendation apparatus according to son with reference to every song in song list is ceased, target song is determined from reference song list, recommends
Target song.
Illustratively, song A is referred to reference in song one with the favorite n of userLFor, L is greater than 0 and little
In the integer of n, information recommending apparatus can be sung in list from default full dose and determine comprising referring to song ALSon it is single with reference to song, the son
It is singly expressed as { song list L1, sings list L2 ..., sings list LN } with reference to song, N is the integer greater than 2, counts the son with reference to every head in song list
The number of repetition of song, the number of repetition to the son with reference to all songs in song list are ranked up, and obtain the son with reference in song list
The number sequence serial number of every song;Q songs are determined as with reference to song A before number sequence serial number is belonged toLIt is corresponding to
Recommend song TL, q is the positive integer greater than 0, and then obtains referring at least one that the corresponding song to be recommended of song forms by n
A song to be recommended, wherein song T to be recommendedLIt may include at least one song.
In some embodiments, information recommending apparatus is ranked up every historical statistics evaluation data with reference to song,
Obtain every data sorting serial number with reference to song;Antithetical phrase is ranked up with reference to the number of repetition of every song in song list, is obtained
Number sequence serial number of the son with reference to every song in song list;According to number sequence serial number, every head is determined from son reference song list
With reference to the corresponding song to be recommended of song, at least one song to be recommended is obtained;According to data sorting serial number and song to be recommended
Number sort serial number, determine the hobby score value of song to be recommended;According to hobby score value, from least one song to be recommended really
Make target song.
Illustratively, information recommending apparatus carries out descending sort with reference to the historical statistics evaluation data of song to n, obtains
Every data sorting serial number g with reference to song;Again in the number sequence serial number from the son with reference to every song in song list, determine
Every number sequence serial number h with reference to the corresponding song to be recommended of song;The hobby point of song to be recommended is calculated according to formula (3)
Value Score:
According to the hobby score value of song to be recommended, the hobby higher mesh of score value is determined from least one song to be recommended
Song is marked, and recommends user.
It should be noted that information recommending apparatus is sung reference single according to default song attributes after step S2052
In the single song of each song quantified, obtain the reference attributive character single with reference to song each in song list;It is singly commented according to default song
Sub-model and attributive character is referred to, obtained comprising the score data with reference to the single appraisal result of each song in song list;Default song is single
The corresponding relationship of Rating Model characterization attributes feature and score data;It is preset when being less than in score data there are the first score data
When the threshold value that scores, it will delete, obtain updated with reference to single sung in list from reference of the corresponding song of song the first score data of Dan Zhongyu
It is single with reference to song.
Illustratively, information recommending apparatus is the influence for avoiding song inferior single, single to each song according to default song attributes
In every song quantified, obtain the multidimensional property vector of every song, it is corresponding to preset each song attributes in song attributes
One-dimensional data;Multidimensional property matrix is generated using the attribute vector of all songs in each song list, multidimensional property matrix is exactly every
The single reference attributive character of a song;Using the default multidimensional property matrix for singing single Rating Model and each song list, each song is calculated
Single appraisal result;Appraisal result finally single according to each song, singly screens reference song.
Illustratively, default song attributes include Songs age, singer's age, singer gender information, song rhythm
Information, song tone information etc., singer gender information include that the singer gender information of songster is equal to the song of 0 and songstress
Hand gender information is equal to 1, and song rhythm information is song beat number per minute, and song tone information includes the CDEFGAB of song
Tune respectively corresponds 1-7, and with the songster of birth in 1979, in distribution song in 2007, which was A tune, per minute 60
It claps, the multidimensional property vector of the distribution song is expressed as { 2007,1979,0,60,6 };Wherein, each song in song attributes is preset
Position of the bent attribute in multidimensional property vector can change, it is only necessary to guarantee multidimensional property vector song attributes sequence with it is pre-
If singing the corresponding song attributes sequence consensus of single Rating Model.
In some embodiments, information recommending apparatus obtains target user and is directed to full dose sample within default cycle of training
The sample operations data of song;According to sample operations data, positive sample song list and negative sample are determined from default full dose song list
Song is single;The single corresponding positive score data of positive sample song is obtained, positive sample song is single and positive score data constitutes positive sample;Obtain negative sample
The single corresponding negative score data of this song, negative sample song is single and negative score data constitutes negative sample;It is right using positive sample and negative sample
Default initial list Rating Model of singing is trained, until the initial accuracy rate for singing list Rating Model after training is not less than default standard
True threshold value;By the single Rating Model of initial song after training, as the single Rating Model of default song.
Illustratively, information recommending apparatus obtains target user for full dose sample song within default cycle of training
Sample operations data;According to sample operations data, the sample statistics evaluation data of every first sample song are obtained;According to sample statistics
Data are evaluated, full dose sample song is ranked up, the sample ranking results of full dose sample song are obtained;It is sorted and is tied according to sample
Fruit determines that positive sample song is single and negative sample song is single from default full dose song list, and single positive sample song is favorite comprising user
The song list of several songs, negative sample song are single for the song list for the several songs least liked comprising user;According to default song attributes,
The song single to each song in positive sample song list quantifies, and obtains the positive attributive character that each song is single in positive sample song list, according to
Default song attributes, the song single to each song in negative sample song list quantify, and obtain single negative of each song in negative sample song list
Attributive character;Positive attributive character and positive score data constitute positive sample, and negative attributive character and negative score data constitute negative sample.
Illustratively, when positive score data is 1, negative score data is -1, information recommending apparatus is according to each song list
Appraisal result deletes song list of the appraisal result less than 0 from reference song list.
It should be noted that full dose sample song corresponds to full dose history above-mentioned selection song, full dose sample song and complete
Measuring history selection song is that acquisition time is different, and similarly, sample operations data correspond to historical operating data above-mentioned, sample
This operation data and historical operation song are that acquisition time is different;Therefore, sample system is obtained based on sample operations data
The process of meter evaluation data and sample ranking results evaluates data and sequence with historical statistics is obtained based on historical operating data
As a result process is consistent.
Illustratively, the single Rating Model of default song can be obtained using Adboost algorithm, is specifically included: the initial song of setting
Single Rating Model, initial list Rating Model of singing includes the corresponding Weak Classifier of each adjacent two dimension data;According to multidimensional data
In each dimension data primary data weight, utilize positive sample and negative sample corresponding each adjacent two dimension data training one
A Weak Classifier calculates the classifier weight of each Weak Classifier, generates strong point using each Weak Classifier and classifier weight
Class device;The accuracy rate for calculating strong classifier adjusts primary data weight when accuracy rate is less than default accuracy threshold, until strong
The accuracy rate of classifier is not less than default accuracy threshold;Single Rating Model is sung using strong classifier as default.
Illustratively, presetting song attributes includes 5 song attributes, is Songs age, singer's age, singer respectively
Gender information, song rhythm information and song tone information, strong classifier are generated by K=5 Weak Classifier, for example, working as every phase
When adjacent two dimension datas are expressed as { the first song attributes x, the second song attributes y }, sample is obtained according to positive sample and negative sample
This input is { (x1, y1),(x2,y2) ..., (xm,ym), sample output is { -1 ,+1 }, and m is the integer greater than 0;According to initial
Data weighting exports the default initial single Rating Model of song of training using sample input and sample, it is corresponding to obtain the first song attributes x
Weak Classifier G1(x), and then the corresponding Weak Classifier of each song attributes is obtained;Wherein, primary data weight D (1) is such as public
Shown in formula (4):
D (1)=(w11,w12,…,w1m) (4)
Wherein,
Further, to Weak Classifier Gk(x) it is trained, specifically includes: using Weak Classifier Gk(x) corresponding data
Weight D (k) is trained initial list Rating Model of singing, obtains Weak Classifier Gk(x), wherein k=1,2 ... K, data power
Weight D (k) is such as shown in formula (5):
D (k)=(wk1,wk2,…,wkm) (5)
Wherein,Weak Classifier G is calculated using formula (6)k(x) error in classification rate ek:
Wherein, to Weak Classifier Gk(xi) input (xi, yi) appraisal result is exported afterwards, which is not equal to yiIt is corresponding
Preset threshold I.
Weak Classifier G is calculated using formula (7)k(x) classifier weight ak:
The corresponding data weighting D (k+1) of Weak Classifier k+1 is calculated using formula (8):
Wherein, ZkIt is standardizing factor, as shown in formula (9):
Strong classifier is generated using formula (10):
It should be noted that the error in classification rate rate of Weakly separated device is bigger, the classifier weight of the Weak Classifier is smaller.
It is understood that information recommending apparatus carries out at time decaying the history evaluation data in default measurement period
Data are evaluated in reason, the historical statistics after obtaining time attenuation processing, and then obtain the ranking results of full dose history selection song, root
Recommend target song according to ranking results, since history evaluation data indicate that target user likes song within a preset period of time
Degree can obtain indicating that user is current then carrying out time attenuation processing to the history evaluation data in default measurement period
Data are evaluated in the historical statistics of hobby, and then using the ranking results of the historical statistics of full dose historical song evaluation data, are obtained
Target song more meet target user and currently like, improve the accuracy of recommendation information.
Embodiment two
Same inventive concept based on embodiment one, is further detailed.
The embodiment of the present invention provides a kind of information recommending apparatus 4, as shown in figure 4, the device 4 includes: acquiring unit 40, meter
Calculate unit 41, attenuation processing unit 42 and recommendation unit 43;Wherein,
The acquiring unit 40, for obtaining the target user in default measurement period for full dose history selection song
Historical operating data;
The computing unit 41 for the corresponding relationship based on preset operation data and operation score value and described is gone through
History operation data obtains the history evaluation data of every history selection song in the full dose history selection song;The history
Evaluation data like degree to song for characterizing the target user;
The attenuation processing unit 42 obtains described every for carrying out time attenuation processing to the history evaluation data
History selects the historical statistics of song to evaluate data;
The recommendation unit 43, for according to the historical statistics evaluate data, to the full dose history select song into
Row sequence obtains the ranking results of the full dose history selection song;And list is sung according to the ranking results and default full dose,
Recommend target song.
In some embodiments, the default measurement period includes at least one preset time period;The history evaluation number
Data are evaluated according to including period each at least one the described preset time period corresponding period;
The attenuation processing unit 42, specifically for according to presetting attenuation coefficient and the corresponding relationship of time and described
The temporal information of each period characterization, obtains corresponding attenuation coefficient of each period;And it is according to the decaying
Number sums at least one described preset time period corresponding period evaluation data, obtains the historical statistics review number
According to.
In some embodiments, the recommendation unit 43 is specifically used for according to the ranking results, from the full dose history
It determines in selection song with reference to song;And song is referred to according to described, it determines from the default full dose song list with reference to song
It is single;And recommend the target song with reference to song list according to described.
In some embodiments, described single including each with reference to the corresponding sub- reference of song in the reference song with reference to song
Song is single;
The recommendation unit 43, specifically for the number of repetition according to the son with reference to every song in song list, from described
With reference to the target song is determined in song list, recommend the target song.
In some embodiments, the recommendation unit 43, specifically for commenting every historical statistics with reference to song
Valence mumber obtains every data sorting serial number with reference to song according to being ranked up;And to the son with reference to first song every in song list
Bent number of repetition is ranked up, and obtains the son with reference to the number sequence serial number of every song in song list;And according to described time
Number sequence serial number, it is described every with reference to the corresponding song to be recommended of song with reference to being determined in song list from the son, it obtains at least
One song to be recommended;And according to the number of the data sorting serial number and the song to be recommended sort serial number, determine described in
The hobby score value of song to be recommended;And according to the hobby score value, institute is determined from least one described song to be recommended
State target song.
In some embodiments, the recommendation unit 43 is also used to described be preset with reference to song from described according to described
It determines, according to default song attributes, to sing the song that each song is single in list to the reference with reference to after song list in full dose song list
Quantified, obtains the reference attributive character single with reference to each song in song list;And according to the single Rating Model of default song and institute
It states and obtains the score data comprising the appraisal result single with reference to each song in song list with reference to attributive character;The default song
The corresponding relationship of single Rating Model characterization attributes feature and score data;And when there are the first scoring numbers in the score data
When according to less than presetting scoring threshold value, the corresponding song list of first score data with reference to described in song Dan Zhongyu is sung from the reference
It deletes, obtains updated single with reference to song in list.
In some embodiments, model training unit, in the default single Rating Model of song of the basis and the reference
Attributive character, before obtaining the score data comprising the appraisal result single with reference to each song in song list, in default training week
In phase, the sample operations data that target user is directed to full dose sample song are obtained;And according to the sample operations data, from described
Determine that positive sample song is single and negative sample song is single in default full dose song list;And obtain the single corresponding positive scoring number of positive sample song
According to the single and described positive score data of positive sample song constitutes positive sample;And obtain the single corresponding negative scoring of negative sample song
Data, the single and described negative score data of negative sample song constitute negative sample;And the positive sample and the negative sample are utilized, it is right
Default initial list Rating Model of singing is trained, until the initial accuracy rate for singing list Rating Model after training is not less than default standard
True threshold value;And by the single Rating Model of initial song after the training, as the single Rating Model of the default song.
It should be noted that in practical applications, above-mentioned acquiring unit 40, computing unit 41,42 and of attenuation processing unit
Recommendation unit 43 can be realized, specially CPU (Central Processing by the processor 44 being located on information recommending apparatus 4
Unit, central processing unit), MPU (Microprocessor Unit, microprocessor), DSP (Digital Signal
Processing, digital signal processor) or field programmable gate array (FPGA, Field Programmable Gate
) etc. Array realize.
The embodiment of the invention also provides a kind of information recommending apparatus 4, as shown in figure 5, the device 4 include: processor 44,
Memory 45 and communication bus 46, memory 45 is communicated by communication bus 46 with processor 44, at 45 storage of memory
One or more executable program of device 44 is managed, when one or more program is performed, is executed such as by processor 44
Any one information recommendation method described in previous embodiment.
In practical applications, memory 45 can be volatibility first memory (volatile memory), such as at random
It accesses first memory (Random-Access Memory, RAM);Or non-volatile first memories (non-volatile
Memory), such as read-only first memory (Read-Only Memory, ROM), quick flashing first memory (flash
Memory), hard disk (Hard Disk Drive, HDD) or solid state hard disk (Solid-State Drive, SSD);Or above-mentioned kind
The combination of the first memory of class, and program and data are provided to processor 44.
The embodiment of the invention provides a kind of computer readable storage medium, computer-readable recording medium storage has one
Or multiple programs, one or more program can be executed by one or more processor, described program is held by processor 44
Any one information recommendation method as in the foregoing embodiment is realized when row.
It should be understood by those skilled in the art that, the embodiment of the present invention can provide as method, system or computer program
Product.Therefore, the shape of hardware embodiment, software implementation or embodiment combining software and hardware aspects can be used in the present invention
Formula.Moreover, the present invention, which can be used, can use storage in the computer that one or more wherein includes computer usable program code
The form for the computer program product implemented on medium (including but not limited to magnetic disk storage and optical memory etc.).
The present invention be referring to according to the method for the embodiment of the present invention, the process of equipment (system) and computer program product
Figure and/or block diagram describe.It should be understood that every one stream in flowchart and/or the block diagram can be realized by computer program instructions
The combination of process and/or box in journey and/or box and flowchart and/or the block diagram.It can provide these computer programs
Instruct the processor of general purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices to produce
A raw machine, so that being generated by the instruction that computer or the processor of other programmable data processing devices execute for real
The device for the function of being specified in present one or more flows of the flowchart and/or one or more blocks of the block diagram.
These computer program instructions, which may also be stored in, is able to guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works, so that it includes referring to that instruction stored in the computer readable memory, which generates,
Enable the manufacture of device, the command device realize in one box of one or more flows of the flowchart and/or block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device, so that counting
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, thus in computer or
The instruction executed on other programmable devices is provided for realizing in one or more flows of the flowchart and/or block diagram one
The step of function of being specified in a box or multiple boxes.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.
Claims (16)
1. a kind of information recommendation method, which is characterized in that the described method includes:
The target user in default measurement period is obtained for the historical operating data of full dose history selection song;
Corresponding relationship and the historical operating data based on preset operation data and operation score value, obtain the full dose
History selects the history evaluation data of every history selection song in song;The history evaluation data are for characterizing the target
User likes degree to song;
Time attenuation processing is carried out to the history evaluation data, obtains the historical statistics evaluation of every history selection song
Data;
Data are evaluated according to the historical statistics, full dose history selection song is ranked up, the full dose history is obtained
Select the ranking results of song;
It is single according to the ranking results and default full dose song, recommend target song.
2. the method according to claim 1, wherein the default measurement period includes at least one preset time
Section;The history evaluation data evaluate data including period each at least one the described preset time period corresponding period;
It is described that time attenuation processing is carried out to the history evaluation data, obtain the historical statistics of every history selection song
Evaluate data, comprising:
According to default attenuation coefficient and the corresponding relationship of time and the temporal information of each period characterization, institute is obtained
State corresponding attenuation coefficient of each period;
According to the attenuation coefficient, sums, obtain at least one described preset time period corresponding period evaluation data
Data are evaluated in the historical statistics.
3. the method according to claim 1, wherein it is described sung according to the ranking results and default full dose it is single,
Recommend target song, comprising:
According to the ranking results, determine from full dose history selection song with reference to song;
Song is referred to according to described, is determined from the default full dose song list single with reference to song;
According to described with reference to song list, recommend the target song.
4. according to the method described in claim 3, it is characterized in that, described single including each ginseng in the reference song with reference to song
It is single with reference to song to examine the corresponding son of song;
It is described single with reference to song according to described, recommend the target song, comprising:
According to the son with reference to the number of repetition of every song in song list, the target song is determined from the reference song list
Song recommends the target song.
5. according to the method described in claim 4, it is characterized in that, the weight according to every song in the son reference song list
Again it counts, determines the target song from the reference song list, comprising:
Every historical statistics evaluation data with reference to song are ranked up, every data with reference to song is obtained and arranges
Sequence serial number;
Number of repetition to the son with reference to every song in song list is ranked up, and obtains the son with reference to every song in song list
Number sort serial number;
It is described every corresponding to be recommended with reference to song with reference to being determined in song list from the son according to number sequence serial number
Song obtains at least one song to be recommended;
According to the number of the data sorting serial number and the song to be recommended sequence serial number, the happiness of the song to be recommended is determined
Good score value;
According to the hobby score value, the target song is determined from least one described song to be recommended.
6. according to the method described in claim 3, it is characterized in that, refer to song according to described described, from described default complete
After determining in amount song list with reference to list is sung, the method also includes:
According to default song attributes, the song single with reference to each song in song list is quantified, is obtained described single with reference to song
In the single reference attributive character of each song;
According to the single Rating Model of default song and it is described refer to attributive character, obtain commenting comprising described with reference to each song is single in song list
Divide the score data of result;The default corresponding relationship for singing single Rating Model characterization attributes feature and score data;
When being less than default scoring threshold value there are the first score data in the score data, by described with reference to described in song Dan Zhongyu
The corresponding song of first score data is single to be deleted from the reference song list, is obtained updated single with reference to song.
7. according to the method described in claim 6, it is characterized in that, in the default single Rating Model of song of the basis and the reference
Attributive character, before obtaining the score data comprising the appraisal result single with reference to each song in song list, the method is also wrapped
It includes:
Within default cycle of training, the sample operations data that target user is directed to full dose sample song are obtained;
According to the sample operations data, determine that positive sample song is single and negative sample song is single from the default full dose song list;
The single corresponding positive score data of positive sample song is obtained, the single and described positive score data of positive sample song constitutes positive sample
This;
The single corresponding negative score data of negative sample song is obtained, the single and described negative score data of negative sample song constitutes negative sample
This;
Using the positive sample and the negative sample, default list Rating Model of initially singing is trained, until first after training
The accuracy rate for beginning to sing single Rating Model is not less than default accuracy threshold;
By the single Rating Model of initial song after the training, as the single Rating Model of the default song.
8. a kind of information recommending apparatus, which is characterized in that described device includes: acquiring unit, computing unit, attenuation processing unit
And recommendation unit;Wherein,
The acquiring unit, for obtaining the target user in default measurement period for the history behaviour of full dose history selection song
Make data;
The computing unit, for based on preset operation data and operation score value corresponding relationship and the historical operation
Data obtain the history evaluation data of every history selection song in the full dose history selection song;The history evaluation number
Degree is liked to song according to for characterizing the target user;
The attenuation processing unit obtains every first history for carrying out time attenuation processing to the history evaluation data
The historical statistics of song is selected to evaluate data;
The recommendation unit is ranked up full dose history selection song for evaluating data according to the historical statistics,
Obtain the ranking results of the full dose history selection song;And single, recommendation mesh is sung according to the ranking results and default full dose
Mark song.
9. device according to claim 8, which is characterized in that the default measurement period includes at least one preset time
Section;The history evaluation data evaluate data including period each at least one the described preset time period corresponding period;
The attenuation processing unit, specifically for according to preset attenuation coefficient and the corresponding relationship of time and it is described each when
Between segment table levy temporal information, obtain corresponding attenuation coefficient of each period;And according to the attenuation coefficient, to institute
It states the corresponding period evaluation data of at least one preset time period to sum, obtains the historical statistics evaluation data.
10. device according to claim 8, which is characterized in that
The recommendation unit is specifically used for determining to refer to from full dose history selection song according to the ranking results
Song;And song is referred to according to described, it is determined from the default full dose song list single with reference to song;And according to described with reference to song
It is single, recommend the target song.
11. device according to claim 10, which is characterized in that described single including each in the reference song with reference to song
It is single with reference to song with reference to the corresponding son of song;
The recommendation unit, specifically for the number of repetition according to the son with reference to every song in song list, from described with reference to song
The target song is determined in list, recommends the target song.
12. device according to claim 11, which is characterized in that
The recommendation unit obtains institute specifically for being ranked up to every historical statistics evaluation data with reference to song
State every data sorting serial number with reference to song;And the number of repetition to the son with reference to every song in song list is ranked up,
The son is obtained with reference to the number sequence serial number of every song in song list;And according to number sequence serial number, from the sub- ginseng
Examine song list in determine it is described it is every refer to the corresponding song to be recommended of song, obtain at least one song to be recommended;And according to
The number of the data sorting serial number and the song to be recommended sequence serial number, determines the hobby score value of the song to be recommended;
And according to the hobby score value, the target song is determined from least one described song to be recommended.
13. device according to claim 10, which is characterized in that
The recommendation unit is also used to described determine to refer to from the default full dose song list with reference to song according to described
After singing list, according to default song attributes, the song single with reference to each song in song list is quantified, the reference is obtained
Sing the reference attributive character that each song is single in list;And according to the single Rating Model of default song and it is described refer to attributive character, wrapped
Score data containing the appraisal result single with reference to each song in song list;The single Rating Model characterization attributes feature of default song
With the corresponding relationship of score data;And when being less than default scoring threshold value there are the first score data in the score data,
Single sing in list from the reference of the corresponding song of the first score data with reference to described in song Dan Zhongyu is deleted, is obtained updated
It is single with reference to song.
14. device according to claim 13, which is characterized in that described device further include:
Model training unit, for the basis it is default sing single Rating Model and it is described refer to attributive character, obtain comprising institute
Before stating the score data with reference to the single appraisal result of song each in song list, within default cycle of training, target user's needle is obtained
To the sample operations data of full dose sample song;And it according to the sample operations data, is determined from the default full dose song list
Positive sample song is single out and negative sample song is single;And the single corresponding positive score data of positive sample song is obtained, the positive sample song is single
Positive sample is constituted with the positive score data;And obtain the single corresponding negative score data of negative sample song, the negative sample song
The single and described negative score data constitutes negative sample;And the positive sample and the negative sample are utilized, to the default initial single scoring of song
Model is trained, until the initial accuracy rate for singing single Rating Model after training is not less than default accuracy threshold;And by institute
The single Rating Model of initial song after stating training, as the single Rating Model of the default song.
15. a kind of information recommending apparatus, which is characterized in that described device includes: processor, memory and communication bus, described
Memory is communicated by the communication bus with the processor, and the memory stores executable one of the processor
A or multiple programs pass through the processor and execute such as claim 1-7 when one or more of programs are performed
Described in any item methods.
16. a kind of computer readable storage medium, which is characterized in that the computer-readable recording medium storage has program, when
When described program is executed by least one processor, at least one described processor perform claim is caused to require any one of 1-7 institute
The method stated.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811642134.8A CN109408665B (en) | 2018-12-29 | 2018-12-29 | Information recommendation method and device and storage medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811642134.8A CN109408665B (en) | 2018-12-29 | 2018-12-29 | Information recommendation method and device and storage medium |
Publications (2)
Publication Number | Publication Date |
---|---|
CN109408665A true CN109408665A (en) | 2019-03-01 |
CN109408665B CN109408665B (en) | 2021-11-23 |
Family
ID=65462618
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811642134.8A Active CN109408665B (en) | 2018-12-29 | 2018-12-29 | Information recommendation method and device and storage medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109408665B (en) |
Cited By (20)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109982155A (en) * | 2019-03-25 | 2019-07-05 | 北京奇艺世纪科技有限公司 | One kind broadcasting single recommended method and system |
CN109992694A (en) * | 2019-04-17 | 2019-07-09 | 重庆誉存大数据科技有限公司 | A kind of music intelligent recommendation method and system |
CN110363569A (en) * | 2019-06-17 | 2019-10-22 | 深圳壹账通智能科技有限公司 | Data product recommended method, device, computer equipment and storage medium |
CN110599052A (en) * | 2019-09-19 | 2019-12-20 | 携程计算机技术(上海)有限公司 | OTA hotel evaluation method, system, electronic device and medium |
CN110659425A (en) * | 2019-09-25 | 2020-01-07 | 秒针信息技术有限公司 | Resource allocation method and device, electronic equipment and readable storage medium |
CN110910209A (en) * | 2019-11-12 | 2020-03-24 | 腾讯科技(深圳)有限公司 | Data processing method and device and computer readable storage medium |
CN110990621A (en) * | 2019-12-16 | 2020-04-10 | 腾讯科技(深圳)有限公司 | Song recommendation method and device |
CN110996177A (en) * | 2019-11-27 | 2020-04-10 | 北京爱奇艺智慧娱乐科技有限公司 | Video recommendation method, device and equipment for video-on-demand cinema |
CN111078931A (en) * | 2019-12-10 | 2020-04-28 | 腾讯科技(深圳)有限公司 | Singing sheet pushing method and device, computer equipment and storage medium |
CN111383049A (en) * | 2020-02-21 | 2020-07-07 | 中国平安财产保险股份有限公司 | Product recommendation method, device and storage medium |
CN111488484A (en) * | 2020-04-16 | 2020-08-04 | 北京雷石天地电子技术有限公司 | Song recommendation method, medium, and apparatus based on class index distribution and TF-IDF characteristics |
CN111581452A (en) * | 2020-03-26 | 2020-08-25 | 浙江口碑网络技术有限公司 | Method and device for obtaining recommendation object data and electronic equipment |
CN111626615A (en) * | 2020-05-27 | 2020-09-04 | 腾讯音乐娱乐科技(深圳)有限公司 | Method, device and computer storage medium for determining singing order score |
CN112052379A (en) * | 2020-01-20 | 2020-12-08 | 河南紫联物联网技术有限公司 | Intelligent terminal, and recommendation method and system of intelligent household equipment |
CN112231511A (en) * | 2020-10-20 | 2021-01-15 | 腾讯音乐娱乐科技(深圳)有限公司 | Neural network model training method and song mining method and device |
CN113127674A (en) * | 2019-12-31 | 2021-07-16 | 中移(成都)信息通信科技有限公司 | Singing bill recommendation method and device, electronic equipment and computer storage medium |
CN113536027A (en) * | 2021-07-27 | 2021-10-22 | 咪咕音乐有限公司 | Music recommendation method, device, equipment and computer readable storage medium |
CN113570257A (en) * | 2021-07-30 | 2021-10-29 | 北京房江湖科技有限公司 | Index data evaluation method and device based on scoring model, medium and equipment |
CN113688314A (en) * | 2021-08-13 | 2021-11-23 | 今彩慧健康科技(苏州)有限公司 | Physiotherapy store recommendation method and device |
CN110910209B (en) * | 2019-11-12 | 2024-05-31 | 腾讯科技(深圳)有限公司 | Data processing method, device and computer readable storage medium |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103678388A (en) * | 2012-09-19 | 2014-03-26 | 腾讯科技(深圳)有限公司 | Online music recommendation method and device |
CN104090912A (en) * | 2014-06-10 | 2014-10-08 | 腾讯科技(深圳)有限公司 | Information pushing method and device |
CN108595550A (en) * | 2018-04-10 | 2018-09-28 | 南京邮电大学 | A kind of music commending system and recommendation method based on convolutional neural networks |
US20180285896A1 (en) * | 2017-03-31 | 2018-10-04 | International Business Machines Corporation | Determining music to influence customer behavior |
-
2018
- 2018-12-29 CN CN201811642134.8A patent/CN109408665B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103678388A (en) * | 2012-09-19 | 2014-03-26 | 腾讯科技(深圳)有限公司 | Online music recommendation method and device |
CN104090912A (en) * | 2014-06-10 | 2014-10-08 | 腾讯科技(深圳)有限公司 | Information pushing method and device |
US20180285896A1 (en) * | 2017-03-31 | 2018-10-04 | International Business Machines Corporation | Determining music to influence customer behavior |
CN108595550A (en) * | 2018-04-10 | 2018-09-28 | 南京邮电大学 | A kind of music commending system and recommendation method based on convolutional neural networks |
Cited By (29)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109982155A (en) * | 2019-03-25 | 2019-07-05 | 北京奇艺世纪科技有限公司 | One kind broadcasting single recommended method and system |
CN109982155B (en) * | 2019-03-25 | 2021-10-12 | 北京奇艺世纪科技有限公司 | Playlist recommendation method and system |
CN109992694A (en) * | 2019-04-17 | 2019-07-09 | 重庆誉存大数据科技有限公司 | A kind of music intelligent recommendation method and system |
CN110363569A (en) * | 2019-06-17 | 2019-10-22 | 深圳壹账通智能科技有限公司 | Data product recommended method, device, computer equipment and storage medium |
CN110599052A (en) * | 2019-09-19 | 2019-12-20 | 携程计算机技术(上海)有限公司 | OTA hotel evaluation method, system, electronic device and medium |
CN110599052B (en) * | 2019-09-19 | 2023-07-21 | 携程计算机技术(上海)有限公司 | OTA hotel evaluation method, system, electronic equipment and medium |
CN110659425A (en) * | 2019-09-25 | 2020-01-07 | 秒针信息技术有限公司 | Resource allocation method and device, electronic equipment and readable storage medium |
CN110659425B (en) * | 2019-09-25 | 2022-05-17 | 秒针信息技术有限公司 | Resource allocation method and device, electronic equipment and readable storage medium |
CN110910209B (en) * | 2019-11-12 | 2024-05-31 | 腾讯科技(深圳)有限公司 | Data processing method, device and computer readable storage medium |
CN110910209A (en) * | 2019-11-12 | 2020-03-24 | 腾讯科技(深圳)有限公司 | Data processing method and device and computer readable storage medium |
CN110996177A (en) * | 2019-11-27 | 2020-04-10 | 北京爱奇艺智慧娱乐科技有限公司 | Video recommendation method, device and equipment for video-on-demand cinema |
CN111078931A (en) * | 2019-12-10 | 2020-04-28 | 腾讯科技(深圳)有限公司 | Singing sheet pushing method and device, computer equipment and storage medium |
CN111078931B (en) * | 2019-12-10 | 2023-08-01 | 腾讯科技(深圳)有限公司 | Song list pushing method, device, computer equipment and storage medium |
CN110990621B (en) * | 2019-12-16 | 2023-10-13 | 腾讯科技(深圳)有限公司 | Song recommendation method and device |
CN110990621A (en) * | 2019-12-16 | 2020-04-10 | 腾讯科技(深圳)有限公司 | Song recommendation method and device |
CN113127674A (en) * | 2019-12-31 | 2021-07-16 | 中移(成都)信息通信科技有限公司 | Singing bill recommendation method and device, electronic equipment and computer storage medium |
CN113127674B (en) * | 2019-12-31 | 2023-07-21 | 中移(成都)信息通信科技有限公司 | Song list recommendation method and device, electronic equipment and computer storage medium |
CN112052379A (en) * | 2020-01-20 | 2020-12-08 | 河南紫联物联网技术有限公司 | Intelligent terminal, and recommendation method and system of intelligent household equipment |
CN111383049A (en) * | 2020-02-21 | 2020-07-07 | 中国平安财产保险股份有限公司 | Product recommendation method, device and storage medium |
CN111581452A (en) * | 2020-03-26 | 2020-08-25 | 浙江口碑网络技术有限公司 | Method and device for obtaining recommendation object data and electronic equipment |
CN111581452B (en) * | 2020-03-26 | 2023-10-17 | 浙江口碑网络技术有限公司 | Recommendation object data obtaining method and device and electronic equipment |
CN111488484B (en) * | 2020-04-16 | 2023-04-28 | 北京雷石天地电子技术有限公司 | Song recommendation method, medium and device based on class index distribution and TF-IDF characteristics |
CN111488484A (en) * | 2020-04-16 | 2020-08-04 | 北京雷石天地电子技术有限公司 | Song recommendation method, medium, and apparatus based on class index distribution and TF-IDF characteristics |
CN111626615A (en) * | 2020-05-27 | 2020-09-04 | 腾讯音乐娱乐科技(深圳)有限公司 | Method, device and computer storage medium for determining singing order score |
CN112231511A (en) * | 2020-10-20 | 2021-01-15 | 腾讯音乐娱乐科技(深圳)有限公司 | Neural network model training method and song mining method and device |
CN113536027A (en) * | 2021-07-27 | 2021-10-22 | 咪咕音乐有限公司 | Music recommendation method, device, equipment and computer readable storage medium |
CN113570257A (en) * | 2021-07-30 | 2021-10-29 | 北京房江湖科技有限公司 | Index data evaluation method and device based on scoring model, medium and equipment |
CN113688314A (en) * | 2021-08-13 | 2021-11-23 | 今彩慧健康科技(苏州)有限公司 | Physiotherapy store recommendation method and device |
CN113688314B (en) * | 2021-08-13 | 2024-03-19 | 今彩慧健康科技(苏州)有限公司 | Physical therapy store recommending method and device |
Also Published As
Publication number | Publication date |
---|---|
CN109408665B (en) | 2021-11-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109408665A (en) | A kind of information recommendation method and device, storage medium | |
US11921778B2 (en) | Systems, methods and apparatus for generating music recommendations based on combining song and user influencers with channel rule characterizations | |
CN110704674B (en) | Video playing integrity prediction method and device | |
CN108009228A (en) | A kind of method to set up of content tab, device and storage medium | |
CN110532479A (en) | A kind of information recommendation method, device and equipment | |
CN106294830A (en) | The recommendation method and device of multimedia resource | |
CN109543111A (en) | Recommendation information screening technique, device, storage medium and server | |
CN107273436A (en) | The training method and trainer of a kind of recommended models | |
CN110428298A (en) | A kind of shop recommended method, device and equipment | |
CN110263189A (en) | Recommended method, device, storage medium and the computer equipment of media content | |
WO2018040069A1 (en) | Information recommendation system and method | |
CN105975472A (en) | Method and device for recommendation | |
CN106446078A (en) | Information recommendation method and recommendation apparatus | |
CN106454536B (en) | The determination method and device of information recommendation degree | |
CN104991899A (en) | Identification method and apparatus of user property | |
CN107832437A (en) | Audio/video method for pushing, device, equipment and storage medium | |
CN103167330A (en) | Method and system for audio/video recommendation | |
CN105426514A (en) | Personalized mobile APP recommendation method | |
CN106168980A (en) | Multimedia resource recommends sort method and device | |
CN103455538B (en) | Information processing unit, information processing method and program | |
CN107885745A (en) | A kind of song recommendations method and device | |
Li et al. | A method of purchase prediction based on user behavior log | |
CN103942328B (en) | A kind of video retrieval method and video-unit | |
CN108304399A (en) | The recommendation method and device of Web content | |
CN108665148B (en) | Electronic resource quality evaluation method and device and storage medium |
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 |