CN110321478A - A kind of information recommendation method, device, equipment and medium - Google Patents
A kind of information recommendation method, device, equipment and medium Download PDFInfo
- Publication number
- CN110321478A CN110321478A CN201910443788.6A CN201910443788A CN110321478A CN 110321478 A CN110321478 A CN 110321478A CN 201910443788 A CN201910443788 A CN 201910443788A CN 110321478 A CN110321478 A CN 110321478A
- Authority
- CN
- China
- Prior art keywords
- user
- target
- classification
- information
- data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000009826 distribution Methods 0.000 claims abstract description 34
- 239000000284 extract Substances 0.000 claims abstract description 9
- 230000006399 behavior Effects 0.000 claims description 103
- 230000003542 behavioural effect Effects 0.000 claims description 31
- 230000015654 memory Effects 0.000 claims description 15
- 238000003860 storage Methods 0.000 claims description 11
- 230000004044 response Effects 0.000 claims description 7
- 238000009499 grossing Methods 0.000 claims description 5
- 239000012141 concentrate Substances 0.000 claims description 4
- 238000007405 data analysis Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 230000000875 corresponding effect Effects 0.000 description 37
- 238000010586 diagram Methods 0.000 description 13
- 238000012545 processing Methods 0.000 description 10
- 239000011159 matrix material Substances 0.000 description 6
- 230000006870 function Effects 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000004891 communication Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000001914 filtration Methods 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000000354 decomposition reaction Methods 0.000 description 2
- 239000004744 fabric Substances 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000033001 locomotion Effects 0.000 description 2
- 239000000047 product Substances 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 241001269238 Data Species 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 238000009825 accumulation Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 238000002360 preparation method Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
- 239000002356 single layer Substances 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000001360 synchronised effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
Landscapes
- Engineering & Computer Science (AREA)
- Databases & Information Systems (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of information recommendation method, device, equipment and media, and the method includes obtaining user behavior data, the user behavior data is for characterizing user to the interest level of various preliminary classifications;The reference number for generating each preliminary classification extracts target classification from the preliminary classification according to the reference number;The generation of the reference number meets probability distribution, and the user behavior data of the corresponding preliminary classification of the probability distribution is related;Corresponding target information is obtained in each target classification to obtain target information collection;The information extracted for recommendation is concentrated from the target information.The statistical result of the generation of reference number of the present invention meets user behavior data, so that user rarely has preliminary classification that chance touches or less interested to be chosen for target classification compared with small probability, the probing that interest other for user are completed on the basis of taking into account user and being both interested in, avoids the content side recommended for user more and more narrow.
Description
Technical field
The present invention relates to information recommendation field more particularly to a kind of information recommendation method, device, equipment and media.
Background technique
The recommendation list drawn a portrait based on user is obtained usually using various strategies of recalling in existing recommender system, and is based on
Recommendation is presented to user in recommendation list.And there are following problems for this technical solution:
(1) recall strategy be based on user portrait recall, and user portrait be merely able to description user historical behavior,
Therefore the historical behavior for recalling result only with reference to user of strategy is recalled, and can not be the behavior in user in predicting future.Such as it is every
The secondary article recalled all is article relevant with the click behavior of the history of user, and the related article read before user can go out always
It is existing, and user has no chance to read other articles.Recall strategy can not to user not in contact with to content recall, to lead
Cause recommendation covering surface more and more narrow.
(2) every kind in recommender system is recalled strategy and is their respective businesses, do not plan as a whole it is various recall strategy, also can not be based on each
Kind recalls the corresponding result of recalling of strategy and does global consideration to obtain objective recommendation.
Summary of the invention
In order to solve only to consider in the prior art that user's portrait is recommended and causes recommendation covering surface more and more narrow
The technical issues of, the embodiment of the present invention provides a kind of information recommendation method, device, equipment and medium.
On the one hand, the present invention provides a kind of information recommendation methods, which comprises
User behavior data is obtained, the user behavior data is for characterizing user to the journey interested of various preliminary classifications
Degree;
The reference number for generating each preliminary classification extracts target classification from the preliminary classification according to the reference number;
The generation of the reference number meets probability distribution, and the user behavior data of the corresponding preliminary classification of the probability distribution
It is related;
Corresponding target information is obtained in each target classification to obtain target information collection;
The information extracted for recommendation is concentrated from the target information.
On the other hand, the present invention provides a kind of information recommending apparatus, and described device includes:
User behavior data obtains module, and for obtaining user behavior data, the user behavior data is for characterizing
Interest level of the user to various preliminary classifications;
User behavior data analysis module, for generating the reference number of each preliminary classification, according to the reference number from institute
It states and extracts target classification in preliminary classification;The generation of the reference number meets probability distribution, and the probability distribution is right with it
The user behavior data for the preliminary classification answered is related;
Target information collection obtains module, for obtaining in each target classification corresponding target information to obtain target information
Collection;
Recommendation information extraction module, for concentrating the information extracted for recommendation from the target information.
On the other hand, the present invention provides a kind of equipment, which is characterized in that and the equipment includes processor and memory,
It is stored at least one instruction, at least a Duan Chengxu, code set or instruction set in the memory, at least one instruction,
An at least Duan Chengxu, the code set or instruction set are loaded by the processor and are executed to realize a kind of information recommendation side
Method.
On the other hand, the present invention provides a kind of computer storage mediums, which is characterized in that stores in the storage medium
There are at least one instruction, at least a Duan Chengxu, code set or instruction set, at least one instruction, at least a Duan Chengxu, code
Collection or instruction set are loaded by processor and execute a kind of information recommendation method.
The present invention provides a kind of information recommendation method, device, equipment and media.By to user behavior number in the present invention
The target classification that not only compatible subscribers draw a portrait but also can spy user interest has been obtained according to analysis, has been called together to realize a kind of exploration
It returns, and each information in target classification is extracted again, to realize the recommendation of pool.Further, this hair
The exploration of bright offer, which is recalled, makes the organic information that can be appreciated that under the classification not being exposed of user, realizes and recalls result to each road
Global optimization modeling, improve the accuracy rate and coverage rate that information is recalled.By constantly excavating the interest of user, to widen use
The diversity of family recommendation information and rich, avoids the more and more narrow problem of the covering surface of recommendation information.
Detailed description of the invention
It in order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology and advantage, below will be to implementation
Example or attached drawing needed to be used in the description of the prior art are briefly described, it should be apparent that, the accompanying drawings in the following description is only
It is only some embodiments of the present invention, for those of ordinary skill in the art, without creative efforts,
It can also be obtained according to these attached drawings other attached drawings.
Fig. 1 is a kind of schematic diagram of implementation environment provided by the invention;
Fig. 2 is a kind of information recommendation method flow chart provided by the invention;
Fig. 3 is the preliminary classification schematic diagram provided by the invention in news recommendation scene;
Fig. 4 is the preliminary classification schematic diagram provided by the invention in book recommendation scene;
Fig. 5 is the preliminary classification schematic diagram provided by the invention in video recommendations scene;
Fig. 6 is the method flow diagram that a kind of pair of user behavior data provided by the invention is smoothly changed;
Fig. 7 is the reference number flow chart provided by the invention for generating each preliminary classification;
Fig. 8 is provided by the invention to extract target classification flow chart from the preliminary classification according to the reference number;
Fig. 9 is provided by the invention to obtain in each target classification corresponding target information to obtain target information collection process
Figure;
Figure 10 is a kind of schematic diagram of information recommendation method provided by the invention;
Figure 11 is a schematic diagram of the embodiment of the present invention provided by the invention in news scenes in user terminal;
Figure 12 is a kind of information recommending apparatus block diagram provided by the invention;
Figure 13 is that user behavior data provided by the invention smoothly changes module frame chart;
Figure 14 is a kind of hardware knot of equipment for realizing method provided by the embodiment of the present invention provided by the invention
Structure schematic diagram.
Specific embodiment
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation description, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, those of ordinary skill in the art without making creative work it is obtained it is all its
His embodiment, shall fall within the protection scope of the present invention.
It should be noted that description and claims of this specification and term " first " in above-mentioned attached drawing, "
Two " etc. be to be used to distinguish similar objects, without being used to describe a particular order or precedence order.It should be understood that using in this way
Data be interchangeable under appropriate circumstances, so as to the embodiment of the present invention described herein can in addition to illustrating herein or
Sequence other than those of description is implemented.In addition, term " includes " and " having " and their any deformation, it is intended that cover
Cover it is non-exclusive include, for example, containing the process, method of a series of steps or units, system, product or server need not limit
In step or unit those of is clearly listed, but may include be not clearly listed or for these process, methods, produce
The other step or units of product or equipment inherently.
In order to which objects, technical solutions and advantages disclosed by the embodiments of the present invention are more clearly understood, below in conjunction with attached drawing
And embodiment, the embodiment of the present invention is further elaborated.It should be appreciated that specific embodiment described herein is only used
To explain the embodiment of the present invention, it is not intended to limit the present invention embodiment.
Fig. 1 is a kind of schematic diagram of implementation environment provided in an embodiment of the present invention, and referring to Fig. 1, which includes: to push away
Server 01 is recommended, at least one recommends terminal 02, and the recommendation server 01 is communicated to connect with the recommendation terminal 02.
The recommendation terminal 02 can based on Browser/Server Mode (Browser/Server, B/S) or client/
Server mode (Client/Server, C/S) is communicated with recommendation server 01.The recommendation terminal 02 may include: intelligence
The entity device of the types such as energy mobile phone, tablet computer, laptop, digital assistants, intelligent wearable device, car-mounted terminal,
It also may include the software run in entity device, such as application program etc..For example, the recommendation terminal can run news
Recommend class software, video recommendations class software, microblogging recommends class software, and class software is recommended in shopping.
The recommendation server 01 is used to obtain each user's portrait for recommending the corresponding client of terminal 02, and is based on user
Portrait obtains recommendation, and recommendation is pushed to each recommendation terminal 02.Recommendation server 01 may include one only
Found the server run perhaps distributed server or the server cluster being made of multiple servers.
Referring to FIG. 2, it illustrates a kind of information recommendation method provided in an embodiment of the present invention, the method can more than
Stating the recommendation server in implementation environment is subject of implementation, which comprises
S101. user behavior data is obtained, the user behavior data is for characterizing sense of the user to various preliminary classifications
Level of interest.
Specifically, the preliminary classification may be different under different scenes.Preliminary classification can be specific field
The classification of most coarseness under scape.
As shown in figure 3, recommending in scene in news, the information classification that news is recommended is preliminary classification.News it is initial
Classification is the first-level class of news pages, includes society, amusement, news, sport, life range, big Suzhou, finance, the people in Fig. 3
It is raw.
As shown in figure 4, the classification information of books is preliminary classification in book recommendation scene.In Fig. 4 include describing love affairs,
Swordsman, mythology, science fiction, grave-robbery, reasoning.
As shown in figure 5, the classification information of video is preliminary classification in video recommendations scene.In Fig. 5 include self-control,
Ancient costume, describing love affairs, idol, family, youth, city, comedy, spy war.
In a feasible embodiment, the user behavior data can not select value evidence and user by user
Selection value is according to indicating.Specifically, the value evidence of not selecting of the user is the data exposed to user and the selection of the user
It is worth the difference of evidence.User behavior data is a kind of user's portrait, and which characterizes interest of the user for each preliminary classification to take
To.User, which draws a portrait, to be depicted according to the historical behavior of user come various dimensions user behavior is accustomed to the description data of attribute.
The user behavior data can realize real-time storage by Spark platform real-time capture, and based on Red i s.
Spark platform is the computing engines for the Universal-purpose quick for aiming at large-scale data processing and designing.Spark is that one kind is opened
Source cluster computing environment, Spark enables memory distributed data collection, and other than being capable of providing interactive inquiry, it can be with excellent
Change iteration workload.
Red i s is a kind of storage system based on key-value pair.Red i s supports master-slave synchronisation.Data can be from main clothes
Business device is synchronized from server to any number of, be can be from server and is associated with other primary servers from server.This makes
Obtaining Redis can be performed single layer tree copying.Deposit consciously or unconsciously can carry out write operation to data.Due to fully achieving hair
Cloth/subscribing mechanism so that from database anywhere synchronization tree when, a channel can be subscribed to and to receive primary server complete
News release record.It synchronizes helpful to the scalability and data redundancy of read operation.
Specifically, user can realize the selection of data by the various motions such as click, sliding, circle choosing or gesture.
In a feasible embodiment, can for some user in some period for some preliminary classification
User behavior data is by user behavior data to expression, i.e., user behavior data is not to (a: b: selection value selects value), accordingly
Some user in multiple periods (or the time interval being made of multiple continuous periods) for some preliminary classification
User behavior data can by user behavior data to sequence indicate.Accordingly, for each user in each preliminary classification
In user behavior sequence can also be indicated by user behavior data.
For example, with day for a period, user behavior data of some user to some preliminary classification in continuous 30 days
It can be by including that the user behavior data of 30 user behavior datas pair indicates sequence.
In a preferred embodiment, for the exposed preliminary classification of user its user behavior data to note
Record true user behavior, and for not to its user behavior data of the exposed preliminary classification of user to recording virtual user
Behavior, the virtual user behavior indicate a kind of virtual user behavior with the lesser data of absolute value.Such as
Not to the exposed preliminary classification of user<amusement>, for user behavior data to that can be (0,0.001), 0 indicates selection value,
0.001 is expressed as selection value, correspondingly, its impression is 0+0.001=0.001.
Since interest orientation of the user for each preliminary classification has apparent subjective tendency, user is in difference
User behavior data centering numerical value under classification differs greatly, and for a user behavior data to different in sequence
User behavior data pair, numerical fluctuations are also biggish.Each selection behavior of user does not select behavior that can cause certain
The change of the data of a user behavior data centering, the otherness of each user behavior data pair may also be further aggravated in this.
More smooth user behavior data in order to obtain, in user behavior data generation phase, the embodiment of the present invention discloses a kind of right
The method that user behavior data is smoothly changed, as shown in Figure 6, comprising:
S1. in response to user instruction, target user's behavioral data pair that the user instruction is directed toward is determined.
Specifically, user instruction can for some or certain information in some target preliminary classification optional directive or
Non- optional directive.The corresponding user behavior data of the target preliminary classification is to for target user's behavioral data pair.
S2. it obtains the selection value of target user's behavioral data pair and does not select the total value of value.
S3. according to the user instruction, the target data to be changed of target user's behavioral data centering is determined.
Specifically, target user's behavioral data is to (selection value, do not select value) composition.If user instruction is one
Optional directive, for example click some information chart in target preliminary classification, then target data to be changed is target use
The selection value of family behavioral data centering;If user instruction is a non-optional directive, for example directly closes target preliminary classification
The corresponding page of all information, then target data to be changed is that value is not selected in target user's behavioral data centering.
If S4. the total value is not more than preset threshold, the data to be changed increase one certainly.
If S5. the total value is greater than preset threshold, the data to be changed from increasing one, multiplied by with the default threshold
It is worth relevant smoothing factor.
Specifically, by taking the preset threshold is 30 as an example.The centering of target user's behavioral data indicates selection value with a, with b table
Show and do not select value, as (a+b)≤30, in response to optional directive a+1, in response to non-optional directive b+1;As (a+b) > 30,
In response to optional directive (a+1) * (30/31), in response to non-optional directive (b+1) * (30/31), wherein 30/31 is smooth system
Number.
Such effect may be implemented in this smooth change: with the accumulation of user behavior data, selection value and not selecting
Value and be all not more than preset threshold, and be able to reflect the behavior interesting of user, wherein the preset threshold can according to when
Between the effect of scene set.
S103. the reference number for generating each preliminary classification extracts target from the preliminary classification according to the reference number
Classification;The generation of the reference number meets probability distribution, and user's row of the corresponding preliminary classification of the probability distribution
For data correlation.
Specifically, the user behavior data can be the corresponding user behavior number of each preliminary classification in the default time limit
According to sequence.If the default time limit is 30 days, the user behavior data is based on the row in user nearly 30 days to sequence
For obtained from.
Specifically, the target classification can be according to the corresponding user behavior data of each preliminary classification, based on symbol
It closes the data processed result of statistical law and obtains.The target classification is that can be used for comprehensive point for spying user interest
Class not only may include user's interested classification in the past, but also may include the classification that user rarely had chance to touch in the past.
Specifically, in a feasible embodiment, the reference number for generating each preliminary classification, as shown in fig. 7, packet
It includes:
S1031. the corresponding target user's behavioral data pair of each preliminary classification, institute are obtained according to the user behavior data
Target user's behavioral data is stated to including selection value and do not select value.
Specifically, target user's behavioral data pair and preliminary classification are one-to-one relationship, target user's row
It is data to including the selection value to the preliminary classification and do not select value.
In a feasible embodiment, the selection value of the corresponding target user's behavioral data centering of the preliminary classification
It can be the corresponding user behavior data of the preliminary classification to the summation of user behavior data centering selection value each in sequence;
Correspondingly, the corresponding target user's behavioral data centering of the preliminary classification does not select value corresponding for the preliminary classification
User behavior data do not select the summation of value to user behavior data centering each in sequence.
In another feasible embodiment, the selection of the corresponding target user's behavioral data centering of the preliminary classification
Value can be based on the default method of sampling from the corresponding user behavior data of the preliminary classification to user behavior number each in sequence
According to pair selection value in extract;Correspondingly, the corresponding target user's behavioral data centering of the preliminary classification does not select value can
To be based on the default method of sampling from the corresponding user behavior data of the preliminary classification to user behavior number each in sequence
According to pair do not select in value and extract.
S1033. its corresponding probability distribution is based on for each preliminary classification and generates a reference number, the probability distribution
Desired value target user's behavior centering selection value corresponding with the preliminary classification divided by selection value and not selecting value summation
As a result related.
Specifically, the desired value can be equal to, be proportional to or be positively correlated with the corresponding target user of the preliminary classification
Behavior centering selection value is divided by selection value and the result for not selecting value summation.
Specifically, the probability distribution can for beta probability distribution, Gamma distribution (Gamma Distribution) or
Other, the embodiment of the present invention does not limit specific form of probability.
Beta distribution (Beta Distribution) is a conjugate prior as Bernoulli Jacob's distribution and binomial distribution
The density function of distribution has important application in machine learning and mathematical statistics.In probability theory, beta distribution, also referred to as B points
Cloth refers to one group of continuous probability distribution for being defined on (0,1) section.
Gamma distribution is one of statistics continuous probability function, includes two parameter alphas and β, wherein α is known as shape ginseng
Number, β are known as scale parameter.
Correspondingly, described extract target classification from the preliminary classification according to the reference number, as shown in Figure 8, comprising:
S1035. it is ranked up according to reference number descending each preliminary classification corresponding to reference number.
S1037. target classification is extracted according to ranking results.
Specifically, the TOPN of the ranking results can be chosen as target classification.For example, if having 20 in ranking results
Preliminary classification, N=5, then choosing 5 preliminary classifications of sequence up front is target classification.N can be carried out based on actual conditions
Adaptability selection.
It can be generated based on user behavior data for the reference number as sequence reference, and refer in the embodiment of the present invention
The statistical result of several generations meets user behavior data again, so that the interested preliminary classification of user is with greater probability quilt
It is chosen for target classification, and user rarely has preliminary classification that chance touches or less interested to be selected compared with small probability
For target classification, but it is unlikely to absolutely not chance again and is chosen for target classification.To take into account what user had both been interested in
On the basis of complete the probing of other for user interest, avoid the content side recommended for user more and more narrow.
S105. corresponding target information is obtained in each target classification to obtain target information collection.
Specifically, corresponding target information is in each target classification of acquisition to obtain target information collection, such as Fig. 9 institute
Show, comprising:
S1051. the first classification and the second classification in the target classification are obtained;Described first is classified as the information that is called back
The classification of hit;Described second is classified as removing other classification outside first classification in target classification.
Specifically, strategy can be recalled in the embodiment of the present invention based at least one and obtains call back message from all angles.
The strategy of recalling includes but is not limited to that hot spot is recalled, region is recalled, interest is recalled, clicks sequence recalls, matrix decomposition is recalled
One of or a variety of combinations.
Hot spot is recalled: recalling global newest hot information for recommendation of personalized information.
It recalls region: relevant information is recalled to be used for according to the permanent residence of user, local and real-time geographical location information
Recommendation of personalized information.
Interest is recalled: being clicked behavior according to the history of user and is recalled relevant information for recommendation of personalized information.Specifically
Ground, the interest, which is recalled, can be used the collaborative filtering of ICF and UCF and recalls.ICF refers to Item Collaborative
Filtering is to carry out analysis recommendation from information viewpoint.UCF refers to User Collaborative Filtering, and referring to is
The similar people that knows how to behave in a delicate situation is found out from user perspective, is recommended.
Matrix decomposition is recalled: user and project being expressed as to the form of two-dimensional matrix, regard recommendation task as a square
The task of battle array completion (Matrix Completion), i.e., do not generated based on data existing in matrix to fill up in matrix
The element of record, to obtain recommendation results.
Click sequence to recall: the click sequence based on user is recalled.
According to recalling available at least one information of strategy, and every information has its corresponding preliminary classification, by root
It is first-level class according to the classification for recalling the information hit that strategy obtains, other target classifications are secondary classification.
By taking target classification includes 10 preliminary classifications as an example, target classification include<life><international><domestic><Suzhou><
Amusement><sport><finance and economics><society><joke><automobile>, and being classified as of the information hit that the strategy that is called back obtains<amusement><
Sport><finance and economics><society><joke><automobile>, then secondary classification is<life><international><domestic><Suzhou>.
S1053. it is the information of each the first quantity of first classifying and selecting to obtain first object information collection, is each second
The information of the second quantity of classifying and selecting is to obtain the second target information collection.
First quantity can be identical or different with the second quantity.The selection of information can be from base in first classification
Recommendation or the higher information of temperature are chosen in the information that strategy is recalled in recalling.The selection of information can be based in second classification
Other users to it is described second classification in information access temperature, or be based on actual conditions (such as real-time, significance level)
It is chosen.
S1055. it integrates the first object information collection and the second target information collection obtains target information collection.
S107. the information extracted for recommendation is concentrated from the target information.
Specifically, can be upset based on the intrinsic sequence for the information that preset algorithm concentrates the target information, then from
The middle information choosing user and recommending, or the random information extracted for recommendation is concentrated in the target information.The present invention is implemented
Example does not limit the quantity of the information for recommendation.The upset inherently the sorted or effect extracted at random is so that target
Each information that information is concentrated is recommended with same or similar probability, to guarantee the covering surface of information recommended to the user.
The target classification for spying user interest has been obtained by analyzing user behavior data in the embodiment of the present invention,
It is recalled to realize a kind of exploration.Referring to FIG. 10, it illustrates information recommendation methods a kind of in the embodiment of the present invention
Schematic diagram, by increasing the logic that exploration is recalled in intrinsic supplement on the basis of recalling strategy, and will be based on exploration
The result recalled is upset or is broken up together, also has an opportunity to be recommended to user so that exploration recalls result.
Please refer to Figure 11, it illustrates in news scenes the embodiment of the present invention in a schematic diagram of user terminal.With
Family often browses<amusement><content of the classification such as life><automobile>, it is also possible to recommend the hot spot about<house property>for user
Information.News scenes are not terminated in, the embodiment of the present invention can also be widely used in video, microblogging, music, game, shopping etc.
Among the various scenes for needing to carry out information recommendation.
A kind of information recommendation method disclosed by the embodiments of the present invention increases the logic for exploring user interest, is recommending every time
Exploration optimization is done to Shi Douneng to the interest of user, makes the organic information that can be appreciated that under the classification not being exposed of user, realizes
The global optimization modeling that result is recalled to each road, improves the accuracy rate and coverage rate that information is recalled.By constantly excavating user
Interest, the diversity of Lai Tuokuan user's recommendation information and rich, the covering surface for avoiding recommendation information more and more narrow asks
Topic.
The embodiment of the present invention also provides a kind of information recommending apparatus, and as shown in figure 12, described device includes:
User behavior data obtains module 201, and for obtaining user behavior data, the user behavior data is used for table
Family is taken over for use to the interest level of various preliminary classifications;
User behavior data analysis module 203, for generating the reference number of each preliminary classification, according to the reference number from
Target classification is extracted in the preliminary classification;The generation of the reference number meets probability distribution, and the probability distribution and its
The user behavior data of corresponding preliminary classification is related;
Target information collection obtains module 205, for obtaining in each target classification corresponding target information to obtain target
Information collection;
Recommendation information extraction module 207, for concentrating the information extracted for recommendation from the target information.
User behavior data smoothly changes module 209, and the user behavior data smoothly changes module 209, such as Figure 13 institute
Show, comprising:
Target user's behavior that target user's behavioral data is directed toward to determination unit 2091, for determining the user instruction
Data pair;
Total value computing unit 2093, for obtaining the selection value of target user's behavioral data pair and not selecting value
Total value;
Target data determination unit 2095 to be changed, for determining target user's behavior according to the user instruction
The target data to be changed of data centering;
First branch units 2097, if being not more than preset threshold for the total value, the data to be changed increase certainly
One;
Second branch units 2099, if being greater than preset threshold for the total value, the data to be changed increase one certainly,
Multiplied by smoothing factor relevant to the preset threshold.
Specifically, a kind of information recommending apparatus and embodiment of the method described in the embodiment of the present invention are based on identical invention structure
Think.
The embodiment of the invention also provides a kind of computer storage medium, the computer storage medium can store more
Item instruction, described instruction are suitable for being loaded as processor and being executed a kind of the various of information recommendation method described in the embodiment of the present invention
Step, specific implementation procedure include:
A kind of information recommendation method, which comprises
User behavior data is obtained, the user behavior data is for characterizing user to the journey interested of various preliminary classifications
Degree;
The reference number for generating each preliminary classification extracts target classification from the preliminary classification according to the reference number;
The generation of the reference number meets probability distribution, and the user behavior data of the corresponding preliminary classification of the probability distribution
It is related;
Corresponding target information is obtained in each target classification to obtain target information collection;
The information extracted for recommendation is concentrated from the target information.
It further, further include smoothly being changed to user behavior data before obtaining user behavior data, it is described
User behavior data is smoothly changed, comprising:
In response to user instruction, target user's behavioral data pair that the user instruction is directed toward is determined;
It obtains the selection value of target user's behavioral data pair and does not select the total value of value;
According to the user instruction, the target data to be changed of target user's behavioral data centering is determined;
If the total value is not more than preset threshold, the data to be changed increase one certainly;
If the total value is greater than preset threshold, the data to be changed from increasing one, multiplied by with the preset threshold
Relevant smoothing factor.
Further, the reference number for generating each preliminary classification, comprising:
The corresponding target user's behavioral data pair of each preliminary classification, the target are obtained according to the user behavior data
User behavior data is to including selection value and do not select value;
Its corresponding probability distribution is based on for each preliminary classification and generates a reference number, the expectation of the probability distribution
Value target user's behavior centering selection value corresponding with the preliminary classification is divided by selection value and the result phase for not selecting value summation
It closes.
It is further, described that target classification is extracted from the preliminary classification according to the reference number, comprising:
It is ranked up according to reference number descending each preliminary classification corresponding to reference number;
Target classification is extracted according to ranking results.
It is further, described to obtain in each target classification corresponding target information to obtain target information collection, comprising:
Obtain the first classification and the second classification in the target classification;Described first is classified as the information hit that is called back
Classification;Described second is classified as removing other classification outside first classification in target classification;
It is the information of each the first quantity of first classifying and selecting to obtain first object information collection, for each second classification choosing
The information of the second quantity is taken to obtain the second target information collection;
The comprehensive first object information collection and the second target information collection obtain target information collection.
It is further, described that the information extracted for recommendation is concentrated from the target information, comprising:
Intrinsic sequence upset based on the information that preset algorithm concentrates the target information, therefrom chooses what user recommended
Information, or, concentrating the random information extracted for recommendation in the target information.
Further, Figure 14 shows a kind of hardware knot of equipment for realizing method provided by the embodiment of the present invention
Structure schematic diagram, the equipment can participate in constituting or comprising device provided by the embodiment of the present invention.As shown in figure 14, equipment 10
It may include that (processor 102 can be with for one or more (using 102a, 102b ... ... in figure, 102n to show) processors 102
The including but not limited to processing unit of Micro-processor MCV or programmable logic device FPGA etc.), memory for storing data
104 and for communication function transmitting device 106.It in addition to this, can also include: display, input/output interface (I/
O Interface), the port universal serial bus (USB) (a port that can be used as in the port of I/O interface is included), network connect
Mouth, power supply and/or camera.It will appreciated by the skilled person that structure shown in Figure 14 is only to illustrate, not to upper
The structure for stating electronic device causes to limit.For example, equipment 10 may also include the more or less component than shown in Figure 14, or
Person has the configuration different from shown in Figure 14.
It is to be noted that said one or multiple processors 102 and/or other data processing circuits lead to herein
Can often " data processing circuit " be referred to as.The data processing circuit all or part of can be presented as software, hardware, firmware
Or any other combination.In addition, data processing circuit can set for single independent processing module or all or part of be integrated to
In any one in other elements in standby 10 (or mobile devices).As involved in the embodiment of the present application, the data
Processing circuit controls (such as the selection for the variable resistance end path connecting with interface) as a kind of processor.
Memory 104 can be used for storing the software program and module of application software, as described in the embodiment of the present invention
Corresponding program instruction/the data storage device of method, the software program that processor 102 is stored in memory 104 by operation
And module realizes a kind of above-mentioned information recommendation method thereby executing various function application and data processing.Memory
104 may include high speed random access memory, may also include nonvolatile memory, and such as one or more magnetic storage device dodges
It deposits or other non-volatile solid state memories.In some instances, memory 104 can further comprise relative to processor
102 remotely located memories, these remote memories can pass through network connection to equipment 10.The example of above-mentioned network includes
But be not limited to internet, intranet, local area network, mobile radio communication and combinations thereof.
Transmitting device 106 is used to that data to be received or sent via a network.Above-mentioned network specific example may include
The wireless network that the communication providers of equipment 10 provide.In an example, transmitting device 106 includes a network adapter
(NetworkInterfaceController, NIC), can be connected by base station with other network equipments so as to internet
It is communicated.In an example, transmitting device 106 can be radio frequency (RadioFrequency, RF) module, be used to pass through
Wireless mode is communicated with internet.
Display can such as touch-screen type liquid crystal display (LCD), the liquid crystal display aloow user with
The user interface of equipment 10 (or mobile device) interacts.
It should be understood that embodiments of the present invention sequencing is for illustration only, do not represent the advantages or disadvantages of the embodiments.
And above-mentioned this specification specific embodiment is described.Other embodiments are within the scope of the appended claims.One
In a little situations, the movement recorded in detail in the claims or step can be executed according to the sequence being different from embodiment and
Still desired result may be implemented.In addition, process depicted in the drawing not necessarily requires the particular order shown or company
Continuous sequence is just able to achieve desired result.In some embodiments, multitasking and parallel processing it is also possible or
It may be advantageous.
All the embodiments in this specification are described in a progressive manner, same and similar portion between each embodiment
Dividing may refer to each other, and each embodiment focuses on the differences from other embodiments.Especially for device and
For server example, since it is substantially similar to the method embodiment, so being described relatively simple, related place is referring to side
The part of method embodiment illustrates.
Those of ordinary skill in the art will appreciate that realizing that all or part of the steps of above-described embodiment can pass through hardware
It completes, relevant hardware can also be instructed to complete by program, the program can store in a kind of computer-readable
In storage medium, storage medium mentioned above can be read-only memory, disk or CD etc..
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all in spirit of the invention and
Within principle, any modification, equivalent replacement, improvement and so on be should all be included in the protection scope of the present invention.
Claims (10)
1. a kind of information recommendation method, which is characterized in that the described method includes:
User behavior data is obtained, the user behavior data is for characterizing user to the interest level of various preliminary classifications;
The reference number for generating each preliminary classification extracts target classification from the preliminary classification according to the reference number;It is described
The generation of reference number meets probability distribution, and the user behavior data phase of the corresponding preliminary classification of the probability distribution
It closes;
Corresponding target information is obtained in each target classification to obtain target information collection;
The information extracted for recommendation is concentrated from the target information.
2. the method according to claim 1, wherein further including to user before obtaining user behavior data
Behavioral data is smoothly changed, described smoothly to be changed to user behavior data, comprising:
In response to user instruction, target user's behavioral data pair that the user instruction is directed toward is determined;
It obtains the selection value of target user's behavioral data pair and does not select the total value of value;
According to the user instruction, the target data to be changed of target user's behavioral data centering is determined;
If the total value is not more than preset threshold, the data to be changed increase one certainly;
If the total value is greater than preset threshold, the data to be changed are from increasing one, multiplied by related to the preset threshold
Smoothing factor.
3. the method according to claim 1, wherein the reference number for generating each preliminary classification, comprising:
The corresponding target user's behavioral data pair of each preliminary classification, the target user are obtained according to the user behavior data
Behavioral data is to including selection value and do not select value;
Its corresponding probability distribution is based on for each preliminary classification and generates a reference number, the desired value of the probability distribution with
The corresponding target user's behavior centering selection value of the preliminary classification is related to the result of value summation is not selected divided by selection value.
4. according to the method described in claim 3, it is characterized in that, described mention from the preliminary classification according to the reference number
Take target classification, comprising:
It is ranked up according to reference number descending each preliminary classification corresponding to reference number;
Target classification is extracted according to ranking results.
5. the method according to claim 1, wherein described obtain corresponding target information in each target classification
To obtain target information collection, comprising:
Obtain the first classification and the second classification in the target classification;Described first point for being classified as being called back information hit
Class;Described second is classified as removing other classification outside first classification in target classification;
It is the information of each the first quantity of first classifying and selecting to obtain first object information collection, is each second classifying and selecting the
The information of two quantity is to obtain the second target information collection;
The comprehensive first object information collection and the second target information collection obtain target information collection.
6. the method according to claim 1, wherein described concentrate from the target information is extracted for recommendation
Information, comprising:
Intrinsic sequence upset based on the information that preset algorithm concentrates the target information, therefrom chooses the letter that user recommends
Breath, or, concentrating the random information extracted for recommendation in the target information.
7. a kind of information recommending apparatus, which is characterized in that described device includes:
User behavior data obtains module, and for obtaining user behavior data, the user behavior data is for characterizing user
To the interest level of various preliminary classifications;
User behavior data analysis module, for generating the reference number of each preliminary classification, according to the reference number from it is described just
Begin to extract target classification in classification;The generation of the reference number meets probability distribution, and the probability distribution is corresponding
The user behavior data of preliminary classification is related;
Target information collection obtains module, for obtaining in each target classification corresponding target information to obtain target information collection;
Recommendation information extraction module, for concentrating the information extracted for recommendation from the target information.
8. device according to claim 7, which is characterized in that it further include that user behavior data smoothly changes module, it is described
User behavior data smoothly changes module, comprising:
Target user's behavioral data pair that target user's behavioral data is directed toward to determination unit, for determining the user instruction;
Total value computing unit, for obtaining the selection value of target user's behavioral data pair and not selecting the total value of value;
Target data determination unit to be changed, for determining target user's behavioral data centering according to the user instruction
Target data to be changed;
First branch units, if being not more than preset threshold for the total value, the data to be changed increase one certainly;
Second branch units, if being greater than preset threshold for the total value, the data to be changed from increasing one, multiplied by with
The relevant smoothing factor of the preset threshold.
9. a kind of equipment, which is characterized in that the equipment includes processor and memory, and at least one is stored in the memory
Item instruction, at least a Duan Chengxu, code set or instruction set, at least one instruction, an at least Duan Chengxu, the code
Collection or instruction set are loaded by the processor and are executed to realize a kind of information recommendation side as claimed in any one of claims 1 to 6
Method.
10. a kind of computer storage medium, which is characterized in that be stored at least one instruction, at least one in the storage medium
Duan Chengxu, code set or instruction set, at least one instruction, an at least Duan Chengxu, code set or instruction set are added by processor
It carries and executes a kind of information recommendation method as claimed in any one of claims 1 to 6.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910443788.6A CN110321478A (en) | 2019-05-27 | 2019-05-27 | A kind of information recommendation method, device, equipment and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201910443788.6A CN110321478A (en) | 2019-05-27 | 2019-05-27 | A kind of information recommendation method, device, equipment and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN110321478A true CN110321478A (en) | 2019-10-11 |
Family
ID=68119159
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201910443788.6A Pending CN110321478A (en) | 2019-05-27 | 2019-05-27 | A kind of information recommendation method, device, equipment and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN110321478A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116304351A (en) * | 2023-03-29 | 2023-06-23 | 黑龙江青策网络科技有限公司 | Intelligent data information statistical management system and method based on big data |
Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833552A (en) * | 2009-03-10 | 2010-09-15 | 郝瑞林 | Method for marking and recommending streaming media |
CN104008138A (en) * | 2014-05-08 | 2014-08-27 | 南京邮电大学 | Music recommendation method based on social network |
CN106022865A (en) * | 2016-05-10 | 2016-10-12 | 江苏大学 | Goods recommendation method based on scores and user behaviors |
CN109033241A (en) * | 2018-07-04 | 2018-12-18 | 北京第视频科学技术研究院有限公司 | News recommended method, device and electronic equipment |
CN109145210A (en) * | 2018-08-17 | 2019-01-04 | 腾讯科技(深圳)有限公司 | A kind of method, apparatus and storage medium of information recommendation |
CN109472286A (en) * | 2018-09-30 | 2019-03-15 | 浙江工业大学 | Books in University Library recommended method based on interest-degree model Yu the type factor |
CN109543111A (en) * | 2018-11-28 | 2019-03-29 | 广州虎牙信息科技有限公司 | Recommendation information screening technique, device, storage medium and server |
-
2019
- 2019-05-27 CN CN201910443788.6A patent/CN110321478A/en active Pending
Patent Citations (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101833552A (en) * | 2009-03-10 | 2010-09-15 | 郝瑞林 | Method for marking and recommending streaming media |
CN104008138A (en) * | 2014-05-08 | 2014-08-27 | 南京邮电大学 | Music recommendation method based on social network |
CN106022865A (en) * | 2016-05-10 | 2016-10-12 | 江苏大学 | Goods recommendation method based on scores and user behaviors |
CN109033241A (en) * | 2018-07-04 | 2018-12-18 | 北京第视频科学技术研究院有限公司 | News recommended method, device and electronic equipment |
CN109145210A (en) * | 2018-08-17 | 2019-01-04 | 腾讯科技(深圳)有限公司 | A kind of method, apparatus and storage medium of information recommendation |
CN109472286A (en) * | 2018-09-30 | 2019-03-15 | 浙江工业大学 | Books in University Library recommended method based on interest-degree model Yu the type factor |
CN109543111A (en) * | 2018-11-28 | 2019-03-29 | 广州虎牙信息科技有限公司 | Recommendation information screening technique, device, storage medium and server |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116304351A (en) * | 2023-03-29 | 2023-06-23 | 黑龙江青策网络科技有限公司 | Intelligent data information statistical management system and method based on big data |
CN116304351B (en) * | 2023-03-29 | 2024-02-02 | 陕西维纳数字科技股份有限公司 | Intelligent data information statistical management system and method based on big data |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN110442790B (en) | Method, device, server and storage medium for recommending multimedia data | |
CN110543598B (en) | Information recommendation method and device and terminal | |
Liu et al. | Exploring personal impact for group recommendation | |
CN104969224B (en) | Improved user experience for unacknowledged and new users | |
CN108108821A (en) | Model training method and device | |
Sun et al. | Contextual intent tracking for personal assistants | |
Agarwal et al. | Statistical methods for recommender systems | |
CN105976161A (en) | Time axis-based intelligent recommendation calendar and user-based presentation method | |
CN111680254B (en) | Content recommendation method and device | |
CN106326391A (en) | Method and device for recommending multimedia resources | |
CN103886090A (en) | Content recommendation method and device based on user favorites | |
CN110020094A (en) | A kind of methods of exhibiting and relevant apparatus of search result | |
CN106131601A (en) | Video recommendation method and device | |
CN109118336A (en) | Information recommendation method, device, computer equipment and storage medium | |
WO2009023067A1 (en) | System and method for invitation targeting in a web-based social network | |
CN110163703B (en) | Classification model establishing method, file pushing method and server | |
CN102855269A (en) | Content extracting device, content extracting method and program | |
Nathanson et al. | Eigentaste 5.0: constant-time adaptability in a recommender system using item clustering | |
CN103838834A (en) | Mentioning recommending method, information processing method and system | |
CN108595493A (en) | Method for pushing and device, storage medium, the electronic device of media content | |
CN108228844A (en) | A kind of picture screening technique and device, storage medium, computer equipment | |
US20190278819A1 (en) | Systems and methods of providing recommendations of content items | |
CN110472016B (en) | Article recommendation method and device, electronic equipment and storage medium | |
CN112100221B (en) | Information recommendation method and device, recommendation server and storage medium | |
WO2017210644A1 (en) | System and method for a platform to identify and connect like-minded individuals based on interaction |
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 |