CN110674412A - Resource recommendation information pushing method and device and electronic equipment - Google Patents
Resource recommendation information pushing method and device and electronic equipment Download PDFInfo
- Publication number
- CN110674412A CN110674412A CN201910766710.8A CN201910766710A CN110674412A CN 110674412 A CN110674412 A CN 110674412A CN 201910766710 A CN201910766710 A CN 201910766710A CN 110674412 A CN110674412 A CN 110674412A
- Authority
- CN
- China
- Prior art keywords
- day
- attention
- increment
- resource
- resources
- 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
Images
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
- G06F16/9536—Search customisation based on social or collaborative filtering
-
- 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
- G06F16/958—Organisation or management of web site content, e.g. publishing, maintaining pages or automatic linking
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)
Abstract
The invention relates to a resource recommendation information pushing method and device and electronic equipment. The resource recommendation information pushing method comprises the following steps: acquiring the attention amount of the resources in a statistical period every day; acquiring the attention increment of each day of the resource according to the attention amount of each day of the resource and the attention amount of the day before the day of the resource; and pushing recommendation information of the resources with larger attention increment according to the attention increment of the resources in the statistical period every day. According to the method for pushing the recommendation information of the resources, the recommendation information of the resources with larger attention increment is pushed to the user according to the difference value, namely the attention increment, between the attention amount of the resources in each day and the attention amount of the resources in the day before the day in the statistical period, the hot resources can be more accurately recommended to the user, and meanwhile, the continuous hot resources can be recommended, and meanwhile, more recently hot resources can be recommended.
Description
Technical Field
The invention relates to the technical field of information processing, in particular to a resource recommendation information pushing method and device and electronic equipment.
Background
The popular recommendation method is a typical non-personalized recommendation, and the common implementation modes of the popular recommendation method include the following four types: the number of interactions per unit time, the total number of interactions, the number of comments, and the time sequence. In the industry, most of production judges the popularity of resources by using total interaction numbers, namely counting the total interaction numbers (praise numbers, comment numbers, click volumes and the like) of the resources within a period of time, and sorts the first few most popular resources according to the interaction numbers of the resources to recommend the anonymous or long-time unregistered users to stimulate flow or register, wherein the popularity ranking method based on the total interaction numbers comprises some defects, for example, if the popularity of the resources starts to suddenly and continuously rise due to some factors (the popularity of the resources increases, the stimulation of external factors and the like) at the recent time point, the statistical ranking mode based on the total interaction numbers will definitely weaken the existence of the resources in recommendation results, and the longer the time for incorporating the popularity statistics, the more unfavorable the resources; finally, repeated and focused recommendations due to the Martha effect can cause a resource dip for hot recommendations.
Disclosure of Invention
Accordingly, an object of the present invention is to provide a method and an apparatus for pushing recommendation information of resources, and an electronic device, which can recommend not only continuously popular resources but also more recently popular resources.
According to a first aspect of the embodiments of the present disclosure, an embodiment of the present disclosure provides a method for pushing recommendation information of a resource, where the method includes:
acquiring the attention amount of the resources in a statistical period every day;
acquiring the attention increment of each day of the resource according to the attention amount of each day of the resource and the attention amount of the day before the day of the resource;
and pushing recommendation information of the resources with larger attention increment according to the attention increment of the resources in the statistical period every day.
According to the resource recommendation information pushing method, the recommendation information of the resource with the larger attention increment is pushed to the user according to the difference value between the attention amount of the resource in each day and the attention amount of the resource in the day before the day in the statistical period, namely the attention increment, so that the hot resource can be recommended to the user more accurately, and meanwhile, the continuous hot resource can be recommended and more recently hot resources can be recommended.
Further, according to the magnitude of the attention increment of the resource every day in the statistical period, pushing recommendation information of the resource with a large attention increment, comprising the following steps:
summing the attention increments of the resources in the statistical period every day, acquiring a total attention increment, and pushing recommendation information of the resources with larger total attention increments;
alternatively, the first and second electrodes may be,
summing the attention increment of the resource every day in the statistical period, acquiring the average attention increment of the resource every day in the period of putting the resource on the shelf according to the number of days of putting the resource on the shelf in the statistical period, and pushing the recommendation information of the resource with larger average attention increment.
Further, summing the increments of interest for each day of the resource over a statistical period, comprising:
and weighting and summing the concerned increment of each day of the resource in the statistical period.
Further, the weighted summation of the attention increment of each day of the resource in the statistical period comprises the following steps:
and multiplying the attention increment of the resource in the statistical period by a preset attenuation function and then summing, wherein the closer the attention increment is to the starting time of the statistical period, the smaller the attenuation function value is, so that the weight of days with longer time is reduced, and the current heat of the resource is reflected more accurately.
Further, the weighted summation of the attention increment of each day of the resource in the statistical period comprises the following steps:
carrying out weighted calculation on the concerned increment from the first day to the day in the statistical period to obtain the weighted increment of the day, wherein the calculation formula is as follows:
vd＝vd-1+θd
wherein s isdIs the weighted increment of the day, vdFor the middle of the day, vd-1For intermediate calculations, theta, of the day preceding the daydThe day's increment of interest;
and multiplying the weighted increment of the resource in the statistical period by a preset attenuation function, and then summing, wherein the closer the weighted increment is to the starting time of the statistical period, the smaller the attenuation function value is.
Further, the weighted summation of the attention increment of each day of the resource in the statistical period comprises the following steps:
calculating an exponentially weighted average of the concerned increment, wherein the calculation formula of the exponentially weighted average is as follows:
wd＝β·wd-1+(1-β)θd
wherein, wdIs an exponentially weighted average of the increments of interest over the last day of the statistical period, wd-1Is an exponentially weighted average of the increments of interest for the day before the cutoff statistical period, beta is an adjustable weight parameter, thetadThe attention increment of the day is increased, so that the weight of days with longer time is reduced, and the current heat of the resource is reflected more accurately.
Further, obtaining the attention amount of the resource in each day in the statistical period comprises:
acquiring the average value and the standard deviation of the attention amount of different resources each day;
carrying out data standardization processing on the attention amount of different resources each day according to the average value and the standard deviation of the attention amount of different resources each day, wherein the calculation formula is as follows:
wherein, ω isdIs the value of interest, omega, for the day after normalization of the calculated datadFor the amount of attention on the day,
the data standardization processing is carried out on the attention amount of different resources every day, so that on one hand, the data can be in the same scale, the data processing amount is reduced, and on the other hand, the day-to-day difference can be eliminated.
Further, after obtaining the attention increment of the resource each day according to the attention amount of the resource each day and the attention amount of the resource on the day before the day, the method further includes:
acquiring the average value and the standard deviation of the concerned increment of different resources every day;
carrying out data standardization processing on the attention increment of different resources each day according to the average value and the standard deviation of the attention increment of different resources each day, wherein the calculation formula is as follows:
wherein, thetadIs the incremental value of interest, θ, for the day after normalization of the calculated datadFor the increment of interest on that day,
the method is characterized in that the average value of the attention increment of different resources on the day is sigma, the standard deviation of the attention increment of different resources on the day is provided, data can be enabled to be in the same scale on one hand by carrying out data standardization processing on the attention increment of different resources on each day, the data processing amount is reduced, and the difference between days can be eliminated on the other hand.
Further, the amount of attention includes any one of:
click volume, number of comments, number of praise.
According to a second aspect of the embodiments of the present disclosure, there is also provided a device for pushing recommendation information of a resource, the device including:
the attention amount obtaining module is used for obtaining the attention amount of the resources in the statistical period every day;
the attention increment obtaining module is used for obtaining the attention increment of the resource each day according to the attention amount of the resource each day and the attention amount of the resource on the day before the day;
and the pushing module is used for pushing the recommendation information of the resource with larger attention increment according to the attention increment of the resource every day in the statistical period.
According to a third aspect of the embodiments of the present disclosure, there is also provided an electronic apparatus, including:
a memory and a processor;
the memory for storing one or more programs;
when the one or more programs are executed by the one or more processors, the one or more processors implement the recommendation information pushing method for resources according to the first aspect of the embodiments of the present disclosure.
For a better understanding and practice, the invention is described in detail below with reference to the accompanying drawings.
Drawings
FIG. 1 is a flowchart illustrating a recommendation pushing method for resources of the present invention in an exemplary embodiment;
FIG. 2 is a flow chart illustrating a method for pushing recommendation information for resources of the present invention focusing on incremental weighted summation in an exemplary embodiment;
FIG. 3 is a flow chart illustrating a data normalization process for attention amounts of different resources per day in a recommendation information pushing method for resources according to the present invention in an exemplary embodiment;
FIG. 4 is a flowchart illustrating data normalization of attention increments for different resources per day in a recommendation information pushing method for resources of the present invention in an exemplary embodiment;
FIG. 5 is a flowchart illustrating a recommendation pushing method for resources of the present invention in an exemplary embodiment;
FIG. 6 is a graphical illustration of a recommendation information push method decay function red (d) of the resource of the present invention shown in an exemplary embodiment;
FIG. 7a is a graphical illustration of the number of clicks that the recommendation information push method for resources of the present invention shown in an exemplary embodiment recently started to hit the resource 30 days old;
FIG. 7b is a plot of the Δ cnt area after the recommendation information push method for resources of the present invention shown in an exemplary embodiment has recently begun to warm the resources for 30 days of normalization;
FIG. 7c is a schematic diagram illustrating a change in click rate of popular resources that have been off-shelf in the recommendation information pushing method for resources of the present invention for 30 days in an exemplary embodiment;
FIG. 7d is a diagram illustrating the area of Δ cnt after the normalized hit rate for 30 days of popular resources that have been off-shelf in the recommendation information pushing method for resources of the present invention in an exemplary embodiment;
FIG. 7e is a schematic diagram illustrating a change in click volume of the recommendation information pushing method for resources of the present invention shown in an exemplary embodiment for 30 days continuously trending the resources;
FIG. 7f is a diagram illustrating the area of Δ cnt after normalization of the hit rate for 30 days of the resource in the recommendation pushing method for resources according to the present invention in an exemplary embodiment;
fig. 8 is a diagram illustrating a distribution of recommendation results of a recommendation information pushing method using TopN and resources according to an embodiment of the present invention in an exemplary embodiment;
FIG. 9 is a block diagram showing the structure of a recommendation information pushing device of resources of the present invention in an exemplary embodiment;
fig. 10 is a block diagram showing the structure of an electronic device of the present invention in an exemplary embodiment.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some but not all of the relevant aspects of the present invention are shown in the drawings.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. As used in this specification and the appended claims, the singular forms "a", "an", and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It is to be understood that although the terms first, second, third, etc. may be used herein to describe various information, these information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present invention. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.
The resource recommendation information pushing method provided by the embodiment of the invention can be applied to software programs of electronic commerce websites and intelligent terminals, and the resources can be resources, services, network course resources and the like. The working environment of the electronic equipment where the resource recommendation information pushing method of the embodiment of the invention is located is preferably a resource recommendation server, and the resource recommendation server can comprise a plurality of databases. The electronic device to which the recommendation information pushing method for resources is applied is not limited to a personal computer, a multiprocessor system, a consumer electronic device, a minicomputer, a mainframe computer, a distributed computing environment including any of the above systems or devices, and the like.
As shown in fig. 1, in an exemplary embodiment, the method for pushing recommendation information of resources in the present invention includes the following steps:
step S101: and acquiring the daily attention amount of the resources in the statistical period.
The statistical period is a period of time before the preset current time, and is usually an integer multiple of a day, and may be, for example, one month or one quarter. The attention amount represents the attention heat degree of the user to the resource, and the attention amount is larger, the attention heat degree of the user to the resource is higher. For different resources, the attention amount may have a plurality of different specific expressions, for example, the attention amount may be a number of praise, a number of comments, or a click amount, or may be a combination of a plurality of the number of praise, the number of comments, or the click amount, or may be a number of good comments, a good comment proportion, or the like, in some resources with a focus on evaluation.
Step S102: and acquiring the attention increment of the resource each day according to the attention amount of the resource each day and the attention amount of the resource on the day before the day.
Step S103: and pushing recommendation information of the resources with larger attention increment according to the attention increment of the resources in the statistical period every day.
The attention increment may be obtained by subtracting the attention amount of the day before the day from the attention amount of the day, and if the attention amount of the day is greater than the attention amount of the day before the day, that is, if the heat of the resource of the day is in an increasing trend, the attention increment is a positive value, otherwise, if the heat of the resource of the day is in a decreasing trend, the attention increment is a negative value.
According to the embodiment of the invention, the heat degree of the resource in a period of time before the current time is reflected according to the attention increment of the resource every day in the statistical period, and the recommendation information of the resource with larger heat degree, namely attention increment is larger is recommended to the user according to the heat degree of different resources in the statistical period.
The recommendation basis can be to perform different operations on the daily attention increment of the resource, obtain a total attention increment capable of reflecting the heat degree of the resource in the whole statistical period, and push recommendation information of the resource with a larger total attention increment to the user according to the size sequence of the total attention increment.
The pushing of the recommendation information of the resource with the larger attention increment may be a way of pushing an advertisement to the user or showing a recommended resource list to the user in a webpage or an application program when the user browses the webpage or uses the application program, or a way of sending a recommendation short message to the user, and the like, and the recommendation information may be an icon with a hyperlink of the resource, or detailed information of the resource, and the like.
According to the resource recommendation information pushing method, the recommendation information of the resource with the larger attention increment is pushed to the user according to the difference value between the attention amount of the resource in each day and the attention amount of the resource in the day before the day in the statistical period, namely the attention increment, so that the hot resource can be recommended to the user more accurately, and meanwhile, the continuous hot resource can be recommended and more recently hot resources can be recommended.
In an exemplary embodiment, in step S103, according to the size of the attention increment of the resource in the statistics period per day, a specific implementation manner of pushing the recommendation information of the resource with a larger attention increment may be to sum the attention increments of the resource in the statistics period per day, obtain a total attention increment, and push the recommendation information of the resource with a larger total attention increment to the user.
In an exemplary embodiment, after the total attention increment is acquired, an average attention increment of the resource during the period of putting on shelf every day is acquired according to the days of putting on shelf of the resource in the statistical period, and recommendation information of the resource with a large average attention increment is pushed to a user. Wherein the average focus increment may be obtained by dividing the total focus increment by the number of days in the rack.
The method for characterizing the hot degree of a resource in a statistical period generally comprises the following forms: the door is continuously heated, the door heating degree is reduced, and the door heating is started recently. Based on the embodiment that the continuous hot resource is embodied, in order to better embody the resource which starts to be hot recently and reduce the proportion of the resource with the reduced hot degree, in an exemplary embodiment, when the attention increment is summed, the weight of the attention increment of the resource in each day in the statistical period may be adjusted according to the distance from the starting time of the statistical period, that is, the summing manner may be that the attention increment of the resource in each day in the statistical period is summed, specifically, the attention increment of the resource in each day in the statistical period is weighted and summed.
The weighted summation may be performed by, when summing the attention increments of each day, assigning different weight values to the attention increments of each day, so as to reduce the proportion of the attention increments of the day corresponding to the low weight values in the total attention increments, and increase the proportion of the attention increments of the day corresponding to the high weight values in the total attention increments. The weight value may be determined in such a manner that the closer the weight value is to the start time of the statistical period, the smaller the weight value is, thereby reducing the proportion of the resource hot degree at a longer time.
In an example, the weight value may be determined according to a preset attenuation function, that is, the weighted summation may be performed by multiplying a focus increment of the resource each day in the statistical period by the preset attenuation function, and the closer the focus increment is to the start time of the statistical period, the smaller the focus function value is, where the focus increment of the resource each day corresponds to the focus increment of the resource each day.
In an exemplary embodiment, the weighted summation of the interest increments for each day of the resource in the statistical period may further be performed by:
step S201: carrying out weighted calculation on the concerned increment from the first day to the day in the statistical period to obtain the weighted increment of the day, wherein the calculation formula is as follows:
vd＝vd-1+θd(2)
wherein s isdIs the weighted increment of the day, vdFor the middle of the day, vd-1For intermediate calculations, theta, of the day preceding the daydThe day's increment of interest;
step S202: and multiplying the weighted increment of the resource in the statistical period by a preset attenuation function, and then summing, wherein the closer the weighted increment is to the starting time of the statistical period, the smaller the attenuation function value is.
In the above formula, the weighting increment s per daydUltimately determined by the increment of interest over a statistical period for all days prior to the day, i.e.
Wherein, theta0Is zero.
After multiplying the above formula by a preset attenuation function, the formula is as follows:
wherein f (d) is an attenuation function. As can be seen from the above formula, unlike the previous embodiment in which the weight value of the increment of interest per day is smaller as the time of the start of the statistical period is closer, the weight value of the increment of interest per day in formula (1) and formula (2) of the present embodiment is larger as the time of the start of the statistical period is closer, and the attenuation function f (d) of the present embodiment is the same as the attenuation function of the previous embodiment in which the value of the attenuation function is larger as the time of the start of the statistical period is closer. In the final weighted summation formula of this embodiment, the weight value of the concerned increment is smaller as the closer or farther from the start time of the statistical period, and the weight value of the middle time period of the statistical period is larger. Therefore, the embodiment focuses more on the attention heat of the user to the resource in the middle period of the statistical period, can better reflect the resource which is hot from the middle period of the statistical period, and simultaneously reduces the proportion of the resource of which the hot degree is reduced.
The order of the steps of calculating the weighted increment and multiplying the weighted increment by the preset attenuation function is not limited, and in other examples, the weighted increment may be calculated by multiplying the preset attenuation function first and then calculating the weighted increment according to the concerned increment multiplied by the preset attenuation function.
In other examples, the magnitude of the weighted value of the attention increment in this embodiment is smaller as the attention increment is closer to or farther from the start time of the statistical period, and the weighted value in the middle time period of the statistical period is larger, and may also be implemented by multiplying the attention increment per day in the statistical period by a preset attenuation function, where the attenuation function value is smaller as the attention increment is closer to or farther from the start time of the statistical period.
In an exemplary embodiment, the closer to the starting time of the counting period, the smaller the weight value is, so as to reduce the weight of the resource trending degree of the longer time, and the weighted summation of the attention increment of the resource per day in the counting period may also be implemented by:
calculating an exponentially weighted average of the concerned increment, wherein the calculation formula of the exponentially weighted average is as follows:
wd＝β·wd-1+(1-β)θd
wherein, wdIs an exponentially weighted average of the increments of interest over the last day of the statistical period, wd-1Is an exponentially weighted average of the increments of interest for the day before the cutoff statistical period, beta is an adjustable weight parameter, thetadIncrement of interest for the day, wherein θ0Is zero.
After converting the above formula into the attention increment according to the day, the formula is:
wn＝(1-β)θn+β·(1-β)θn-1+β2·(1-β)θn-1+βn-1·(1-β)θ1
wherein n is the total number of days of the statistical period, wnIs an exponentially weighted average of the daily increments of interest over a statistical period, as can be seen from the above equation, the exponentially weighted average wnAnd the sum is obtained by multiplying the concerned increment of each day by a specified weight, wherein the closer the concerned increment is to the starting time of the statistical period, the smaller the weight is.
In some scenarios, the same resource provider may provide multiple types of different resources at the same time, and different resources have large exposure rates for different customer requirements or facing different types of customers, that is, the difference of the attention amounts may be large, and if statistics is performed only according to the attention amount, the resources facing the small consumer group may not be pushed. In order to more fairly push resources with different levels of attention to users and perform data standardization processing on the attention of different resources each day, the step S101 of acquiring the attention of the resources each day in a statistical period specifically includes the following steps:
step S301: the mean and standard deviation of the amount of interest for different resources per day are obtained.
Step S302: carrying out data standardization processing on the attention amount of different resources each day according to the average value and the standard deviation of the attention amount of different resources each day, wherein the calculation formula is as follows:
wherein, ω isdIs the value of interest, omega, for the day after normalization of the calculated datadFor the amount of attention on the day,
is the average value of the attention amount of different resources on the day, and delta is the standard deviation of the attention amount of different resources on the day.
In another example, the data normalization processing may be performed on the attention increment of different resources each day, and after acquiring the attention increment of a resource each day according to the attention amount of the resource each day and the attention amount of the resource on the day before the day, the method further includes:
step S401: the mean and standard deviation of the increments of interest for different resources per day are obtained.
Step S402: carrying out data standardization processing on the attention increment of different resources each day according to the average value and the standard deviation of the attention increment of different resources each day, wherein the calculation formula is as follows:
wherein, thetadIs the incremental value of interest, θ, for the day after normalization of the calculated datadFor the increment of interest on that day,
is the average value of the concerned increment of different resources in the day, and sigma is the standard deviation of the concerned increment of different resources in the day.
The formulas of the standardization processing are z-score standardization formulas, and through the data standardization processing in the two examples, the attention amount or the attention increment of different resources per day is determined to be a unified data standard, so that the attention amount deviation caused by facing different customer requirements or facing different types of customers is overcome, meanwhile, the attention amount is reduced to a smaller value, and the data calculation amount is reduced. In other examples, the data may not be normalized, and the ranking results of the hot resources are recommended according to the arrangement of the total attention increment or the average attention increment from large to small for the resources with large resource amount, rich types and uniform heat of each type (the exposure rates of the resources of different types are not different), and the hot recommendations may be made by grouping the first few resources in the types.
As shown in fig. 5, in a specific embodiment, a method for pushing recommendation information of a resource according to an embodiment of the present invention includes the following steps:
step S501: and acquiring the attention amount cnt of the resource every day in the statistical period.
Step S502: and obtaining the attention increment delta cnt of the resource each day according to the attention amount of the resource each day and the attention amount of the resource on the day before the day.
Step S503: the delta of interest Δ cnt for different resources per day is normalized (normalization) to yield Δ zcnt, normalized by z-score.
Step S504: the normalized attention increment Δ zcnt is multiplied by a preset attenuation function red (d), wherein the value of the attenuation function red (d) is smaller as it is closer to the start time of the statistical period.
First, the normalized attention increment Δ zcnt is reduced by a smaller value as it is closer to the start time of the statistical period using the mapping function index (d), for example, the statistical period is 30 days, and then the mapping function index (d) on the first day (d ═ 1) may be set to a valueThe value of the mapping function index (d) for the next day (d ═ 2) may be
The mapping function index (d) on the last day (d 30, i.e. the day farthest from the start time of the counting period) has a value of 1. Selecting the decay function red (d) as ex-3Wherein the image of the decay function red (d) is shown in fig. 6. The range of results obtained because of the index (d) function is
In order to substitute the value obtained by the index (d) function into the attenuation function, the value obtained by the function is directly enlarged by 3 times and the range is changed
Will attenuate the function ex-3And index (d) function, the formula for obtaining the attenuation function red (d) is as follows:
red(Δzcnt)＝Δzcnt·e(3·index(d)-3)
as can be seen from fig. 6, the impact of the attention increment retention for days closer to the present will be greater.
Step S505: carrying out weighted calculation on the concerned increment from the first day to the day in the statistical period to obtain the weighted increment of the day, wherein the calculation formula is as follows:
vd＝vd-1+θd
wherein s isdIs the weighted increment of the day, vdFor the middle of the day, vd-1For intermediate calculations, theta, of the day preceding the daydIncrease of interest on that day, θdEqual to red (d) in the previous step. In some examples, if a resource is not on shelf for that day, the weighted increment s for the resource for that day may be setdIs set to 0, and vdThe value of the day before that day is maintained, thereby reducing the impact of the resource not being on shelf.
Step S506: for the weighted increment s of each daydAnd summing to obtain the total weighted increment total _ s in the statistical period. The calculation formula is as follows:
wherein n is the number of days of the statistical cycle.
Step S507: and acquiring an average attention increment aver _ s of the resource every day during the period of uploading according to the online _ d of the resource on the days of uploading in the statistical period, sequencing the average attention increment aver _ s, and pushing recommendation information of the resource with a larger average attention increment aver _ s. The calculation formula is as follows:
the method comprises the steps of outputting ranking results of hot resources according to the arrangement of average attention increment aver _ s from large to small aiming at the resources with large resource quantity, rich types and uniform heat of various types (the exposure rates of different types of resources are not different), recommending, and grouping and then taking the first few types of resources in the types to be hot recommended aiming at the resources with small resource quantity, small types and concentrated heat in a certain type.
In the aspect of time complexity, a statistical period D is set, the total number of resources is N, the time complexity of the algorithm is O (DN), but the complexity of the algorithm can be reduced by maintaining a historical calculation table, the table can record the standardized results of the resources every day before D-1 days, so that when the recommendation result is calculated for D days, the standardized results of all the resources of the day and the new average attention increment aver _ s of each resource are calculated only once, the time complexity is reduced to O (N), and the calculation speed is greatly improved.
As shown in fig. 7a and 7b, fig. 7a and 7b are graphs of area of attention increment Δ zcnt after the change of attention amount and the normalization of the recent 30 days of the popular resource, respectively, in an example, the resource is on line on day 9 of month 4, the area of attention amount and the area of attention increment Δ zcnt at the beginning of the resource are not large, and the attention amount of the resource is increased from day 17 of month 4. If a traditional ranking algorithm based on the sum of the attention cnt of 30 days is used, compared with the continuously popular resources (i.e. the resources are always in a higher attention amount from the beginning of statistics), the resources do not have the advantage of the accumulation of the attention amount in the previous period and cannot appear in the recommended results.
As shown in fig. 7c and 7d, fig. 7c and 7d are graphs of area of interest increment Δ zcnt after change and normalization of interest amount of popular resources that were on-shelf for 30 days in an example, the resources were on-shelf for 4 months and 10 days, and the resources were on-shelf for a period of time from 4 months and 13 days to 4 months and 16 days, respectively, if a conventional ranking algorithm based on the sum of interest amounts cnt for 30 days is used, the probability of such resources is that competition recommendation fails because the interest amount of a certain period of time is lost.
As shown in fig. 7e and 7f, fig. 7e and 7f are area diagrams of attention increment Δ zcnt after change and normalization of attention amount of the continuously hot resource for 30 days in an example, respectively, and it can be seen that the daily attention amount cnt and attention increment Δ zcnt of the resource fluctuate steadily, which indicates that the algorithm still maintains the screening recommendation for the continuously hot resource.
As shown in fig. 8, fig. 8 shows the distribution of recommendation results of the recommendation information pushing method using TopN (traditional ranking algorithm based on the total of the attention cnt for 30 days) and resources according to the embodiment of the present invention, and it can be seen that the new algorithm effectively inhibits the existence of continuously trending resources in the recommendation results, and gives more exposure opportunities to recently trended and resources that have been on-shelf. Wherein, continuously warming up: the resource is always in a hot state from the beginning of statistics, namely the attention amount is always large and the attention amount generally tends to rise; 2) recently, the heat: the resource is always in a low-heat state before, and the resource becomes popular in the last period of time, namely, the attention amount continuously increases from the last period of time; 3) once off shelf: resources are placed on shelves continuously or intermittently several times after the shelf, i.e. the amount of interest is zero for a period of time within the statistical period. In the three groups of histograms in fig. 8, the histogram on the left side of each group is the recommendation result using the conventional sorting algorithm based on the cnt sum over 30 days, and the histogram on the right side of each group is the recommendation result using the recommendation information pushing method for resources according to the embodiment of the present invention.
Corresponding to the aforementioned method for pushing recommendation information of resources, the present invention further provides a device for pushing recommendation information of resources, where the device may be installed in any intelligent terminal, and for example, may be specifically a computer, a server, a mobile phone, a tablet computer, an interactive smart tablet, a PDA (Personal Digital Assistant), an e-book reader, a multimedia player, and the like. According to the resource recommendation information pushing device provided by the embodiment of the invention, the recommendation information of the resource with larger attention increment is pushed to the user according to the difference value, namely the attention increment, between the attention amount of the resource every day and the attention amount of the resource in the day before the day in the statistical period, so that the hot resource can be more accurately recommended to the user, and meanwhile, the continuously hot resource can be recommended, and more recently hot resources can be recommended.
In an exemplary embodiment, as shown in fig. 9, the recommendation information pushing apparatus 900 for resources includes:
an attention amount obtaining module 901, configured to obtain an attention amount of a resource each day in a statistical period;
an attention increment obtaining module 902, configured to obtain an attention increment of a resource each day according to an attention amount of the resource each day and an attention amount of a day before the resource each day;
and the pushing module 903 is configured to push recommendation information of a resource with a larger attention increment according to the attention increment of the resource every day in the statistical period.
In an exemplary embodiment, the push module 903 comprises:
the first pushing unit 9031 is configured to sum the attention increments of the resources in the statistical period every day, obtain a total attention increment, and push recommendation information of the resources with the larger total attention increment.
Alternatively, the first and second electrodes may be,
the second pushing unit 9032 is configured to sum the attention increments of the resource each day in the statistical period, acquire the average attention increment of the resource each day in the period of the resource being on shelf according to the number of days of the resource being on shelf in the statistical period, and push recommendation information of the resource with a larger average attention increment.
In an exemplary embodiment, the first push unit 9031 and/or the second push unit 9032 further includes:
and the weighted summation unit 9033 is used for weighted summation of the concerned increment of each day of the resource in the statistical period.
In an exemplary embodiment, the weighted sum unit 9033 further includes:
and the first attenuation unit 9034 is configured to multiply the attention increment of the resource in the statistical period by a preset attenuation function, and then sum the result, wherein the closer the attention increment is to the start time of the statistical period, the smaller the attenuation function value is.
In an exemplary embodiment, the weighted sum unit 9033 further includes:
a weighted increment obtaining unit 9035, configured to perform weighted calculation on the attention increment from the first day to the day in the statistical period to obtain a weighted increment of the day, where the calculation formula is:
vd＝vd-1+θd
wherein s isdIs the weighted increment of the day, vdFor the middle of the day, vd-1For intermediate calculations, theta, of the day preceding the daydThe day's increment of interest;
and the second attenuation unit 9035 is configured to multiply the weighted increment of the resource in the statistical period by a preset attenuation function, and then sum the weighted increment, where the closer the weighted increment is to the start time of the statistical period, the smaller the attenuation function value is.
In an exemplary embodiment, the weighted sum unit 9033 further includes:
an exponentially weighted average obtaining unit 9036, configured to calculate an exponentially weighted average of the attention increment, where a calculation formula of the exponentially weighted average is:
wd＝β·wd-1+(1-β)θd
wherein, wdIs an exponentially weighted average of the increments of interest over the last day of the statistical period, wd-1Is an exponentially weighted average of the increments of interest for the day before the cutoff statistical period, beta is an adjustable weight parameter, thetadThe day's attention was increased.
In an exemplary embodiment, the attention amount obtaining module 901 further includes:
a first obtaining unit 9011, configured to obtain an average value and a standard deviation of the amount of interest of different resources per day.
The first normalization unit 9012 is configured to perform data normalization processing on the attention amounts of different resources per day according to an average value and a standard deviation of the attention amounts of the different resources per day, where a calculation formula is as follows:
wherein, ω isdIs the value of interest, omega, for the day after normalization of the calculated datadFor the amount of attention on the day,
is the average value of the attention amount of different resources on the day, and delta is the standard deviation of the attention amount of different resources on the day.
In an exemplary embodiment, the focus delta acquisition module 902 further comprises:
the second obtaining unit 9021 obtains an average value and a standard deviation of the attention increment of different resources each day.
A second normalization unit 9022, configured to perform data normalization processing on the attention increments of different resources per day according to an average value and a standard deviation of the attention increments of different resources per day, where a calculation formula is as follows:
wherein, thetadIs the incremental value of interest, θ, for the day after normalization of the calculated datadFor the increment of interest on that day,
is the average value of the concerned increment of different resources in the day, and sigma is the standard deviation of the concerned increment of different resources in the day.
In an exemplary embodiment, the amount of interest includes any one of:
click volume, number of comments, number of praise.
Corresponding to the aforementioned method for pushing the recommendation information of the resource, the present invention further provides an electronic device applied to the device for pushing the recommendation information of the resource, where the electronic device may be any intelligent terminal, and for example, may be a computer, a server, a mobile phone, a tablet computer, an interactive intelligent tablet, a PDA (Personal Digital Assistant), an e-book reader, a multimedia player, and the like. The electronic equipment pushes recommendation information of the resources with larger attention increment to the user according to the difference value, namely the attention increment, between the attention amount of the resources in each day and the attention amount of the previous day in the statistical period, so that the hot resources can be recommended to the user more accurately, and meanwhile, the latest hot resources can be recommended while the continuous hot resources are recommended.
The electronic device includes:
a processor;
a memory for storing a computer program executable by the processor;
when the processor executes the program, the method for pushing recommendation information of a resource described in any of the above embodiments is implemented.
As shown in fig. 10, fig. 10 is a block diagram illustrating a structure of an electronic device according to an exemplary embodiment of the invention.
Referring to fig. 10, electronic device 1200 includes a processing component 1222 that further includes one or more processors, and memory resources, represented by memory 1232, for storing instructions, such as application programs, that are executable by processing component 1222. The application programs stored in memory 1232 may include one or more modules that each correspond to a set of instructions. Further, the processing component 1222 is configured to execute the instructions to execute the recommendation information pushing method of the resource recited in any of the above embodiments.
The electronic device 1200 may also include a power component 1226 configured to perform power management of the apparatus 1200, a wired or wireless network interface 1250 configured to connect the apparatus 1200 to a network, and an input output (I/O) interface 1258. The device 1200 may operate based on an operating system stored in memory 1232, such as Android, IOS, Windows Server, Mac OS XTM, UnixTM, LinuxTM, FreeBSDTM, or the like.
The implementation process of the functions and actions of each component in the above device is specifically described in the implementation process of the corresponding step in the above method, and is not described herein again.
For the apparatus embodiment, since it basically corresponds to the method embodiment, reference may be made to the partial description of the method embodiment for relevant points. The above-described device embodiments are merely illustrative, wherein the components described as separate parts may or may not be physically separate, and the parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the disclosed solution. One of ordinary skill in the art can understand and implement it without inventive effort. The electronic device provided by the above can be used to execute the resource recommendation information pushing method provided by any of the above embodiments, and has corresponding functions and beneficial effects. The implementation process of the functions and actions of each component in the device is specifically described in the implementation process of the corresponding step in the resource recommendation information pushing method, and is not described herein again.
Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention following, in general, the principles of the invention and including such departures from the present disclosure as come within known or customary practice within the art to which the invention pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.
It will be understood that the invention is not limited to the precise arrangements described above and shown in the drawings and that various modifications and changes may be made without departing from the scope thereof. The scope of the invention is limited only by the appended claims.
The above-mentioned embodiments only express several embodiments of the present invention, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the inventive concept, which falls within the scope of the present invention.
Claims (11)
1. A method for pushing recommendation information of resources is characterized by comprising the following steps:
acquiring the attention amount of the resources in a statistical period every day;
acquiring the attention increment of each day of the resource according to the attention amount of each day of the resource and the attention amount of the day before the day of the resource;
and pushing recommendation information of the resources with larger attention increment according to the attention increment of the resources in the statistical period every day.
2. The method for pushing recommendation information of resources according to claim 1, wherein pushing recommendation information of resources with larger attention increment according to attention increment of the resources per day in the statistical period comprises:
summing the attention increments of the resources in the statistical period every day, acquiring a total attention increment, and pushing recommendation information of the resources with larger total attention increments;
alternatively, the first and second electrodes may be,
summing the attention increment of the resource every day in the statistical period, acquiring the average attention increment of the resource every day in the period of putting the resource on the shelf according to the number of days of putting the resource on the shelf in the statistical period, and pushing the recommendation information of the resource with larger average attention increment.
3. The method for pushing recommendation information of resource according to claim 2, wherein summing the attention increments of each day of the resource in the statistical period comprises:
and weighting and summing the concerned increment of each day of the resource in the statistical period.
4. The method for pushing recommendation information of resource according to claim 3, wherein the weighted summation of the attention increments of the resource each day in the statistical period comprises:
and multiplying the attention increment of the resource in the statistical period by a preset attenuation function, and summing, wherein the closer the attention increment is to the starting time of the statistical period, the smaller the attenuation function value is.
5. The method for pushing recommendation information of resource according to claim 3, wherein the weighted summation of the attention increments of the resource each day in the statistical period comprises:
carrying out weighted calculation on the concerned increment from the first day to the day in the statistical period to obtain the weighted increment of the day, wherein the calculation formula is as follows:
vd＝vd-1+θd
wherein s isdIs the weighted increment of the day, vdFor the middle of the day, vd-1For intermediate calculations, theta, of the day preceding the daydThe day's increment of interest;
and multiplying the weighted increment of the resource in the statistical period by a preset attenuation function, and then summing, wherein the closer the weighted increment is to the starting time of the statistical period, the smaller the attenuation function value is.
6. The method for pushing recommendation information of resource according to claim 3, wherein the weighted summation of the attention increments of the resource each day in the statistical period comprises:
calculating an exponentially weighted average of the concerned increment, wherein the calculation formula of the exponentially weighted average is as follows:
wd＝β·wd-1+(1-β)θd
wherein, wdIs an exponentially weighted average of the increments of interest over the last day of the statistical period, wd-1Is an exponentially weighted average of the increments of interest of the day before the cutoff statistical period, and beta is an adjustable weightHeavy parameter, θdThe day's attention was increased.
7. The method for pushing recommendation information of resources according to claim 1, wherein obtaining the attention amount of the resources per day in the statistical period comprises:
acquiring the average value and the standard deviation of the attention amount of different resources each day;
carrying out data standardization processing on the attention amount of different resources each day according to the average value and the standard deviation of the attention amount of different resources each day, wherein the calculation formula is as follows:
wherein, ω isdIs the value of interest, omega, for the day after normalization of the calculated datadFor the amount of attention on the day,
is the average value of the attention amount of different resources on the day, and delta is the standard deviation of the attention amount of different resources on the day.
8. The method of claim 1, wherein after obtaining the attention increment of each day of the resource according to the attention amount of each day of the resource and the attention amount of the day before the day of the resource, the method further comprises:
acquiring the average value and the standard deviation of the concerned increment of different resources every day;
carrying out data standardization processing on the attention increment of different resources each day according to the average value and the standard deviation of the attention increment of different resources each day, wherein the calculation formula is as follows:
wherein, thetadIs the incremental value of interest, θ, for the day after normalization of the calculated datadFor the increment of interest on that day,
is the average value of the concerned increment of different resources in the day, and sigma is the standard deviation of the concerned increment of different resources in the day.
9. The method according to claim 1, wherein the amount of interest includes any one of:
click volume, number of comments, number of praise.
10. An apparatus for pushing recommendation information of a resource, the apparatus comprising:
the attention amount obtaining module is used for obtaining the attention amount of the resources in the statistical period every day;
the attention increment obtaining module is used for obtaining the attention increment of the resource each day according to the attention amount of the resource each day and the attention amount of the resource on the day before the day;
and the pushing module is used for pushing the recommendation information of the resource with larger attention increment according to the attention increment of the resource every day in the statistical period.
11. An electronic device, comprising:
a memory and a processor;
the memory for storing one or more programs;
when executed by the one or more processors, cause the one or more processors to implement the method for pushing recommendation information for resources of any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910766710.8A CN110674412A (en) | 2019-08-20 | 2019-08-20 | Resource recommendation information pushing method and device and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910766710.8A CN110674412A (en) | 2019-08-20 | 2019-08-20 | Resource recommendation information pushing method and device and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110674412A true CN110674412A (en) | 2020-01-10 |
Family
ID=69075413
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910766710.8A Pending CN110674412A (en) | 2019-08-20 | 2019-08-20 | Resource recommendation information pushing method and device and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110674412A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113824980A (en) * | 2021-09-09 | 2021-12-21 | 广州方硅信息技术有限公司 | Video recommendation method, system and device and computer equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500213A (en) * | 2013-09-30 | 2014-01-08 | 北京搜狗科技发展有限公司 | Page hot-spot resource updating method and device based on pre-reading |
| CN107026892A (en) * | 2016-11-03 | 2017-08-08 | 阿里巴巴集团控股有限公司 | Message recommends method and apparatus |
| CN107666516A (en) * | 2017-09-20 | 2018-02-06 | 重庆邮电大学 | A kind of method for ensureing kafka company-data uniformity based on message temperature |
| CN108363788A (en) * | 2018-02-12 | 2018-08-03 | 广州市贝聊信息科技有限公司 | Model intelligently ranks method, apparatus and computer readable storage medium |
| CN108710635A (en) * | 2018-04-08 | 2018-10-26 | 达而观信息科技（上海）有限公司 | A kind of content recommendation method and device |
| WO2019000782A1 (en) * | 2017-06-30 | 2019-01-03 | 深圳创维数字技术有限公司 | Search term recommendation method, set top box and storage medium |
| CN109344316A (en) * | 2018-08-14 | 2019-02-15 | 优视科技（中国）有限公司 | News temperature calculates method and device |
-
2019
- 2019-08-20 CN CN201910766710.8A patent/CN110674412A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103500213A (en) * | 2013-09-30 | 2014-01-08 | 北京搜狗科技发展有限公司 | Page hot-spot resource updating method and device based on pre-reading |
| CN107026892A (en) * | 2016-11-03 | 2017-08-08 | 阿里巴巴集团控股有限公司 | Message recommends method and apparatus |
| WO2019000782A1 (en) * | 2017-06-30 | 2019-01-03 | 深圳创维数字技术有限公司 | Search term recommendation method, set top box and storage medium |
| CN107666516A (en) * | 2017-09-20 | 2018-02-06 | 重庆邮电大学 | A kind of method for ensureing kafka company-data uniformity based on message temperature |
| CN108363788A (en) * | 2018-02-12 | 2018-08-03 | 广州市贝聊信息科技有限公司 | Model intelligently ranks method, apparatus and computer readable storage medium |
| CN108710635A (en) * | 2018-04-08 | 2018-10-26 | 达而观信息科技（上海）有限公司 | A kind of content recommendation method and device |
| CN109344316A (en) * | 2018-08-14 | 2019-02-15 | 优视科技（中国）有限公司 | News temperature calculates method and device |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113824980A (en) * | 2021-09-09 | 2021-12-21 | 广州方硅信息技术有限公司 | Video recommendation method, system and device and computer equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109274987B (en) | Video set ordering method, server and readable storage medium | |
| CN109275031B (en) | Video popularity evaluation method and device and electronic equipment | |
| CN110796477A (en) | Advertisement display method and device, electronic equipment and readable storage medium | |
| CN109359217B (en) | User interest degree calculation method, server and readable storage medium | |
| US10559004B2 (en) | Systems and methods for establishing and utilizing a hierarchical Bayesian framework for ad click through rate prediction | |
| CN108965951B (en) | Advertisement playing method and device | |
| CN111767429A (en) | Video recommendation method and device and electronic equipment | |
| US11277352B2 (en) | Method, apparatus, and server for dynamically adjusting resource competition thresholds | |
| CN110363580B (en) | Information recommendation method and device, computer equipment and storage medium | |
| CN113422986B (en) | Method, apparatus, device, medium, and program product for live room recommendation | |
| CN110825977A (en) | Data recommendation method and related equipment | |
| CN108667877B (en) | Method and device for determining recommendation information, computer equipment and storage medium | |
| CN112150182B (en) | Multimedia file pushing method and device, storage medium and electronic device | |
| JP7085486B2 (en) | User behavior data processing method and equipment | |
| CN110674412A (en) | Resource recommendation information pushing method and device and electronic equipment | |
| CN110570271A (en) | information recommendation method and device, electronic equipment and readable storage medium | |
| CN109684546A (en) | Recommended method, device, storage medium and terminal | |
| CN108769107A (en) | A kind of video dispatching method, device and electronic equipment | |
| CN109472455B (en) | Activity evaluation method, activity evaluation device, electronic equipment and storage medium | |
| CN110020129B (en) | Click rate correction method, prediction method, device, computing equipment and storage medium | |
| CN110717653A (en) | Risk identification method and device and electronic equipment | |
| CN113038242B (en) | Method, device and equipment for determining display position of live broadcast card and storage medium | |
| CN112669091B (en) | Data processing method, device and storage medium | |
| CN114201696A (en) | Message pushing method and device, storage medium and computer equipment | |
| CN113408817A (en) | Traffic distribution method, device, equipment 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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20200110 |