CN113821574A - User behavior classification method and device and storage medium - Google Patents
User behavior classification method and device and storage medium Download PDFInfo
- Publication number
- CN113821574A CN113821574A CN202111017409.0A CN202111017409A CN113821574A CN 113821574 A CN113821574 A CN 113821574A CN 202111017409 A CN202111017409 A CN 202111017409A CN 113821574 A CN113821574 A CN 113821574A
- Authority
- CN
- China
- Prior art keywords
- behavior
- operation behavior
- current user
- user operation
- target
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 57
- 230000006399 behavior Effects 0.000 claims abstract description 440
- 238000004590 computer program Methods 0.000 claims description 9
- 230000003542 behavioural effect Effects 0.000 claims description 5
- 238000005516 engineering process Methods 0.000 abstract description 4
- 238000004422 calculation algorithm Methods 0.000 description 13
- 230000009471 action Effects 0.000 description 11
- 238000010586 diagram Methods 0.000 description 11
- 238000012545 processing Methods 0.000 description 7
- 238000013468 resource allocation Methods 0.000 description 6
- 238000004364 calculation method Methods 0.000 description 5
- 238000004458 analytical method Methods 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000003993 interaction Effects 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 230000008569 process Effects 0.000 description 3
- 230000000694 effects Effects 0.000 description 2
- 238000005259 measurement Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 238000013500 data storage Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 230000002250 progressing effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000003595 spectral effect Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
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
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/28—Databases characterised by their database models, e.g. relational or object models
- G06F16/284—Relational databases
- G06F16/285—Clustering or classification
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3438—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment monitoring of user actions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2462—Approximate or statistical queries
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2474—Sequence data queries, e.g. querying versioned data
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Databases & Information Systems (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Probability & Statistics with Applications (AREA)
- Data Mining & Analysis (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Quality & Reliability (AREA)
- Computer Hardware Design (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The disclosure relates to a user behavior classification method and device and a storage medium. The method comprises the following steps: acquiring time sequence data corresponding to current user operation behaviors to be classified, wherein the time sequence data is used for recording a statistical result of the current user operation behaviors in a target time period, and the current user operation behaviors are operation behaviors executed on multimedia resources displayed in a target platform by a current user account; determining behavior similarity between the current user operation behavior and each core operation behavior according to the time sequence data, wherein the core operation behavior is the operation behavior matched with each cluster and determined from various user operation behaviors; and under the condition that the behavior similarity meets the matching condition of the target core operation behavior, classifying the current user operation behavior into a target cluster matched with the target core operation behavior. Therefore, the problem of low accuracy in the user behavior classification method in the related technology is solved.
Description
Technical Field
The present disclosure relates to the field of computers, and in particular, to a user behavior classification method and apparatus, and a storage medium.
Background
Nowadays, many multimedia resources are often released on many resource releasing platforms to achieve the purpose of advertising certain information carried in the multimedia resources. In order to improve the delivery effect of the multimedia resource, the related behavior of the user watching the multimedia resource is usually analyzed, so as to determine the preference of the user according to the classification result of the user behavior, thereby implementing targeted delivery of the resource.
Currently, a common way to classify user behavior is to perform unsupervised classification based on statistics. After the classification is performed by adopting the classification method, a larger deviation actually exists between the result obtained by the classification and the real preference of the user.
In view of the above problems, no effective solution has been proposed.
Disclosure of Invention
The present disclosure provides a user behavior classification method and apparatus, and a storage medium, to at least solve the problem of low accuracy in the user behavior classification method in the related art. The technical scheme of the disclosure is as follows:
according to a first aspect of the embodiments of the present disclosure, there is provided a user behavior classification method, including: acquiring time sequence data corresponding to current user operation behaviors to be classified, wherein the time sequence data is used for recording a statistical result of the current user operation behaviors in a target time period, and the current user operation behaviors are operation behaviors executed on multimedia resources displayed in a target platform by a current user account; determining behavior similarity between the current user operation behavior and each core operation behavior according to the time sequence data, wherein the core operation behavior is the operation behavior matched with each cluster determined from various user operation behaviors; and under the condition that the behavior similarity meets the matching condition of the target core operation behavior, classifying the current user operation behavior into a target cluster matched with the target core operation behavior.
According to a second aspect of the embodiments of the present disclosure, there is provided a user behavior classification apparatus, including: the device comprises a first obtaining unit, a second obtaining unit and a third obtaining unit, wherein the first obtaining unit is configured to obtain time sequence data corresponding to current user operation behaviors to be classified, the time sequence data is used for recording a statistical result of the current user operation behaviors in a target time period, and the current user operation behaviors are operation behaviors executed on multimedia resources displayed in a target platform by a user account; a first determining unit configured to determine behavior similarity between the current user operation behavior and each core operation behavior according to the time series data, wherein the core operation behavior is an operation behavior matched with each cluster determined from various user operation behaviors; and the classifying unit is configured to classify the current user operation behavior into a target cluster matched with the target core operation behavior under the condition that the behavior similarity meets the matching condition of the target core operation behavior.
According to a third aspect of the embodiments of the present disclosure, there is provided a computer-readable storage medium, wherein instructions, when executed by a processor of an electronic device, cause the electronic device to execute the user behavior classification method.
According to a fourth aspect of the embodiments of the present disclosure, there is provided an electronic apparatus including: a processor; a memory for storing the processor-executable instructions; the processor is configured to execute the instructions to implement the user behavior classification method.
According to a fifth aspect of embodiments of the present disclosure, there is provided a computer program product comprising computer programs/instructions which, when executed by a processor, implement the user behavior classification method described above.
The technical scheme provided by the embodiment of the disclosure at least brings the following beneficial effects:
in the embodiment of the invention, after the time sequence data corresponding to the user operation behaviors to be classified are acquired, the behavior similarity between the current user operation behavior and each core operation behavior is determined according to the time sequence data, wherein the core operation behavior is the operation behavior which is determined from various user operation behaviors and is matched with each cluster. And under the condition that the behavior similarity meets the matching condition of the target core operation behavior, classifying the current user operation behavior into a target cluster matched with the target core operation behavior. That is to say, the cluster to which the current user operation behavior belongs is determined by comparing the behavior similarity between the current user operation behavior to be classified and the core operation behavior determined for each cluster, so that more accurate classification processing is performed on the current user operation behavior according to the behavior similarity. And further the problem of lower accuracy in the user behavior classification method in the related technology is solved.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the disclosure and are not to be construed as limiting the disclosure.
FIG. 1 is a schematic diagram of an application environment illustrating a method for user behavior classification in accordance with an exemplary embodiment;
FIG. 2 is a flow diagram illustrating a method of user behavior classification in accordance with an exemplary embodiment;
FIG. 3 is a diagram illustrating an alternative user behavior classification method according to an exemplary embodiment;
FIG. 4 is a schematic diagram illustrating another alternative user behavior classification method in accordance with an exemplary embodiment;
fig. 5 is a block diagram illustrating an alternative user behavior classification apparatus according to an example embodiment.
Detailed Description
In order to make the technical solutions of the present disclosure better understood by those of ordinary skill in the art, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the accompanying drawings.
It should be noted that the terms "first," "second," and the like in the description and claims of the present disclosure and in the above-described drawings are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the disclosure described herein are capable of operation in sequences other than those illustrated or otherwise described herein. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
According to an aspect of the embodiments of the present invention, a user behavior classification method is provided, and optionally, as an optional implementation manner, the user behavior classification method may be applied, but not limited, to a user behavior classification system in a hardware environment as shown in fig. 1. The user behavior classification system may include, but is not limited to, the terminal device 102, the network 104, and the server 106. The terminal device 102 includes a human-computer interaction screen 1022, a processor 1024, and a memory 1026. The human-computer interaction screen 1022 is used to present multimedia assets. The processor 1024 is configured to generate a human-computer interaction instruction in response to a human-computer interaction operation, so as to determine an operation behavior performed on the multimedia resource by the user account. The memory 1026 is used for storing the multimedia resources to be presented.
In addition, the server 106 includes a database 1062 and a processing engine 1064, where the database 1062 is used to store the time series data acquired from the terminal device 102 and the behavior similarity between each core operation behavior and each user operation behavior obtained through calculation. The processing engine 1064 is configured to determine whether the behavior similarity satisfies a matching condition of the target core operation behavior, and classify the user operation behavior into a target cluster matched with the target core operation behavior when it is determined that the matching condition of the target core operation behavior is satisfied.
Assume that the terminal device 102 is a hardware device for displaying multimedia resources in any application scenario, where the specific process includes the following steps: in step S102, the terminal device 102 sends, to the server 106 through the network 104, time series data corresponding to the current user operation behavior to be classified, where the time series data are used to record a statistical result of the current user operation behavior in a target time period, and the current user operation behavior is an operation behavior performed by a user account on a multimedia resource (as shown in the figure, multimedia resource 1) presented by a target platform. Processing engine 1064 in server 106 then performs the following steps using the received time series data stored in database 1064:
in step S104, the behavior similarity between the current user operation behavior and each core operation behavior is determined according to the time series data, wherein the core operation behavior is the operation behavior matched with the classified cluster. In step S106, in the case that the behavior similarity satisfies the matching condition of the target core operation behavior, the user operation behavior is classified into a target cluster matched with the target core operation behavior. Therefore, different user operation behaviors are classified more accurately without using an unsupervised classification mode, and the purpose of improving the classification accuracy of the user behaviors is achieved.
Further, in step S108, the server 106 adjusts the delivered multimedia resource in the terminal device 102 according to the classified target cluster, so as to improve the conversion rate of the delivered multimedia resource.
It should be noted that, in this embodiment, after obtaining time series data corresponding to a current user operation behavior to be classified, behavior similarity between the current user operation behavior and each core operation behavior is determined according to the time series data, where the core operation behavior is an operation behavior that is determined from various user operation behaviors and is matched with each cluster. And under the condition that the behavior similarity meets the matching condition of the target core operation behavior, classifying the current user operation behavior into a target cluster matched with the target core operation behavior. That is to say, the cluster to which the current user operation behavior belongs is determined by comparing the behavior similarity between the current user operation behavior to be classified and the core operation behavior determined for each cluster, so that the user operation behavior is classified more accurately according to the behavior similarity. And further the problem of lower accuracy in the user behavior classification method in the related technology is solved.
Optionally, in this embodiment, the terminal device may be a terminal device configured with a target client, and may include, but is not limited to, at least one of the following: mobile phones (such as Android phones, iOS phones, etc.), notebook computers, tablet computers, palm computers, MID (Mobile Internet Devices), PAD, desktop computers, smart televisions, etc. The target client can be a video client, an instant messaging client, a browser client, an education client, and the like, which allow the delivery of multimedia resources. Such networks may include, but are not limited to: a wired network, a wireless network, wherein the wired network comprises: a local area network, a metropolitan area network, and a wide area network, the wireless network comprising: bluetooth, WIFI, and other networks that enable wireless communication. The server may be a single server, a server cluster composed of a plurality of servers, or a cloud server. The above is merely an example, and this is not limited in this embodiment.
Optionally, as an optional implementation manner, as shown in fig. 2, the user behavior classification method includes:
s202, acquiring time sequence data corresponding to current user operation behaviors to be classified, wherein the time sequence data is used for recording a statistical result of the current user operation behaviors in a target time period, and the current user operation behaviors are operation behaviors executed by a user account on multimedia resources displayed in a target platform;
s204, determining the behavior similarity between the current user operation behavior and each core operation behavior according to the time sequence data, wherein the core operation behavior is the operation behavior matched with each cluster determined from various user operation behaviors;
and S206, under the condition that the behavior similarity meets the matching condition of the target core operation behavior, classifying the current user operation behavior into a target cluster matched with the target core operation behavior.
Optionally, in this embodiment, the user behavior classification method may be applied, but not limited to, in a decision process of a control policy for an online information flow. For example, taking the multimedia resource allocation field as an example, the online control policy herein may include but is not limited to: a strategy for controlling multimedia resource over-allocation, a strategy for improving budget consumption rate, a cold start strategy (here, cold start refers to a release strategy referred to when user preference is unknown at the time of first release), and the like. That is to say, the cluster to which the user operation behavior belongs is determined through analysis of the user operation behaviors, so that the multimedia resource allocation control strategy correspondingly adopted by the current user operation behavior is determined. In addition, the target platform may be, but not limited to, different terminal Applications (APPs), such as a short video platform Application, a video playing Application, a community space sharing Application, an instant messaging Application, an online shopping Application, and the like. The above is an example, and the application scenarios and fields are not limited in any way.
It should be noted that, in this embodiment, the current user operation behavior to be classified refers to the user operation behavior to be currently classified, and the data pool actually stores time series data of a plurality of different user operation behaviors.
Optionally, in this embodiment, the operation performed by each user account on the multimedia resource presented on the platform may include, but is not limited to, at least one of the following: exposure, play, attention, comment, share, pay, etc. That is, each user action has an action tag and carries an action timestamp.
Optionally, in this embodiment, each cluster may be, but is not limited to, divided according to resource characteristics (e.g., service requirements) of the multimedia resource. Taking the multimedia resource allocation procedure as an example, the link cycle designed for the multimedia resource may include, but is not limited to: exposure (Aware), attraction (Appeal), action (Act), user (Advocate), Assets (Assets), five parts, called the 5A model. The main application scene of the 5A model is the measurement of the full life cycle value of brand consumers, and the applied product function modules comprise customer asset overview, 5A link analysis and the like; the latter is also applied to modules of link flow and the like. In this embodiment, the user operation behaviors may be classified into each cluster according to a comparison result of the behavior similarity, so as to achieve the purpose of accurate classification. Here, this is an example, the classification result for the cluster in this embodiment is not limited to the 5A model, and may also include multiple clusters obtained according to other classification strategies, which is not described herein again.
For example, assuming that the target platform is an application of a short video platform as an example, and further assuming that after time series data corresponding to a current user operation behavior of a current user account ID1 to be classified is acquired, it is determined that the current user account ID1 clicks many multimedia resources (such as videos) of "motion types" for playing and viewing, behavior similarity between the current user operation behavior and each core operation behavior may be determined based on the time series data.
Assume that the existing cluster includes: exposure, attraction, action, advocacy, asset. And (3) determining that the behavior similarity between the current user operation behavior of the playing and watching and the core operation behavior in the cluster action reaches a matching condition through comparison, and classifying the current user operation behavior (namely the playing and watching behavior) into the cluster action.
It should be noted that the user account may be, but is not limited to, performing a series of sequential operation behaviors (which may be regarded as a behavior link), such as exposure, playing for 3s (regarded as being attracted), playing for 5s (regarded as completing playing), exiting playing, and the like. After the cluster to which the user account belongs is determined according to the operation behavior in the behavior link of the user account, the characteristics of the user account in the cluster can be calculated in the background when the platform selects and displays the user people in the specified cluster, so that the attention to the multimedia resources can be promoted.
Optionally, in this embodiment, the above-mentioned manner of determining the behavior similarity between the current user operation behavior and each core operation behavior may be, but is not limited to, adopting a Dynamic Time Warping (DTW) algorithm. The DTW algorithm is based on the idea of Dynamic Programming (DP), and describes the time corresponding relation between a test template and a reference template by using a time rule function meeting a certain condition, and solves the regularization function corresponding to the minimum accumulated distance when the two templates are matched. The explanation is assumed to take the time series data of the current user operation behavior and the time series data of the core operation behavior in this embodiment as an example:
to align two time series, a matrix grid may be constructed, but not limited to, with the elements in the matrix being the distance between two points (i.e., the similarity between each point of one series and each point of the other series, the smaller the distance, the higher the similarity), such as euclidean distance. The similarity between two sequences can be concluded by finding a set of distances through the path between points in the matrix grid.
Optionally, in this embodiment, the behavior similarity may be determined directly by using a calculation result of the dynamic warping algorithm (such as a minimum value of DTW for short), and may also be, but is not limited to, jointly determine the behavior similarity by combining at least one of the following distance parameter values: maximum distance value, mean distance value, minimum value of the combined variance, maximum value of the correlation coefficient, etc.
Optionally, in this embodiment, the core operation behavior matched with each cluster may be determined in advance from candidate user operation behaviors, and the number of the core operation behaviors matched with each cluster may be one or multiple, which is not limited herein.
According to the embodiment provided by the application, the cluster to which the current user operation behavior belongs is determined by comparing the behavior similarity between the current user operation behavior to be classified and the core operation behavior determined for each cluster, so that the current user operation behavior can be classified more accurately according to the behavior similarity. And further the problem of lower accuracy in the user behavior classification method in the related technology is solved.
As an alternative, the determining the behavior similarity between the current user operation behavior and each core operation behavior according to the time series data includes:
s1, determining a first time sequence curve corresponding to the current user operation behavior according to the time sequence data, and determining a second time sequence curve corresponding to the core operation behavior;
s2, calculating the distance of a path between any one first point position on the first time sequence curve and any one second point position on the second time sequence curve to obtain a plurality of distance values, wherein the first point position and the second point position meet the path boundary condition between the first time sequence curve and the second time sequence curve;
and S3, determining the behavior similarity according to the plurality of distance values.
It should be noted that, in this embodiment, the time series data may be, but is not limited to, a statistical result of the current user operation behavior in the target time period, where the statistical result includes a behavior tag corresponding to the current user operation behavior and a corresponding behavior timestamp. Therefore, in this embodiment, after a large number of user operation behaviors of different user accounts are acquired, the attribute of the current user operation behavior may be determined based on the behavior tag, and the current user operation behavior may be counted based on the behavior timestamp.
Further, in the present embodiment, a corresponding time series curve can be drawn based on the above time series data. In the time-series curve, the horizontal axis represents time, and the vertical axis represents the statistical result of the execution number of the user operation behaviors at different time points.
It should be noted that, in the present embodiment, the statistical result may include, but is not limited to, statistics in the form of the number of independent visitors, or statistics in the form of the number of visits. The number of independent visitors (uv) refers to the number of access times after deduplication, that is, the number of accesses of the same identity is counted as one independent access. And the access times (Page View, pv for short), namely the single-Page click rate, are used for counting the Page access amount without distinguishing identities.
Further, but not limited to, the time sequence curve is subjected to characteristic engineering, and the minimum value min to the maximum value max are subjected to normalization processing to obtain a curve of 0-1. Further, a similarity matrix in a spectral clustering algorithm is absorbed to calculate the similarity of the two time sequence curves, and the similarity is used as the behavior similarity between the current user operation behavior to be classified and the current core operation behavior for comparison.
Specifically, referring to the example shown in fig. 3, it is assumed that the DTW algorithm is used to perform the calculation, and the two timing curves shown in a diagram a and a diagram B in fig. 3 are respectively a first timing curve (shown as a solid line) corresponding to the current user operation behavior and a second timing curve (shown as a dashed line) corresponding to the core operation behavior. Wherein, a connecting line between two curves in the diagram B is a DTW path of the two curves in the diagram a, and the DTW value is determined according to a sum of respective distance values (e.g., euclidean distances) of the plurality of connecting lines, so as to serve as behavior similarity between respective corresponding operation behaviors of the two timing curves.
It should be noted that, the determining of the path satisfying the path boundary condition in the DTW algorithm may be, but is not limited to, by the following manners: defining the path as a warping path-structured path and denoted by W, the k-th element of W is defined as Wk＝(i，j)kA mapping of sequences Q (e.g., sequences corresponding to the first timing curve) and C (e.g., sequences corresponding to the second timing curve) is defined. The above timing curve will satisfy the following characteristics:
1) path boundary conditions (may be simply referred to as boundary conditions): w is a1(1,1) and wk(m, n). The selected path must start from the lower left corner and end at the upper right corner.
2) Continuity: if w isk-1(a ', b') then for the next point w of the pathk(a, b) is required to satisfy (a-a')<1 and (b-b')<1. I.e. it is not possible to match across a certain point, but only to align with its own neighbouring points. This ensures that each coordinate in Q and C appears in W.
3) Monotonicity: if w isk-1(a ', b') then for the next point w of the pathk(a, b) is required to satisfy 0<(a-a') and 0<(b-b'). The points above W are monotonically progressing over time. To ensure that the dashed lines in diagram B do not intersect.
Combining continuity and monotonicity constraints, the path of each grid point has only three directions. For example, if the path has passed through lattice point (i, j), then the next passing lattice point may be only one of the following three cases: (i +1, j), (i, j +1), or (i +1, j + 1). The effect can be shown in fig. 4, and is not described in detail here.
According to the embodiment provided by the application, after the first time sequence curve corresponding to the current user operation behavior and the second time sequence curve corresponding to the core operation behavior are obtained, the distance value between the point meeting the path boundary condition and the point between the two time sequence curves is determined, and a plurality of distance values are obtained. And calculating the behavior similarity between the current user operation behavior and each core operation behavior according to the plurality of distance values by adopting a dynamic time warping algorithm, so as to determine a target core operation behavior with the similarity to the current user operation behavior based on the behavior similarity, and further determine a target cluster (cluster matched with the target core operation behavior) to which the current user operation behavior belongs.
As an alternative, determining the behavior similarity according to the plurality of distance values includes:
s1, carrying out weighted summation on the squares of the distance values to obtain a reference distance value;
s2, determining the minimum distance value of the square root of the reference distance value within the path section indicated by the path boundary condition as the behavior similarity.
For example, assume that the timing curve shown in FIG. 3 is still used for explanation: the first time sequence curve (shown as a solid line in the figure) corresponding to the current user operation behavior and the second time sequence curve (shown as a dotted line in the figure) corresponding to the core operation behavior. And (3) calculating the two time sequence curves by adopting a dynamic time warping algorithm and a dp dynamic programming algorithm as follows:
wherein x and y respectively represent two timing curves (x corresponds to the current user operation behavior, and y corresponds to the core operation behavior), i represents the ith point on the first timing curve x, and j represents the jth point on the second timing curve y. d (x)i、yj) Represents xi、yjThe distance between them. The value space pi represents the range of the paths that can be traveled by the points on the two timing curves x and y (i.e. the path boundary strip)Piece). It is not required that x, y have the same length and extent, as long as the starting boundaries of the samples are given, ensuring that they are continuous unidirectional.
According to the embodiment provided by the application, the DTW algorithm is adopted to determine the distance values of a plurality of connecting lines between two time sequence curves, wherein the connecting lines meet the path boundary condition, and a plurality of distance values are obtained. The method is favorable for more accurately determining the behavior similarity between the current user operation behavior and the core operation behavior according to the plurality of distance values, so that the accuracy of the classification result of the user operation behavior is ensured.
As an alternative, determining the behavior similarity according to the plurality of distance values includes:
s1, obtaining a distance parameter value between the current user operation behavior and the core operation behavior, wherein the distance parameter value comprises at least one of the following: a maximum distance value of the plurality of distance values, an average distance value of the plurality of distance values, a minimum value of a combined variance of the plurality of distance values, a maximum value of a correlation coefficient of the plurality of distance values;
and S2, carrying out weighted summation on the minimum distance value and the distance parameter value to obtain the behavior similarity.
For example, suppose that the minimum distance value DTW is divided byminIn addition, the distance parameter value: the maximum distance value a among the plurality of distance values, the average distance value b among the plurality of distance values, the minimum value c of the combined variance of the plurality of distance values, and the maximum value d of the correlation coefficient of the plurality of distance values. Based on the above, the behavior similarity can be calculated by referring to the following formula:
P＝DTWmin*w1+a*w2+b*w3+c*w4+d*w5
wherein, w1To w5Is [0, 1 ]]The weighting coefficients between the two groups can be configured according to actual needs, but not limited to.
According to the embodiment provided by the application, the behavior similarity is calculated by combining the minimum distance value and the distance parameter value between the current user operation behavior and the core operation behavior, so that the accuracy of the behavior similarity is further ensured.
As an optional scheme, before classifying the current user operation behavior into a target cluster matched with the target core operation behavior, the method further includes:
1) under the condition that the target core operation behavior comprises a core operation behavior, determining a matching condition meeting the target core operation behavior under the condition that the behavior similarity is smaller than a first threshold value;
2) under the condition that the target core operation behavior comprises at least two core operation behaviors, acquiring a similarity average value of behavior similarities corresponding to the current user operation behavior and each core operation behavior; and determining a matching condition meeting the target core operation behavior under the condition that the similarity average value is smaller than a second threshold value.
It should be noted that the core operation behavior matched with each cluster may be one or multiple. Here, when the target core operation behavior is a core operation behavior, the obtained behavior similarity may be directly compared with a first threshold, and when the behavior similarity is smaller than the first threshold, it is determined that the current user operation behavior satisfies a matching condition of the target core operation behavior, and is similar to the target core operation behavior, and the current user operation behavior may be further classified into a target cluster corresponding to the target core operation behavior. And under the condition that the target core operation behavior is at least two core operation behaviors, calculating the average value of the behavior similarity between the current user operation behavior and each core operation behavior matched with one cluster to obtain the similarity average value. And under the condition that the similarity average value is smaller than a second threshold, classifying the user operation gender into a target cluster corresponding to the target core operation behavior. The first threshold and the second threshold may be, but not limited to, comparison thresholds set according to different scenarios, and the first threshold and the second threshold may be the same value or different values.
In addition, in this embodiment, for the number of core operation behaviors respectively matched with different cluster clusters, behavior similarity or a similarity average value between the current user operation behavior and the core operation behavior of each cluster is respectively determined. And determining a cluster corresponding to the minimum value from the behavior similarity or the similarity average value, and taking the cluster as a target cluster. And then classifying the current user operation behavior into the target cluster.
According to the embodiment provided by the application, whether the current user operation behavior meets the corresponding matching condition is determined in different modes according to the number of the target core operation behaviors in each cluster. Therefore, the target cluster to which the user operation behavior belongs is quickly determined, and the accuracy of classification is ensured.
As an optional scheme, before determining the behavior similarity between the user operation behavior and each core operation behavior according to the time series data, the method further includes:
s1, according to the resource characteristics of the multimedia resources, core operation behaviors respectively matched with each cluster are determined from a plurality of candidate user operation behaviors, wherein the plurality of candidate user operation behaviors comprise current user operation behaviors, and each cluster is matched with at least one core operation behavior.
It should be noted that, in this embodiment, the idea of the Kmeans and the aggregative clustering algorithm may be absorbed at the same time, and the core operation behavior of each cluster is determined from the existing multiple candidate user operation behaviors according to the resource characteristics of the multimedia resource.
For example, assume that three cluster clusters are preset in this example, and respectively include: the core operation behavior of the first cluster A1 is determined as exposure, the core operation behavior of the second cluster A2 is determined as click, comment …, and the core operation behavior of the third cluster A3 is determined as sharing. Further, the plurality of candidate user operation behaviors includes a1 through a 10. And then selecting the core operation behaviors matched with the three clustering clusters from a1 to a10 by adopting the clustering algorithm. Wherein the selection may be, but is not limited to, based on resource characteristics of the multimedia resource.
Assume that the core operation behavior of the first cluster a1 is a1, the core operation behavior of the second cluster a2 is A3, and the core operation behavior of the third cluster A3 is a7, a 10. Then, the classification method described in the above embodiment is further employed.
If the current user operation behavior to be currently classified is a5, acquiring a time sequence curve corresponding to the time sequence data corresponding to a5 and a time sequence curve corresponding to each core operation behavior in a1 to A3, then calculating the similarity between the two time sequence curves by using a DTW algorithm, and further determining the behavior similarity or the average value of the similarities, where the behavior similarity between a5 and a1 in a1 is P1, the behavior similarity between a5 and A3 in a2 is P2, and the average value of the behavior similarities between a7 and a10 in a5 and A3 is P3. Comparing the P1 to the P3 to determine that the minimum value is P2, it can be determined that the current user operation behavior a5 is to be classified into the target cluster a2 corresponding to P2.
According to the embodiment provided by the application, the core operation behaviors matched with each cluster are predetermined, so that the core operation behaviors are compared with the current user operation behaviors to be classified conveniently, and the more accurate classification processing of the current user operation behaviors is realized through the behavior similarity obtained through comparison.
As an optional scheme, before acquiring time-series data corresponding to a current user operation behavior to be classified, the method further includes:
s1, behavior link data corresponding to a plurality of candidate user operation behaviors are obtained, wherein the behavior link data corresponding to each candidate user operation behavior comprises a behavior label and a behavior timestamp of a group of ordered operation behaviors;
and S2, counting the behavior link data according to a target period to obtain time sequence data corresponding to each candidate user operation behavior.
It should be noted that, in this embodiment, the behavior link data of each of the multiple candidate user operations is the corresponding behavior record data, which includes a behavior tag and a behavior timestamp of a set of ordered operation behaviors. After the behavior link data are respectively counted according to the target period, the corresponding time sequence data can be obtained.
For example, after acquiring behavior link data of each of N candidate user operations, summarizing the behavior link data, and counting corresponding uv data or pv data by day to obtain corresponding time series data. Behaviors as in a piece of behavioral link data may include, but are not limited to: expose, play 3s, play 5s, play completed, click, like, comment …, and so on.
Fig. 5 is a schematic structural diagram of a user behavior classification apparatus for implementing a user behavior classification method according to an exemplary embodiment. Referring to fig. 5, the apparatus includes:
a first obtaining unit 502, configured to obtain time series data corresponding to a current user operation behavior to be classified, where the time series data is used to record a statistical result of the current user operation behavior in a target time period, and the current user operation behavior is an operation behavior executed by a user account on a multimedia resource displayed in a target platform;
a first determining unit 504 configured to determine a behavior similarity between the current user operation behavior and each core operation behavior according to the time-series data, wherein the core operation behavior is an operation behavior matched with each cluster determined from various user operation behaviors;
the classifying unit 506 is configured to classify the current user operation behavior into a target cluster matched with the target core operation behavior under the condition that the behavior similarity meets the matching condition of the target core operation behavior.
Optionally, in this embodiment, the user behavior classification method may be applied, but not limited to, in a decision process of a control policy for an online information flow. For example, taking the multimedia resource allocation field as an example, the online control policy herein may include but is not limited to: a strategy for controlling multimedia resource over-allocation, a strategy for improving budget consumption rate, a cold start strategy (here, cold start refers to a release strategy referred to when user preference is unknown at the time of first release), and the like. That is to say, the cluster to which the user operation behavior belongs is determined through analysis of the user operation behaviors, so that the multimedia resource allocation strategy correspondingly adopted by the current user operation behavior is determined. In addition, the target platform may be, but not limited to, different terminal Applications (APPs), such as a short video platform Application, a video playing Application, a community space sharing Application, an instant messaging Application, an online shopping Application, and the like. The above is an example, and the application scenarios and fields are not limited in any way.
It should be noted that, in this embodiment, the current user operation behavior to be classified refers to the current user operation behavior to be currently classified, and the data pool actually stores time series data of a plurality of different user operation behaviors.
Optionally, in this embodiment, the operation performed by each user account on the multimedia resource presented on the platform may include, but is not limited to, at least one of the following: exposure, play, attention, comment, share, pay, etc. That is, each user action has an action tag and carries an action timestamp.
Optionally, in this embodiment, each cluster may be, but is not limited to, divided according to resource characteristics (e.g., service requirements) of the multimedia resource. Taking the multimedia resource allocation procedure as an example, the link cycle designed for the multimedia resource may include, but is not limited to: exposure (Aware), attraction (Appeal), action (Act), user (Advocate), Assets (Assets), five parts, called the 5A model. The main application scene of the 5A model is the measurement of the full life cycle value of brand consumers, and the applied product function modules comprise customer asset overview, 5A link analysis and the like; the latter is also applied to modules of link flow and the like. In this embodiment, the current user operation behavior may be classified into each cluster according to a comparison result of the behavior similarity, so as to achieve the purpose of accurate classification. Here, this is an example, the classification result for the cluster in this embodiment is not limited to the 5A model, and may also include multiple clusters obtained according to other classification strategies, which is not described herein again.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
As an alternative, the first determining unit includes:
the first determining module is configured to determine a first time sequence curve corresponding to the current user operation behavior according to the time sequence data and determine a second time sequence curve corresponding to the core operation behavior;
the calculation module is configured to calculate a distance of a path between any one first point position on the first timing curve and any one second point position on the second timing curve to obtain a plurality of distance values, wherein the first point position and the second point position meet a path boundary condition between the first timing curve and the second timing curve;
a second determination module configured to determine the behavioral similarity according to the plurality of distance values.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
As an alternative, the second determining module includes:
the processing submodule is configured to perform weighted summation on the squares of the plurality of distance values to obtain a reference distance value;
a first determination submodule configured to determine a minimum distance value of a square root of the reference distance value within a path section indicated by the path boundary condition as the behavior similarity.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
As an alternative, the second determining module includes:
an obtaining submodule configured to obtain a distance parameter value between a current user operation behavior and a core operation behavior, wherein the distance parameter value includes at least one of: a maximum distance value of the plurality of distance values, an average distance value of the plurality of distance values, a minimum value of a combined variance of the plurality of distance values, a maximum value of a correlation coefficient of the plurality of distance values;
and the calculation submodule is configured to perform weighted summation on the minimum distance value and the distance parameter value to obtain the behavior similarity.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
As an optional solution, the apparatus further includes:
the second determining unit is configured to determine that the matching condition of the target core operation behavior is met under the condition that the behavior similarity is smaller than the first threshold value under the condition that the target core operation behavior comprises one core operation behavior before the current user operation behavior is classified into the target cluster matched with the target core operation behavior;
the third determining unit is configured to acquire a similarity average value of behavior similarities corresponding to the current user operation behavior and each core operation behavior under the condition that the target core operation behavior comprises at least two core operation behaviors; and determining a matching condition meeting the target core operation behavior under the condition that the similarity average value is smaller than a second threshold value.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
As an optional solution, the apparatus further includes:
and the fourth determining unit is configured to determine core operation behaviors respectively matched with each cluster from a plurality of candidate user operation behaviors according to the resource characteristics of the multimedia resource before determining the behavior similarity between the current user operation behavior and each core operation behavior according to the time sequence data, wherein the plurality of candidate user operation behaviors comprise the current user operation behavior, and each cluster is matched with at least one core operation behavior.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
As an optional solution, the apparatus further includes:
the second obtaining unit is configured to obtain behavior link data corresponding to a plurality of candidate user operation behaviors before obtaining time series data corresponding to current user operation behaviors to be classified, wherein the behavior link data corresponding to each candidate user operation behavior comprises a group of behavior tags and behavior time stamps of ordered operation behaviors;
and the statistical unit is configured to perform statistics on the behavior link data according to a target period to obtain time sequence data corresponding to each candidate user operation behavior.
The embodiments of the apparatus herein may refer to the embodiments of the method described above, but are not described here again.
With regard to the apparatus in the above-described embodiment, the specific manner in which each module performs the operation has been described in detail in the embodiment related to the method, and will not be elaborated here.
In an exemplary embodiment, a computer-readable storage medium comprising instructions, such as a memory comprising instructions, executable by a processor to perform the steps of any of the above method embodiments is also provided. Alternatively, the computer readable storage medium may be a ROM, a Random Access Memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.
In an exemplary embodiment, a computer program product is also provided, the computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The processor of the computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions to cause the computer device to perform the above-described method. Wherein the computer program is arranged to perform the steps of any of the above method embodiments when executed.
Other embodiments of the disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the disclosure disclosed herein. This application is intended to cover any variations, uses, or adaptations of the disclosure following, in general, the principles of the disclosure and including such departures from the present disclosure as come within known or customary practice within the art to which the disclosure pertains. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the disclosure being indicated by the following claims.
It will be understood that the present disclosure 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 present disclosure is limited only by the appended claims.
Claims (10)
1. A user behavior classification method is characterized by comprising the following steps:
acquiring time sequence data corresponding to current user operation behaviors to be classified, wherein the time sequence data is used for recording a statistical result of the current user operation behaviors in a target time period, and the current user operation behaviors are operation behaviors executed on multimedia resources displayed in a target platform by a current user account;
determining behavior similarity between the current user operation behavior and each core operation behavior according to the time sequence data, wherein the core operation behavior is the operation behavior matched with each cluster determined from various user operation behaviors;
and under the condition that the behavior similarity meets the matching condition of the target core operation behavior, classifying the current user operation behavior into a target cluster matched with the target core operation behavior.
2. The method of claim 1, wherein the determining behavior similarities between the current user operational behavior and each core operational behavior from the time series data comprises:
determining a first time sequence curve corresponding to the current user operation behavior according to the time sequence data;
determining a second timing curve corresponding to the core operational behavior;
calculating the distance of a path between any one first point position on the first timing curve and any one second point position on the second timing curve to obtain a plurality of distance values, wherein the first point position and the second point position meet the path boundary condition between the first timing curve and the second timing curve;
determining the behavior similarity according to the plurality of distance values.
3. The method of claim 2, wherein the determining the behavioral similarity from the plurality of distance values comprises:
carrying out weighted summation on the squares of the plurality of distance values to obtain a reference distance value;
determining a minimum distance value of a square root of the reference distance value within a path interval indicated by the path boundary condition as the behavior similarity.
4. The method of claim 3, wherein the determining the behavioral similarity from the plurality of distance values comprises:
obtaining a distance parameter value between the current user operation behavior and the core operation behavior, wherein the distance parameter value includes at least one of: a maximum distance value of the plurality of distance values, an average distance value of the plurality of distance values, a minimum value of a combined variance of the plurality of distance values, a maximum value of a correlation coefficient of the plurality of distance values;
and carrying out weighted summation on the minimum distance value and the distance parameter value to obtain the behavior similarity.
5. The method of claim 1, further comprising, prior to said classifying said current user operational behavior into a target cluster that said target core operational behavior matches:
determining that a matching condition of the target core operation behavior is satisfied under the condition that the behavior similarity is smaller than a first threshold value under the condition that the target core operation behavior comprises one core operation behavior;
under the condition that the target core operation behavior comprises at least two core operation behaviors, acquiring a similarity average value of behavior similarities corresponding to the current user operation behavior and each core operation behavior; determining a matching condition that satisfies the target core operation behavior if the similarity average is less than a second threshold.
6. The method of claim 1, further comprising, prior to said determining a behavioral similarity between the current user operational behavior and each core operational behavior from the time series data:
and according to the resource characteristics of the multimedia resources, determining core operation behaviors which are respectively matched with each cluster from a plurality of candidate user operation behaviors, wherein the plurality of candidate user operation behaviors comprise the current user operation behaviors, and each cluster is matched with at least one core operation behavior.
7. A user behavior classification apparatus, comprising:
the device comprises a first obtaining unit, a second obtaining unit and a third obtaining unit, wherein the first obtaining unit is configured to obtain time sequence data corresponding to current user operation behaviors to be classified, the time sequence data is used for recording a statistical result of the current user operation behaviors in a target time period, and the current user operation behaviors are operation behaviors executed on multimedia resources displayed in a target platform by a user account;
a first determining unit configured to determine behavior similarity between the current user operation behavior and each core operation behavior according to the time series data, wherein the core operation behavior is an operation behavior matched with each cluster, which is determined from various user operation behaviors;
and the classifying unit is configured to classify the current user operation behavior into a target cluster matched with the target core operation behavior under the condition that the behavior similarity meets the matching condition of the target core operation behavior.
8. A computer-readable storage medium, whose instructions, when executed by a processor of an electronic device, cause the electronic device to perform the user behavior classification method of any of claims 1 to 6.
9. An electronic device, comprising:
a processor;
a memory for storing the processor-executable instructions;
wherein the processor is configured to execute the instructions to implement the user behavior classification method of any of claims 1 to 6.
10. A computer program product comprising computer programs/instructions, characterized in that the computer programs/instructions, when executed by a processor, implement the user behavior classification method of any of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111017409.0A CN113821574B (en) | 2021-08-31 | 2021-08-31 | User behavior classification method and device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111017409.0A CN113821574B (en) | 2021-08-31 | 2021-08-31 | User behavior classification method and device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113821574A true CN113821574A (en) | 2021-12-21 |
| CN113821574B CN113821574B (en) | 2024-07-30 |
Family
ID=78913960
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111017409.0A Active CN113821574B (en) | 2021-08-31 | 2021-08-31 | User behavior classification method and device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113821574B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114780606A (en) * | 2022-03-30 | 2022-07-22 | 欧阳安安 | Big data mining method and system |
| CN115222461A (en) * | 2022-09-19 | 2022-10-21 | 杭州数立信息技术有限公司 | Intelligent marketing accurate recommendation method |
Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446078A (en) * | 2016-09-08 | 2017-02-22 | 乐视控股（北京）有限公司 | Information recommendation method and recommendation apparatus |
| CN109754113A (en) * | 2018-11-29 | 2019-05-14 | 南京邮电大学 | Load forecasting method based on dynamic time warping Yu length time memory |
| CN111258435A (en) * | 2020-01-15 | 2020-06-09 | 北京达佳互联信息技术有限公司 | Multimedia resource commenting method and device, electronic equipment and storage medium |
| CN111460910A (en) * | 2020-03-11 | 2020-07-28 | 深圳市新镜介网络有限公司 | Face type classification method and device, terminal equipment and storage medium |
| CN111666351A (en) * | 2020-05-29 | 2020-09-15 | 北京睿知图远科技有限公司 | Fuzzy clustering system based on user behavior data |
| CN112131322A (en) * | 2020-09-22 | 2020-12-25 | 腾讯科技（深圳）有限公司 | Time series classification method and device |
| CN112138403A (en) * | 2020-10-19 | 2020-12-29 | 腾讯科技（深圳）有限公司 | Interactive behavior recognition method and device, storage medium and electronic equipment |
| CN112826514A (en) * | 2019-11-22 | 2021-05-25 | 华为技术有限公司 | Atrial fibrillation signal classification method, device, terminal and storage medium |
| CN113011886A (en) * | 2021-02-19 | 2021-06-22 | 腾讯科技（深圳）有限公司 | Method and device for determining account type and electronic equipment |
| CN113051442A (en) * | 2019-12-26 | 2021-06-29 | 中国电信股份有限公司 | Time series data processing method, device and computer readable storage medium |
| CN113190757A (en) * | 2021-05-17 | 2021-07-30 | 清华大学 | Multimedia resource recommendation method and device, electronic equipment and storage medium |
-
2021
- 2021-08-31 CN CN202111017409.0A patent/CN113821574B/en active Active
Patent Citations (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106446078A (en) * | 2016-09-08 | 2017-02-22 | 乐视控股（北京）有限公司 | Information recommendation method and recommendation apparatus |
| CN109754113A (en) * | 2018-11-29 | 2019-05-14 | 南京邮电大学 | Load forecasting method based on dynamic time warping Yu length time memory |
| CN112826514A (en) * | 2019-11-22 | 2021-05-25 | 华为技术有限公司 | Atrial fibrillation signal classification method, device, terminal and storage medium |
| CN113051442A (en) * | 2019-12-26 | 2021-06-29 | 中国电信股份有限公司 | Time series data processing method, device and computer readable storage medium |
| CN111258435A (en) * | 2020-01-15 | 2020-06-09 | 北京达佳互联信息技术有限公司 | Multimedia resource commenting method and device, electronic equipment and storage medium |
| CN111460910A (en) * | 2020-03-11 | 2020-07-28 | 深圳市新镜介网络有限公司 | Face type classification method and device, terminal equipment and storage medium |
| CN111666351A (en) * | 2020-05-29 | 2020-09-15 | 北京睿知图远科技有限公司 | Fuzzy clustering system based on user behavior data |
| CN112131322A (en) * | 2020-09-22 | 2020-12-25 | 腾讯科技（深圳）有限公司 | Time series classification method and device |
| CN112138403A (en) * | 2020-10-19 | 2020-12-29 | 腾讯科技（深圳）有限公司 | Interactive behavior recognition method and device, storage medium and electronic equipment |
| CN113011886A (en) * | 2021-02-19 | 2021-06-22 | 腾讯科技（深圳）有限公司 | Method and device for determining account type and electronic equipment |
| CN113190757A (en) * | 2021-05-17 | 2021-07-30 | 清华大学 | Multimedia resource recommendation method and device, electronic equipment and storage medium |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114780606A (en) * | 2022-03-30 | 2022-07-22 | 欧阳安安 | Big data mining method and system |
| CN115222461A (en) * | 2022-09-19 | 2022-10-21 | 杭州数立信息技术有限公司 | Intelligent marketing accurate recommendation method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113821574B (en) | 2024-07-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20210326674A1 (en) | Content recommendation method and apparatus, device, and storage medium | |
| CN108476334B (en) | Cross-screen optimization of advertisement placement | |
| WO2020135535A1 (en) | Recommendation model training method and related apparatus | |
| TW202007178A (en) | Method, device, apparatus, and storage medium of generating features of user | |
| CN109783730A (en) | Products Show method, apparatus, computer equipment and storage medium | |
| CN107358247B (en) | Method and device for determining lost user | |
| CN115860833A (en) | Television advertisement slot targeting based on consumer online behavior | |
| CN113821574B (en) | User behavior classification method and device and storage medium | |
| US20180033027A1 (en) | Interactive user-interface based analytics engine for creating a comprehensive profile of a user | |
| CN109993627B (en) | Recommendation method, recommendation model training device and storage medium | |
| CN112613938B (en) | Model training method and device and computer equipment | |
| CN116821475B (en) | Video recommendation method and device based on client data and computer equipment | |
| CN112070564B (en) | Advertisement pulling method, device and system and electronic equipment | |
| CN107633257A (en) | Data Quality Assessment Methodology and device, computer-readable recording medium, terminal | |
| CN111861605A (en) | Business object recommendation method | |
| CN114938458B (en) | Object information display method and device, electronic equipment and storage medium | |
| CN113919923A (en) | Live broadcast recommendation model training method, live broadcast recommendation method and related equipment | |
| CN111738766A (en) | Data processing method and device for multimedia information and server | |
| US20240202588A1 (en) | Systems and methods for cohort-based predictions in clustered time-series data in order to detect significant rate-of-change events | |
| CN115730125A (en) | Object identification method and device, computer equipment and storage medium | |
| CN112115354A (en) | Information processing method, information processing apparatus, server, and storage medium | |
| CN116701896A (en) | Image tag determining method, image tag determining device, computer device, and storage medium | |
| CN116186119A (en) | User behavior analysis method, device, equipment and storage medium | |
| CN113672816B (en) | Account feature information generation method and device, storage medium and electronic equipment | |
| CN113724044A (en) | User portrait based commodity recommendation, apparatus, computer device and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |