CN114661984A - Information pushing method based on artificial intelligence and user portrait and cloud computing system - Google Patents

Abstract

The disclosed embodiment provides an information pushing method based on artificial intelligence and user portrait and a cloud computing system, wherein a basic user portrait corresponding to target user interest data is determined according to a pre-trained user portrait classification model, dynamic expansion data related to the target user interest data is further acquired on the basis, a dynamic expansion portrait corresponding to the dynamic expansion data is determined based on the user portrait classification model, so that a target user portrait corresponding to a service user can be determined according to the basic user portrait and the dynamic expansion portrait, corresponding personalized service item link information is pushed to service using equipment in real time based on the target user portrait corresponding to the service user, not only the user interest data but also the related dynamic expansion data are considered, and the personalized service item link information is pushed after the basic user portrait is optimized, the accuracy of personalized push can be improved.

Description

Information pushing method based on artificial intelligence and user portrait and cloud computing system

Technical Field

The disclosure relates to the technical field of big data, in particular to an information pushing method based on artificial intelligence and user portrait and a cloud computing system.

Background

With the development of big data and cloud computing technology, various internet service platforms aim to provide convenient information content information service for users by providing various personalized services. In the process of developing and operating personalized services, determining the user portrait of the related user is an indispensable step. The user representation is a tagged user model abstracted according to the attention interest data information related to the user attribute, user preference, living habits, user behaviors and the like. Colloquially, a user is labeled, and the label is a highly refined characteristic mark obtained by analyzing user information. By tagging the user with some highly generalized, easily understandable features, the user can be made more easily understandable by the machine and computer processing can be facilitated.

In the related technology, for the determination of the user portrait, large data mining is needed to obtain related user concern interest data, and portrait classification is carried out on the user concern interest data based on a user portrait classification model obtained through artificial intelligence training.

Disclosure of Invention

In order to overcome at least the above disadvantages in the prior art, an object of the present disclosure is to provide an information pushing method based on artificial intelligence and user portrait and a cloud computing system.

In a first aspect, the present disclosure provides an information pushing method based on artificial intelligence and a user portrait, which is applied to a cloud computing system, where the cloud computing system is communicatively connected to a plurality of service usage devices, and the method includes:

determining a basic user portrait corresponding to target user interest data according to a pre-trained user portrait classification model;

acquiring dynamic expansion data related to the target user interest data, and determining a dynamic expansion portrait corresponding to the dynamic expansion data based on the user portrait classification model;

determining a target user portrait corresponding to a service user according to the basic user portrait and the dynamic extension portrait;

and pushing corresponding personalized service item link information to the service using equipment in real time based on the target user portrait corresponding to the service user.

In an embodiment of the first aspect, the target user attention interest data is obtained by:

acquiring first user interest data mined by a first big data mining model to a service user corresponding to the service using equipment, wherein the first user interest data comprises the user interest data mined by the first big data mining model under a first big data mining path sequence;

identifying second user interest data matched with the first user interest data in a user interest database, wherein the second user interest data comprises user interest data mined by a second big data mining model under a second big data mining path sequence;

updating the mining path in the first big data mining path sequence according to the first big data mining path sequence and the second big data mining path sequence to obtain a third big data mining path sequence;

updating the first user interest data according to the third big data mining path sequence to obtain target user interest data, wherein the target user interest data comprise big data mining paths corresponding to the target user interest points;

and carrying out user portrait classification on the target user interest data to obtain a target user portrait corresponding to the service user.

In an embodiment of the first aspect, in an independent concept, the updating the mining path in the first big data mining path sequence according to the first big data mining path sequence and the second big data mining path sequence to obtain a third big data mining path sequence includes:

determining a second big data mining path sequence according to the first big data mining path sequence, wherein the first user attention interest data are obtained by the first big data mining model along a first mining service path, the first mining service path comprises first cluster mining service units, and the first big data mining path sequence comprises unit mining paths of the first big data mining model on every two mining service units related to the first cluster mining service units and global mining paths of the first big data mining model on every mining service unit in the first cluster mining service units;

determining a third mining path sequence and a fourth mining path sequence according to the second big data mining path sequence, wherein the second user interest data is obtained by the first big data mining model along a second mining service path, the second mining service path comprises second cluster mining service units, and the second big data mining path sequence comprises unit mining paths of the second big data mining model on every two mining service units related to the first cluster mining service units and global mining paths of the second big data mining model on every mining service unit in the first cluster mining service units;

and updating the mining path in the first big data mining path sequence by using the global mining path of the second big data mining model according to the unit mining path sequence of the first big data mining model, the global mining path of the first big data mining model and the unit mining path sequence of the second big data mining model to obtain a third big data mining path sequence.

In an embodiment of an independent concept of the first aspect, the updating, according to the sequence of the unit mining paths of the first big data mining model, the global mining path of the first big data mining model, and the sequence of the unit mining paths of the second big data mining model, the mining path in the sequence of the first big data mining path with the global mining path of the second big data mining model to obtain the sequence of the third big data mining path includes:

determining a first mining path relationship network according to the first big data mining path sequence and the second big data mining path sequence, wherein the first mining path relationship network comprises the first cluster mining service unit, a first group of mining service association attributes, the second cluster mining service unit, a second group of mining service association attributes and a third group of mining service association attributes, the first set of mined service association attributes includes mined service association attributes between every two mined service units in the first cluster of mined service units, the second group of mining service association attributes comprise mining service association attributes between every two mining service units in the second cluster of mining service units, and the third group of mining service association attributes comprise mining service association attributes between the mining service units in the first cluster of mining service units and the mining service units in the second cluster of mining service units;

acquiring a unit mining path between two mining service units communicated with each mining service association attribute in the third group of mining service association attributes;

determining whether a redundant mining service association attribute exists in the third group of mining service association attributes according to a unit mining path between two mining service units communicated by each mining service association attribute in the third group of mining service association attributes;

when redundant mining service association attributes exist in the third group of mining service association attributes, hiding the redundant mining service association attributes in the first mining path relationship network to obtain a second mining path relationship network;

and updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence according to the second mining path relationship network to obtain a third mining path sequence and a fourth mining path sequence, wherein the third mining path sequence is a mining path sequence obtained by updating the first group of global mining paths in the first big data mining path sequence into a third group of global mining paths, and the fourth mining path sequence is a mining path sequence obtained by updating the second group of global mining paths in the second big data mining path sequence into a fourth group of global mining paths.

In an embodiment of the first aspect, the determining whether a redundant mining service association attribute exists in the third set of mining service association attributes according to a unit mining path between two mining service units connected by each mining service association attribute in the third set of mining service association attributes includes:

determining a connection relation network formed by each mining service correlation attribute in the third group of mining service correlation attributes, the first group of mining service correlation attributes and the mining service correlation attributes in the second group of mining service correlation attributes to obtain a first group of connection relation networks;

and determining whether redundant mining service association attributes exist in the third group of mining service association attributes according to the unit mining path corresponding to the mining service association attributes in each of the first group of communication relation networks.

In an embodiment of the first aspect, in which an independent concept is implemented, the determining a connectivity network formed by each mining service association attribute in the third set of mining service association attributes, the first set of mining service association attributes, and the mining service association attributes in the second set of mining service association attributes to obtain a first set of connectivity network includes:

when the preset connection relation network association attribute number comprises N attribute numbers, executing the following steps for each attribute number in the N attribute numbers to obtain the first group of connection relation networks, wherein N is 1 or a natural number greater than 1, and when the following steps are executed, each attribute number is the current attribute number:

determining a connected relationship network formed by each mining service correlation attribute in the third group of mining service correlation attributes and the mining service correlation attributes in the first group of mining service correlation attributes and the second group of mining service correlation attributes, wherein the number of the mining service correlation attributes included in the formed connected relationship network is the current attribute number.

In an embodiment of an independent concept of the first aspect, the determining whether a redundant mining service association attribute exists in the third set of mining service association attributes according to a unit mining path corresponding to a mining service association attribute in each of the first set of connected networks includes:

executing the following steps for each mined service association attribute in the third group of mined service association attributes, wherein each mined service association attribute is a current mined service association attribute when the following steps are executed:

determining a second group of connectivity networks comprising the current mining service association attribute in the first group of connectivity networks;

determining redundancy measurement parameters of the current mining service association attribute according to unit mining paths corresponding to mining service association attributes in each of the second group of communication relationship networks;

and when the redundancy measurement parameter of the current mining service associated attribute does not meet the preset condition, determining the current mining service associated attribute as a redundant mining service associated attribute.

In an embodiment of an independent concept of the first aspect, the determining the redundancy measurement parameter of the current mining service association attribute according to the unit mining path corresponding to the mining service association attribute in each of the second group of connected networks includes:

for each connectivity network in the second set of connectivity networks, performing the following steps, wherein each connectivity network is a current connectivity network when performing the following steps:

converting the unit excavation path corresponding to each excavation service correlation attribute in the current connectivity network into an excavation path mesh to obtain a group of excavation path meshes, wherein parameters in the excavation path meshes are used for representing excavation service nodes and excavation dimensions in the unit excavation paths;

performing primary fusion on the group of excavation path grids according to a preset sequence to obtain a target grid, wherein parameters in the target grid are used for representing excavation service nodes and excavation dimensions in a target unit excavation path, and the target unit excavation path is an accumulated unit excavation path formed by unit excavation paths corresponding to each excavation service correlation attribute in the current connectivity network;

converting parameters in the target grid into target redundancy variables;

determining a redundancy measurement parameter corresponding to the current connectivity network according to the target redundancy variable;

determining an accumulated value of the redundancy measurement parameters corresponding to each communication relation network in the second group of communication relation networks as the parameters of the redundancy measurement parameters of the current mining service association attributes;

wherein, when the redundancy measurement parameter of the current mining service association attribute does not meet the preset condition, determining the current mining service association attribute as a redundant mining service association attribute includes:

and when the parameter of the redundancy measurement parameter of the current mining service associated attribute is less than 0, determining the current mining service associated attribute as a redundant mining service associated attribute.

In an embodiment of an independent concept of the first aspect, the updating, according to the second mining path relationship network, a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence includes:

updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence according to a first limiting strategy to obtain a third mining path sequence and a fourth mining path sequence, wherein the first limiting strategy is a limiting strategy determined according to unit mining paths corresponding to a fourth group of mining service correlation attributes in the second mining path relationship network;

wherein, according to a first constraint policy, updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence, and includes:

and updating the global excavation path in the first big data excavation path sequence and the global excavation path in the second big data excavation path sequence, so that the sum of the loss between the unit excavation path obtained by each excavation service correlation attribute in the second excavation path relationship network and the unit excavation path recalculated by each excavation service correlation attribute is the minimum, wherein the unit excavation path recalculated by each excavation service correlation attribute is the unit excavation path calculated by the global excavation path on the excavation service unit communicated by each excavation service correlation attribute.

In an embodiment of an independent concept of the first aspect, the updating, according to the second mining path relationship network, a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence includes:

updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence according to a second limiting strategy to obtain a third mining path sequence and a fourth mining path sequence, wherein the second restriction policy is a restriction policy determined according to a unit mining path corresponding to a fourth group of mining service association attributes in the second mining path relationship network and the first group of prior mining paths, the fourth set of mined service association attributes includes mined service association attributes of the third set of mined service association attributes other than the redundant mined service association attribute, each prior excavation path in the first group of prior excavation paths is a prior excavation path of an excavation service unit in the first cluster excavation service unit;

wherein, according to a second constraint policy, updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence, and includes:

updating a global mining path in the first sequence of big data mining paths and a global mining path in the second sequence of big data mining paths such that a sum of a first global penalty and a second global penalty is minimized, wherein the first global loss is the sum of losses between the unit mining path obtained by each mining service associated attribute in the second mining path relationship network and the unit mining path recalculated by each mining service associated attribute, the unit mining path recalculated by each mining service associated attribute is a unit mining path calculated by a global mining path on the mining service unit communicated with each mining service associated attribute, the second global loss is a sum of losses between each prior mined path in the first set of prior mined paths and a global mined path on a corresponding mined business unit;

preferably, after the global mining path in the first big data mining path sequence and the global mining path in the second big data mining path sequence are updated according to the first big data mining path sequence and the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence, the method further includes:

updating a first group of user attention interest data in the first user attention interest data according to a third group of global mining paths in the third mining path sequence, wherein the third mining path sequence is a mining path sequence obtained by updating the first group of global mining paths in the first big data mining path sequence into the third group of global mining paths, the first user interest data is the user interest data mined by the first big data mining model along the first mining business path, the first set of global mining paths includes global mining paths of the first big data mining model on a third cluster mining business unit of the first cluster mining business units, the first group of user interest data comprises user interest data mined by the first big data mining model on the third cluster mining service unit;

updating a second group of user interest data in second user interest data according to a fourth group of global mining paths in the fourth mining path sequence to obtain the target user interest data, wherein the fourth mining path sequence is a mining path sequence obtained by updating the second group of global mining paths in the second big data mining path sequence to the fourth group of global mining paths, the second user interest data is the user interest data mined by the second big data mining model along the second mining service path, the second set of global mining paths includes global mining paths of the second big data mining model on a fourth cluster mining service unit of the second cluster mining service units, the second group of user interest data comprises user interest data mined by the second big data mining model on the fourth cluster mining service unit.

In a second aspect, an embodiment of the present disclosure further provides an information pushing system based on artificial intelligence and a user portrait, where the information pushing system based on artificial intelligence and a user portrait includes a cloud computing system and a plurality of service usage devices in communication connection with the cloud computing system;

the cloud computing system is configured to:

determining a basic user portrait corresponding to target user interest data according to a pre-trained user portrait classification model;

acquiring dynamic expansion data related to the target user interest data, and determining a dynamic expansion portrait corresponding to the dynamic expansion data based on the user portrait classification model;

determining a target user portrait corresponding to a service user according to the basic user portrait and the dynamic extension portrait;

and pushing corresponding personalized service item link information to the service using equipment in real time based on the target user portrait corresponding to the service user.

According to any one of the above aspects, a base user portrait corresponding to target user interest data is determined according to a pre-trained user portrait classification model, further acquiring dynamic expansion data related to the interest data concerned by the target user on the basis of the user portrait classification model, determining a dynamic expansion portrait corresponding to the dynamic expansion data on the basis of the user portrait classification model, thereby, the target user portrait corresponding to the service user can be determined according to the basic user portrait and the dynamic expansion portrait, and pushes corresponding personalized service item link information to the service using equipment in real time based on the target user figure corresponding to the service user, and considers not only the user interest data but also the related dynamic expansion data, therefore, the personalized business item link information is pushed after the basic user portrait is optimized, and the accuracy of personalized pushing can be improved.

Drawings

FIG. 1 is a schematic diagram of an application environment of an information push system based on artificial intelligence and a user portrait according to an embodiment of the present disclosure;

FIG. 2 is a schematic flowchart of an information pushing method based on artificial intelligence and a user profile according to an embodiment of the present disclosure;

fig. 3 is a schematic block diagram of a cloud computing system for implementing the above-described artificial intelligence and user portrait-based information pushing method according to an embodiment of the present disclosure.

Detailed Description

The present disclosure is described in detail below with reference to the drawings, and the specific operation methods in the method embodiments can also be applied to the device embodiments or the system embodiments.

Fig. 1 is a schematic application environment diagram of an information push system 10 based on artificial intelligence and user representation according to an embodiment of the present disclosure. The information pushing system 10 based on artificial intelligence and user portrayal can comprise a cloud computing system 100 and a business use device 200 which is in communication connection with the cloud computing system 100. The artificial intelligence and user representation based information push system 10 shown in FIG. 1 is only one possible example, and in other possible embodiments, the artificial intelligence and user representation based information push system 10 may also include only at least some of the components shown in FIG. 1 or may also include other components.

In this embodiment, the cloud computing system 100 and the service using apparatus 200 in the artificial intelligence and user portrait based information push system 10 may cooperatively perform the artificial intelligence and user portrait based information push method described in the following method embodiment, and the detailed description of the method embodiment may be referred to in the specific steps of the cloud computing system 100 and the service using apparatus 200.

The artificial intelligence and user portrait based information pushing method provided by the present embodiment may be executed by the cloud computing system 100 shown in fig. 1, and is described in detail below.

And step S110, determining a basic user portrait corresponding to the target user interest data according to a pre-trained user portrait classification model.

Step S120, dynamic expansion data relevant to the target user interest data are obtained, and a dynamic expansion portrait corresponding to the dynamic expansion data is determined based on the user portrait classification model.

Step S130, determining a target user portrait corresponding to the service user according to the basic user portrait and the dynamic extension portrait.

Step S140, based on the target user portrait corresponding to the service user, pushing corresponding personalized service item link information to the service using equipment in real time.

In this embodiment, the pre-trained user portrait classification model may be obtained by training based on pre-collected user interest data samples and user tag vectors corresponding to the user interest data samples.

In this embodiment, the dynamic expansion data related to the interest data of the target user may refer to other interest related data having a business session collaborative relationship with the current interest node of the interest data of the target user.

In this embodiment, in step S130, the base user portrait and the dynamic extended portrait may be fused with portrait tag vectors to determine a target user portrait corresponding to the service user, for example, an extended portrait tag vector related to the dynamic extended portrait, which is not included in the portrait tag vector set of the base user portrait, may be fused with the portrait tag vector set of the base user portrait to obtain the target user portrait.

In this embodiment, in step S140, in the process of pushing corresponding personalized service item link information to the service using device in real time based on the target user profile corresponding to the service user, click tendency parameters of the service user for clicking each candidate personalized service item connection information respectively may be predicted according to the target user profile, and after determining a potential personalized service item of each candidate personalized service item connection information according to each click tendency parameter, the potential personalized service item is pushed to the service using device.

Based on the above steps, the present embodiment determines the basic user portrait corresponding to the target user interest data according to the pre-trained user portrait classification model, further acquiring dynamic expansion data related to the interest data concerned by the target user on the basis of the user portrait classification model, determining a dynamic expansion portrait corresponding to the dynamic expansion data on the basis of the user portrait classification model, thereby, the target user portrait corresponding to the service user can be determined according to the basic user portrait and the dynamic expansion portrait, and pushes corresponding personalized service item link information to the service using equipment in real time based on the target user figure corresponding to the service user, and considers not only the user interest data but also the related dynamic expansion data, therefore, the personalized business item link information is pushed after the basic user portrait is optimized, and the accuracy of personalized pushing can be improved.

In an independently contemplated embodiment, with respect to step S120, the disclosed embodiment further provides a user portrait mining data expansion method based on big data, which may be implemented by the following steps.

Step C11: and acquiring the extension session behavior data of the extension session flow between the current interest-concerned node of the interest-concerned data of the target user and the interest-concerned node of the cooperative interaction of the current interest-concerned node and the interest-concerned node.

In an embodiment based on independent concept, the extended session behavior data may include information of the interactive behavior instance detected by the session application connecting interest nodes, such as: connection flow nodes, etc., and the extended session behavior data may also include state data of interest nodes, such as: a service delivery start node, a service delivery end node, etc.

Step C12: based on the extended session behavior data, reference persistence behavior data for the extended session flow is determined.

In an embodiment based on the independent concept, the reference to the apersistence behavior data may be understood as reference data to the apersistence behavior data. For example, a first activation node that activates a current interest node from the extended session flow by the interactive behavior instance and a second activation node that activates a collaborative interaction interest node of the interest node may be determined based on a range of flow nodes where the first connection flow node and the second connection flow node are located, respectively, so as to determine a behavior engagement degree between the second activation node and the first connection flow node.

A reference apersistence behavior data in an independent concept based embodiment may include, but is not limited to: low-frequency behavior data of the persistent behavior data, and common behavior data of the persistent behavior data.

Step C13: and screening the interaction behavior examples of the expanded conversation process in the preset time sequence range by referring to the persistent behavior data to obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity in the preset time sequence range.

Step C14: and acquiring the behavior data of the interest ending activity, determining the interest ending activity of which the behavior data meets the first screening condition as the interest triggering activity, and loading the interest triggering activity into the interest triggering log.

Step C15: and determining dynamic expansion data of the expansion conversation process within a preset time sequence range through the interest trigger log.

Based on the steps, extended session behavior data of an extended session process between the current interest-interested node and the interest-interested node cooperatively interacting with the current interest-interested node is obtained, reference persistence behavior data of the extended session process is determined, and accordingly, interaction behavior instances of the extended session process within a preset time sequence range are screened by the reference persistence behavior data, an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity within the preset time sequence range are obtained, and the influence of non-interest trigger activities on the determination of dynamic extended data can be avoided. And the interest ending activity of which the behavior data accord with the first screening condition is determined as the interest triggering activity and is loaded to the interest triggering log, so that the dynamic expansion data of the session flow can be expanded in a preset time sequence range through the interest triggering log, and the accuracy of the dynamic expansion data is improved.

In an embodiment based on the independent concept, the reference apersistence behavior data includes common behavior data of the apersistence behavior data and low frequency behavior data of the apersistence behavior data. Step C13 is specifically configured to filter the interaction behavior instances of the extended session flow within a preset time sequence range through the common behavior data of the persistent behavior data, and obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity within the preset time sequence range.

In an embodiment based on independent concept, before determining, by the interest trigger log, dynamic extension data of the extended session flow within the preset time sequence range, the method further includes: and eliminating the interest trigger activities which do not accord with the second screening condition in the interest trigger log through the low-frequency behavior data of the continuous behavior data.

In an embodiment based on independent conception, the extended session behavior data of the extended session flow includes a first connection flow node where an interactive behavior instance connects the current interest-concerned node from the extended session flow and a second connection flow node where the interactive behavior instance connects the collaborative interaction interest-concerned node, and the preset timing range includes a first preset timing range between a service delivery start node and a service delivery end node and a second preset timing range between a subsequent service delivery start node and a subsequent service delivery end node after the service delivery end node;

in an independent concept-based embodiment, determining the reference persistence behavior data of the extended session flow based on the extended session behavior data comprises:

step C101, determining a first activation node of an interactive behavior instance for activating the current interest-concerned node from the extended session flow and a second activation node of the interest-concerned node of the collaborative interaction for activating the interactive behavior instance based on the flow node range of the first connection flow node and the second connection flow node;

step C102, summarizing the behavior data of all interactive behavior instances of the extended session process through the second activation node and the first connection process node;

step C103, analyzing the behavior data of all the interactive behavior instances of the extended session flow, and acquiring the low-frequency behavior data of the continuous behavior data and the common behavior data of the continuous behavior data;

in an embodiment based on independent conception, the screening, by using the common behavior data of the persistent behavior data, the interaction behavior instances of the extended session flow within a preset time sequence range to obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity within the preset time sequence range includes:

step C201, summarizing a first target interaction behavior instance of the first activation node in the first preset time sequence range and a second target interaction behavior instance of the first activation node in the second preset time sequence range;

step C202, a first interest trigger log and a first interest end log are screened from the first target interaction behavior instance through comparison information among the first connection process node, the first activation node, the service delivery starting node, the service delivery ending node, the behavior data and common behavior data of the continuous behavior data;

step C203, screening out a second interest trigger log and a second interest end log from the second target interaction behavior instance through comparison information among the first connection process node, the first activation node, the next service release starting node, the next service release ending node, the behavior data and common behavior data of the persistent behavior data;

in an embodiment based on independent concept, the step C14 of obtaining behavior data of the end-of-interest activity, and regarding the end-of-interest activity whose behavior data meets the first filtering condition as the interest triggering activity and loading the interest triggering log includes:

step C301, acquiring behavior data of the interest ending activities in the first interest ending log, determining the interest ending activities of which the behavior data meet first screening conditions as interest triggering activities, and loading the interest ending activities into the first interest triggering log;

step C302, acquiring the behavior data of the interest ending activities in the second interest ending log, determining the interest ending activities of which the behavior data meet the first screening conditions as interest triggering activities, and loading the interest ending activities into the second interest triggering log;

the removing, by the low-frequency behavior data of the persistent behavior data, the interest trigger activities in the interest trigger log that do not meet the second filtering condition includes: and eliminating the interest trigger activities which do not accord with the second screening condition in the first interest trigger log through the low-frequency behavior data of the continuous behavior data, and eliminating the interest trigger activities which do not accord with the second screening condition in the second interest trigger log.

The determining, by the interest trigger log, dynamic extension data of the extended session flow within the preset time sequence range includes: and determining dynamic expansion data of the expansion session flow in the first preset time sequence range through a first interest trigger log after the removing processing, and determining dynamic expansion data of the expansion session flow in the second preset time sequence range through a second interest trigger log after the removing processing.

In an embodiment based on independent conception, the analyzing the behavior data of all the interaction behavior instances of the extended session flow, and the obtaining the low-frequency behavior data of the persistent behavior data and the common behavior data of the persistent behavior data includes: sequencing and sorting the behavior data of all the interactive behavior instances to obtain a behavior data sequence; acquiring behavior data in a preset frequency range in the behavior data sequence, and considering the behavior data in the preset frequency range as low-frequency behavior data of the continuous behavior data; and screening the behavior data within the range of the set behavior tag in the behavior data sequence, and regarding the shared behavior data part of the screened behavior data as the common behavior data of the continuous behavior data.

In an embodiment based on independent conception, the determining, based on a process node range in which the first connection process node and the second connection process node are located, a first activation node at which an interactive behavior instance activates the current interest-of-interest node from the extended session process, and a second activation node at which an interactive behavior instance activates the collaborative interactive interest-of-interest node includes: if the first connection process node is in the range of the activated process node, the first connection process node is regarded as the first activated node; if the first connection process node is in the range of the inactive process node, the service delivery starting node after the range of the inactive process node is finished is regarded as the first active node; if the second connection process node is in the range of the activation process node, the second connection process node is regarded as the second activation node; and if the second connection process node is in the range of the non-activated process node, the service delivery starting node after the range of the non-activated process node is finished is regarded as the second activated node.

In an embodiment based on an independent concept, the screening out a first interest trigger log and a first interest end log from the first target interactive behavior instance through comparison information between common behavior data of the first connection process node, the first activation node, the service delivery start node, the service delivery end node, the behavior data and the persistent behavior data includes: sequencing the first target interaction behavior instances according to the sequence of the first connection process nodes; judging whether a first interactive behavior example in the sorted first target interactive behavior examples accords with a first preset activity characteristic or not; if yes, all the first target interaction behavior instances are determined as non-interest triggering activities; if not, the first target interaction behavior instances after being managed are sequentially analyzed, and the first target interaction behavior instances are loaded to the first interest trigger log or the first interest end log based on the analysis result.

In an embodiment based on independent conception, the method includes sequentially parsing the sorted first target interaction behavior instances, and loading the first target interaction behavior instances to the first interest trigger log or the first interest end log based on a parsing result, where the method includes: the first end information is configured by default as a first metric parameter.

The analyzing the sorted first target interaction behavior instances in sequence, and the loading the first target interaction behavior instances to the first interest trigger log or the first interest end log based on the analysis result includes: if the second activation node of the current first target interactive behavior instance is prior to the service delivery starting node, the current first target interactive behavior instance and the subsequent first target interactive behavior instance are determined as the non-interest triggering activities; if the behavior engagement degree between the first activation node and the first connection process node of the current first target interaction behavior instance is within the first target behavior engagement degree, and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interaction behavior instance to the first interest trigger log; if the behavior engagement degree between the first activation node and the first connection process node of the current first target interaction behavior instance is within the first target behavior engagement degree and the ending metric parameter of the first ending information is not the first metric parameter, loading the current first target interaction behavior instance to the first interest ending log; if the first connection process node of the current first target interactive behavior instance is smaller than the service delivery starting node and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interactive behavior instance to the first interest trigger log; if the first connection process node of the current first target interactive behavior instance is smaller than the service delivery starting node and the ending measurement parameter of the first ending information is not the first measurement parameter, loading the current first target interactive behavior instance to the first interest ending log; if the behavior engagement degree between the first activation node of the current first target interaction behavior instance and the first activation node of the previous first target interaction behavior instance conforms to a second preset activity characteristic, the current first target interaction behavior instance and the subsequent first target interaction behavior instance are considered as non-interest triggering activities; if the behavior engagement degree of the current first target interaction behavior instance is equal to zero and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interaction behavior instance to the first interest trigger log; if the behavior engagement degree of the current first target interaction behavior instance is equal to zero and the ending measurement parameter of the first ending information is not the first measurement parameter, loading the current first target interaction behavior instance to the first interest ending log; if the behavior engagement degree of the current first target interaction behavior instance is greater than or equal to the first comparison routine of the common behavior data of the continuous behavior data and is the engagement degree, and the ending metric parameter of the first ending information is the first metric parameter, loading the current first target interaction behavior instance to the first interest trigger log; if the behavior engagement degree of the current first target interactive behavior instance is greater than or equal to the first comparison routine of the common behavior data of the continuous behavior data and is the engagement degree, and the ending metric parameter of the first ending information is not the first metric parameter, loading the current first target interactive behavior instance to the first interest ending log; if the behavior engagement degree of the current first target interaction behavior instance is smaller than the second proportion behavior engagement degree of the common behavior data of the continuous behavior data, the current first target interaction behavior instance and the subsequent first target interaction behavior instances are determined as the non-interest triggering activities; and if quantitative comparison information between the behavior data of the current first target interactive behavior instance and the common behavior data of the continuous behavior data is between the first ratio routine degree and the second ratio behavior engagement degree, setting the first end information as a second metric parameter, and loading the current first target interactive behavior instance into a first interest end log.

In an embodiment based on the independent concept, the first ratio routine is a degree of engagement greater than the second ratio behavior degree of engagement; and/or, the first preset activity characteristic comprises any one of: the behavior connection degree between the first connection process node and the service delivery starting node is greater than a first preset behavior connection degree; the behavior connection degree between the first connection process node and the service delivery starting node is greater than a second preset behavior connection degree, the behavior connection degree between the first activation node and the first connection process node is less than a third preset behavior connection degree, and the behavior connection degree is less than the first ratio routine of the common behavior data of the continuous behavior data;

the first preset behavior engagement degree is greater than the second preset behavior engagement degree and the third preset behavior engagement degree; and/or the second preset activity characteristic comprises any one of the following: the behavior connection degree between the first activation node of the current first target interaction behavior example and the first activation node of the previous first target interaction behavior example is larger than a fourth preset behavior connection degree, and the first activation node is smaller than the service release end node; the behavior connection degree between a first activation node of a current first target interactive behavior example and a first activation node of a previous first target interactive behavior example is greater than a fifth preset behavior connection degree, the first activation node is smaller than the service release end node, the behavior connection degree between the first activation node and the first connection process node is smaller than a sixth preset behavior connection degree, and the behavior connection degree is smaller than the first ratio of common behavior data of the continuous behavior data and is a connection degree; and the fourth preset behavior engagement degree is greater than the fifth preset behavior engagement degree and the sixth preset behavior engagement degree.

In an embodiment based on an independent concept, the aforementioned target user interest data may be implemented by the following steps.

Step R110, first user attention interest data mined by a first big data mining model is obtained, wherein the first user attention interest data comprises user attention interest data mined by the first big data mining model under a first big data mining path sequence.

And step R120, identifying second user interest data matched with the first user interest data in a user interest database, wherein the second user interest data comprises user interest data mined by a second big data mining model under a second big data mining path sequence.

And step R130, updating the mining path in the first big data mining path sequence according to the first big data mining path sequence and the second big data mining path sequence to obtain a third big data mining path sequence.

And step R140, updating the first user interest data according to the third big data mining path sequence to obtain target user interest data, wherein the target user interest data comprises a big data mining path corresponding to each target user interest point.

Based on the above steps, in this embodiment, by acquiring the first user attention interest data, searching the second user attention interest data matched with the first user attention interest data in the user attention interest database, mining a path sequence according to the first big data and the second big data, updating the mining path in the first big data mining path sequence to obtain a third mining path sequence, therefore, the mining path is updated and optimized, the interest data concerned by the first user is updated, after the interest data concerned by the target user is obtained, classifying the user portrait of the target user interest data to obtain a target user portrait corresponding to the target user interest data, therefore, the scheme of optimizing and updating the large data mining path is considered by combining the user attention interest data mined in different forms, and the accuracy of user portrait classification can be improved.

In an embodiment based on the same concept, the artificial intelligence and user portrait based information pushing method provided by the present embodiment may be executed by the cloud computing system 100 shown in fig. 1, and the artificial intelligence and user portrait based information pushing method is described in detail below.

Step A110, obtaining first user attention interest data, wherein the first user attention interest data comprise a first big data mining path sequence, the first user attention interest data are user attention interest data determined according to data mined by a first big data mining model along a first mining service path, the first mining service path comprises first cluster mining service units, and the first big data mining path sequence comprises unit mining paths of the first big data mining model on every two mining service units related in the first cluster mining service units and global mining paths of the first big data mining model on every mining service unit in the first cluster mining service units.

Step A120, searching a second user interest data matched with the first user interest data in a user interest database, the second user attention interest data are user attention interest data which are obtained by determining data mined along a second mining service path according to a second big data mining model, mining service path distribution corresponding to the second mining service path and mining service path distribution corresponding to the first mining service path are in cross distribution, the second mining service path comprises second cluster mining service units, the second user attention interest data comprise second big data mining path sequences, and the second big data mining path sequences comprise unit mining paths of the second big data mining model on every two mining service units related in the second cluster mining service units and global mining paths of the second big data mining model on every mining service unit in the second cluster mining service units.

Step A130, updating the global excavation path in the first big data excavation path sequence and the global excavation path in the second big data excavation path sequence according to the first big data excavation path sequence and the second big data excavation path sequence to obtain a third excavation path sequence and a fourth excavation path sequence, wherein the third excavation path sequence and the fourth excavation path sequence meet a target restriction strategy.

Step A140, updating a first group of user attention interest data in the first user attention interest data according to a third group of global mining paths in a third mining path sequence, wherein the third mining path sequence is a mining path sequence obtained by updating the first group of global mining paths in the first big data mining path sequence into the third group of global mining paths, the first user attention interest data is the user attention interest data mined by the first big data mining model along the first mining service path, the first group of global mining paths includes the global mining path of the first big data mining model on a third cluster mining service unit in the first cluster mining service unit, and the first group of user attention interest data includes the user attention interest data mined by the first big data mining model on the third cluster mining service unit. Updating a second group of user attention interest data in the second user attention interest data according to a fourth group of global mining paths in the fourth mining path sequence to obtain target user attention interest data, the fourth mining path sequence is a mining path sequence obtained by updating a second group of global mining paths in the second big data mining path sequence into a fourth group of global mining paths, the second user attention interest data comprises second user attention interest data, the second user attention interest data is user attention interest data mined by the second big data mining model along the second mining service path, the second group of global mining paths comprises global mining paths of the second big data mining model on a fourth cluster mining service unit in the second cluster mining service unit, and the second group of user attention interest data comprises user attention interest data mined by the second big data mining model on the fourth cluster mining service unit.

In an independently conceived embodiment, a big data mining rule is deployed on a first big data mining model, and interest point information in user interest data mined by the first big data mining model along a mining service path can be obtained according to the deployed big data mining rule.

Searching matched second user attention interest data in a user attention interest database according to mining service path distribution corresponding to the first user attention interest data, wherein the mining service path distribution corresponding to the second user attention interest data is partially or completely crossed with the mining service path distribution corresponding to the first user attention interest data. That is to say, the user interest data which has a cross section with the first user interest data is found, the user interest data is also the data mined by the second big data mining model along the second mining service path, the second big data mining model may be the first big data mining model, that is, the first user interest data and the second user interest data are two pieces of user interest data obtained by two rounds of mining by the same big data mining model along the same mining service path, or two pieces of user interest data mined by two different big data mining model objects along two different mining service paths.

In other words, the first user attention interest data and the second user attention interest data mining may be by:

the first method is as follows: the big data mining model object 1 mines interest data of a first user along the mining service path 1, and the big data mining model object 2 mines interest data of a second user along the mining service path 1. And (4) mining user interest data mined by different big data mining model objects through the same mining service path.

The second method comprises the following steps: the big data mining model object 1 mines interest data of a first user along the mining business path 1, and the big data mining model object 1 mines interest data of a second user along the mining business path 1. The same big data mining model object mines the interest data of the users concerned in multiple times through the same mining service path.

The third method comprises the following steps: the big data mining model object 1 mines interest data of a first user along the mining business path 1, and the big data mining model object 1 mines interest data of a second user along the mining business path 2. And the user interest data mined by different big data mining model objects through the same mining service path.

For example, the user interest database may include a plurality of user interest data, each of the plurality of user interest data may include a mining path sequence, and each of the user interest data is determined according to data mined by a big data mining model along a mining service path. The data mining method includes the steps that a plurality of user interest data in a user interest database can be data mined by a data mining model for multiple times, wherein each user interest data in the plurality of user interest data in the user interest database is similar to first user interest data and can comprise a mining path sequence, namely the user interest data in the user interest database and the data in the first user interest data are data with the same dimension.

In an embodiment of an independent concept, the second user interest data may be user interest data determined according to data mined by a second big data mining model along a second mining service path; the data mining method can also be used for mining the data of interest of the user determined by the first big data mining model along the second mining business path, or can also be used for mining the data of interest of the user determined by the first big data mining model along the first mining business path again.

The second mining service path may include a second cluster mining service unit, the second user attention interest data may include a second big data mining path sequence, and the second big data mining path sequence includes a unit mining path of a second big data mining model on each of two mining service units associated with the second cluster mining service unit, and a global mining path of the second big data mining model on each mining service unit in the second cluster mining service unit. The unit mining path is determined according to rule configuration information of a big data mining rule on the big data mining model, and the unit mining path is determined by a root global mining path. And the second user pays attention to the determination mode of the second big data mining path sequence and the determination mode of the first big data mining path sequence in the interest data.

In the embodiment of independent conception, the global service node of the big data mining model on the mining service unit in each mining path sequence can be updated according to the plurality of mining path sequences, and the common updating of interest data concerned by a plurality of users can be realized, namely the integral updating of the plurality of mining path sequences is realized, so that accurate mining data is obtained.

For example, each mining path sequence includes a cluster of mining service units, the mining service units are connected to obtain a mining service unit network, and the whole mining service unit network is updated, that is, the data of each mining path sequence is updated to obtain the whole update of a plurality of mining path sequences.

For example, assume that the mining service unit includes 2 mining path sequences, each mining path sequence corresponds to a user interest data mined by the big data mining model along the mining service path, that is, 2 mining path sequences correspond to 2 paths, each path corresponds to a mining service path, and a point on each path corresponds to a mining service unit.

And updating the global mining paths of the big data mining models on all the mining path sequence mining service units according to the global mining paths and the unit mining paths in the first big data mining path sequence and the global mining paths and the unit mining paths in the second big data mining path sequence.

The design is carried out, first user attention interest data are obtained, wherein the first user attention interest data comprise a first big data mining path sequence, the first user attention interest data are user attention interest data which are obtained by determining data mined along a first mining service path according to a first big data mining model, the first mining service path comprises first cluster mining service units, and the first big data mining path sequence comprises unit mining paths of the first big data mining model on every two mining service units related to the first cluster mining service units and global mining paths of the first big data mining model on every mining service unit in the first cluster mining service units; searching a second user interest data matched with the first user interest data in a user interest database, the second user attention interest data are user attention interest data which are determined according to data mined by a second big data mining model along a second mining service path, mining service path distribution corresponding to the second mining service path and mining service path distribution corresponding to the first mining service path are in cross distribution, the second mining service path comprises second cluster mining service units, the second user attention interest data comprise second big data mining path sequences, the second big data mining path sequences comprise unit mining paths of the second big data mining model on every two mining service units related to the second cluster mining service units in the second cluster mining service units, and global mining paths of the second big data mining model on every mining service unit in the second cluster mining service units; updating the global mining path in the first big data mining path sequence and the global mining path in the second big data mining path sequence according to the first big data mining path sequence and the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence, wherein the third mining path sequence and the fourth mining path sequence meet a target limiting strategy to achieve the aim of integrally updating the interest data of a plurality of users, namely, the mining paths are optimized and updated to ensure that the mining path loss meets a preset limiting strategy, thereby improving the accuracy of subsequent user image classification, so that after the integral updating, the mining path information of the interest data of the users concerned with the mining interest of the users at this time is optimized and updated, and the updating can be synchronized into a user interest database, therefore, the interest characteristics of users in the following portrait classification are considered, the related mining path characteristics which are accurately matched are also considered, and the accuracy of portrait classification is improved.

In an embodiment of independent concept, updating a global mining path in a first big data mining path sequence and a global mining path in a second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence may include:

s201, determining a first mining path relation network according to the first big data mining path sequence and the second big data mining path sequence, the first mining path relation network comprises a first cluster mining service unit, a first group of mining service association attributes, a second cluster mining service unit, a second group of mining service association attributes and a third group of mining service association attributes, wherein the first group of mining service association attributes comprise mining service association attributes between every two mining service units associated in the first cluster mining service unit, the second group of mining service association attributes comprise mining service association attributes between every two mining service units associated in the second cluster mining service unit, and the third group of mining service association attributes comprise mining service association attributes between the mining service units in the first cluster mining service unit and the mining service units in the second cluster mining service unit;

s202, acquiring a unit mining path between two mining service units communicated with each mining service association attribute in the third group of mining service association attributes;

s203, determining whether a redundant mining service correlation attribute exists in the third group of mining service correlation attributes according to a unit mining path between two mining service units communicated by each mining service correlation attribute in the third group of mining service correlation attributes;

s204, when the redundant mining service correlation attributes exist in the third group of mining service correlation attributes, hiding the redundant mining service correlation attributes in the first mining path relation network to obtain a second mining path relation network;

and S205, updating a first group of global excavation paths in the first big data excavation path sequence and a second group of global excavation paths in the second big data excavation path sequence according to a second excavation path relation network to obtain a third excavation path sequence and a fourth excavation path sequence, wherein the third excavation path sequence is an excavation path sequence obtained by updating the first group of global excavation paths in the first big data excavation path sequence into a third group of global excavation paths, and the fourth excavation path sequence is an excavation path sequence obtained by updating the second group of global excavation paths in the second big data excavation path sequence into a fourth group of global excavation paths.

In an embodiment of an independent concept, the second big data mining path sequence may include one or more, for example, the second big data mining path sequence includes 1, and it is assumed that 1 first big data mining path sequence and 1 second big data mining path sequence, i.e., 2 mining path sequences, and the mining path sequence is numbered W, T, where each mining path sequence includes a unit mining path of the first big data mining model on each associated two mining service units in a cluster of mining service units, and a global mining path of the first big data mining model on each mining service unit in a cluster of mining service units. The mining path sequence W comprises 5 mining service units, the mining path sequence T comprises 6 mining service units, the first group of mining service association attributes comprise 4 mining service association attributes W12, W23, W34 and W45 corresponding to the mining path sequence W, the second group of mining service association attributes comprise 5 mining service association attributes T12, T23, T34, T45 and T56 corresponding to the mining path sequence T, and the third group of mining service association attributes comprise mining service association attributes R1 to R10 between the mining service units in the first cluster of mining service units and the mining service units in the second cluster of mining service units.

The determining of the first mining path relationship network according to the first big data mining path sequence and the second big data mining path sequence may be understood as performing connectivity relationship network configuration according to the first big data mining path sequence and the second big data mining path sequence.

In an embodiment of independent concept, it needs to be determined that there exists a redundant mining service association attribute in the first mining path relationship network, where the redundant mining service association attribute is that the unit mining path loss of the mining service association attribute in each connected relationship network is not 0, for example, if the mining service association attribute U4 exists in the connected relationship network 1 formed by the mining service association attributes U3, U4 and the mining service association attribute W23 and also exists in the connected relationship network 2 formed by U4, U5 and the mining service association attribute T23, it is determined according to the global loss of the unit mining path in the two connected relationship networks where U4 exists, whether the mining service association attribute U4 needs to be hidden or not, it needs to be explained that the loss ls1 of the mining service association attribute U4 in the connected relationship network 1 is equal to the connected product of the mining service association attributes U3, U4 and the unit mining service association attribute W23 corresponding to the mining service association attribute, and the loss ls2 of the mining service correlation attribute U4 in the connected relation network 2 is equal to the product of the mining service correlation attribute U4 and the mining service correlation attribute U5 and the unit mining path corresponding to the mining service correlation attribute T23, whether the mining service correlation attribute 4 needs to be hidden is determined according to the losses ls1 and ls2, and if the sum of the losses ls1 and ls2 is less than 0, the mining service correlation attribute 4 is hidden.

For example, the second big data mining path sequence may include 2, and the mining path sequence is numbered W, T, U assuming that it includes 1 first big data mining path sequence and 2 second big data mining path sequences, i.e., 3 mining path sequences, where each mining path sequence includes a unit mining path of the first big data mining model on each associated two mining service units in a cluster of mining service units, and a global mining path of the first big data mining model on each mining service unit in a cluster of mining service units. The mining path sequence W comprises 5 mining service units, the mining path sequence T comprises 6 mining service units, the mining path sequence W comprises 5 mining service units, the first group of mining service associated attributes comprises 4 mining service associated attributes W12, W23, W34 and W45 corresponding to the mining path sequence W, the second group of mining service associated attributes comprises 9 mining service associated attributes T12, T23, T34, T45 and T56, U12, U23, U34 and U45 corresponding to the mining path sequence T and the mining path U, and the third group of mining service associated attributes comprises mining service units in a first cluster of mining service units, mining service units in a second cluster of mining service units and R1-R15 among mining service units in a third cluster of mining service units.

In an embodiment of an independent concept, it needs to be determined that a redundant mining service correlation attribute exists in a first mining path relationship network, where the redundant mining service correlation attribute is that a unit mining path loss of the mining service correlation attribute in each connected relationship network is not 0, it is assumed that a mining service correlation attribute R5 exists in a connected relationship network 1 formed by mining service correlation attributes R5, R4 and mining service correlation attribute T34, and also exists in a connected relationship network 2 formed by R5, R10 and R11, and also exists in a connected relationship network 3 formed by R5, R9, U23 and R11, and also exists in a connected relationship network 4 formed by mining service correlation attributes R5, R6 and W23, and according to global loss determination of unit mining paths in 4 connected relationship networks where R5 exists, whether the mining service correlation attribute R5 needs to be hidden or not, and it needs to be explained that a loss of the mining service correlation attribute R5 in the connected relationship network 1 is equal to a unit mining service correlation attribute R5, The continuous product of the unit mining paths corresponding to the mining service correlation attributes W34 and R4, the loss ls2 of the mining service correlation attribute R5 in the connected relationship network 2 is equal to the continuous product of the unit mining paths corresponding to the mining service correlation attributes R5, R10 and R11, the loss ls3 of the mining service correlation attribute R5 in the connected relationship network 3 is equal to the continuous product of the unit mining paths corresponding to the mining service correlation attributes R5, R9, U23 and R11, the loss ls4 of the mining service correlation attribute R5 in the connected relationship network 4 is equal to the continuous product of the unit mining paths corresponding to the mining service correlation attributes R5, R6 and W23, and whether the mining service correlation attribute R5 needs to be hidden is determined according to the losses ls1, ls2, ls3 and ls4, for example, if the sum of the mining service correlation attributes R380, ls1, ls2, ls3 and ls4 are smaller than the hidden service correlation attribute R5.

It can be seen from the above that, in an embodiment of an independent concept, a mining service association attribute is taken as a unit, all connection relationship networks where the mining service association attribute is located are obtained first, where the connection relationship networks may include a plurality of connection relationship networks where the mining service association attribute is located, such as a 3 mining service association attribute mesh, a 4 mining service association attribute mesh, and the like, and whether the mining service association attribute is hidden or not is determined according to a unit mining path corresponding to the mining service association attribute in each connection relationship network in a group of connection relationship networks where the mining service association attribute is located.

And judging whether each mined service association attribute in the third group of mined service association attributes needs to be hidden or not according to the mode, and after the judgment, hiding the redundant mined service association attributes to obtain a second mined path relation network. And updating the global mining path of the big data mining model according to the second mining path relation network.

It should be noted that whether the mining service association attribute needs to be hidden or not can be understood as connected relation network verification, the connected relation network verification is to hide the mining service association attribute with large constraint loss, and further hide the mining service association attribute which does not meet the constraint to obtain a second mining path relation network, and the global mining path of the big data mining model in each mining path sequence is obtained through overall updating according to the second mining path relation network.

In an embodiment of independent concept, determining whether a redundant mined service association attribute exists in the third group of mined service association attributes according to a unit mining path between two mined service units connected by each of the third group of mined service association attributes may include: determining a connection relation network formed by each mining service correlation attribute in the third group of mining service correlation attributes, the first group of mining service correlation attributes and the mining service correlation attributes in the second group of mining service correlation attributes to obtain a first group of connection relation networks; and determining whether the third group of mining service association attributes have redundant mining service association attributes or not according to the unit mining path corresponding to the mining service association attributes in each of the first group of communication relationship networks.

Determining a connectivity network formed by each mining service association attribute in the third set of mining service association attributes, the first set of mining service association attributes, and the mining service association attributes in the second set of mining service association attributes to obtain a first set of connectivity network, which may include: when the preset associated attribute number of the connectivity relationship network comprises N attribute numbers, executing the following steps for each attribute number in the N attribute numbers to obtain a first group of connectivity relationship networks, wherein N is a natural number which is 1 or more than 1, and when the following steps are executed, each attribute number is the current attribute number: and determining a connected relationship network formed by each mining service correlation attribute in the third group of mining service correlation attributes, the first group of mining service correlation attributes and the mining service correlation attributes in the second group of mining service correlation attributes, wherein the number of the mining service correlation attributes included in the formed connected relationship network is the current attribute number.

Determining whether a redundant mining service association attribute exists in the third group of mining service association attributes according to a unit mining path corresponding to the mining service association attribute in each of the first group of communication relationship networks may include: for each mined service association attribute in the third group of mined service association attributes, executing the following steps, wherein each mined service association attribute is a current mined service association attribute when the following steps are executed: determining a second group of communication relation networks including the current mining service association attribute in the first group of communication relation networks; determining redundancy measurement parameters of the current mining service association attribute according to a unit mining path corresponding to the mining service association attribute in each of the second group of connection relationship networks; and when the redundancy measurement parameter of the current mining service associated attribute does not meet the preset condition, determining the current mining service associated attribute as the redundant mining service associated attribute.

In an embodiment of the independent concept, redundancy measure parameters in the connectivity network 1, the connectivity network 2, and the connectivity network 3 are determined.

In an embodiment of independent concept, determining a redundancy measurement parameter of a current mining service association attribute according to a unit mining path corresponding to a mining service association attribute in each of a second group of connectivity networks may include:

for each connectivity network of the second set of connectivity networks, performing the following steps, wherein each connectivity network is a current connectivity network when performing the following steps: converting the unit excavation path corresponding to each excavation service association attribute in the current connectivity relationship network into excavation path meshes to obtain a group of excavation path meshes, wherein parameters in the excavation path meshes are used for representing excavation service nodes and excavation dimensions in the unit excavation paths; determining redundancy measurement parameters corresponding to the current connectivity network according to a group of excavation path grids; and determining the redundancy measurement parameters of the currently mined service association attributes according to the redundancy measurement parameters corresponding to each communication relation network in the second group of communication relation networks.

It should be noted that, determining the redundancy measurement parameter corresponding to the current connectivity network according to a group of mining path grids may include: performing primary fusion on a group of excavation path grids according to a preset sequence to obtain a target grid, wherein parameters in the target grid are used for representing excavation service nodes and excavation dimensions in a target unit excavation path, and the target unit excavation path is an accumulated unit excavation path formed by unit excavation paths corresponding to each excavation service correlation attribute in the current communication relation network; converting parameters in the target grid into target redundancy variables; and determining a redundancy measurement parameter corresponding to the current communication relation network according to the target redundancy variable.

In the embodiment of an independent concept, a loss value of the edge is obtained by calculation according to the unit excavation path of each excavation service correlation attribute in the communication relation network, the calculation process is that the phase excavation path is converted into an excavation path grid, the excavation path grids of each unit excavation path are multiplied to obtain a target grid, and parameters in the target grid are converted into target redundancy variables; and determining a redundancy measurement parameter corresponding to the current communication relation network according to the target redundancy variable.

In an embodiment of independent concept, determining a redundancy measurement parameter of a currently mined service association attribute according to a redundancy measurement parameter corresponding to each of a second group of connectivity networks may include: and determining the accumulated value of the redundancy measurement parameters corresponding to each communication relation network in the second group of communication relation networks as the parameters of the redundancy measurement parameters of the current mining service correlation attributes.

When the redundancy measurement parameter of the current mining service association attribute does not meet the preset condition, determining the current mining service association attribute as the redundant mining service association attribute may include: and when the parameter of the redundancy measurement parameter of the current mining service associated attribute is less than 0, determining the current mining service associated attribute as the redundant mining service associated attribute.

In an embodiment of independent concept, updating a first group of global mining paths in a first big data mining path sequence and a second group of global mining paths in a second big data mining path sequence according to a second mining path relationship network to obtain a third mining path sequence and a fourth mining path sequence may include: and updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence according to a first limiting strategy to obtain a third mining path sequence and a fourth mining path sequence, wherein the first limiting strategy is a limiting strategy determined according to unit mining paths corresponding to a fourth group of mining service correlation attributes in the second mining path relationship network, and the target limiting strategy can comprise the first limiting strategy.

Wherein, according to the first constraint policy, updating the first group of global mining paths in the first big data mining path sequence and the second group of global mining paths in the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence, which may include: and updating the global excavation path in the first big data excavation path sequence and the global excavation path in the second big data excavation path sequence, so that the sum of the loss between the unit excavation path obtained by each excavation service correlation attribute in the second excavation path relation network and the unit excavation path recalculated by each excavation service correlation attribute is the minimum, wherein the unit excavation path recalculated by each excavation service correlation attribute is the unit excavation path calculated by the global excavation path on the excavation service unit communicated by each excavation service correlation attribute.

In an embodiment of independent concept, updating a first group of global mining paths in a first big data mining path sequence and a second group of global mining paths in a second big data mining path sequence according to a second mining path relationship network to obtain a third mining path sequence and a fourth mining path sequence may include:

updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence according to a second limiting strategy to obtain a third mining path sequence and a fourth mining path sequence, the second restriction strategy is determined according to a unit mining path corresponding to a fourth group of mining service association attributes in the second mining path relationship network and a first group of prior mining paths, the fourth group of mining service association attributes comprise mining service association attributes except redundant mining service association attributes in the third group of mining service association attributes, each prior mining path in the first group of prior mining paths is a prior mining path of a mining service unit in the first cluster mining service unit, and the target restriction strategy comprises the second restriction strategy.

Wherein, according to a second constraint policy, updating a first group of global mining paths in the first big data mining path sequence and a second group of global mining paths in the second big data mining path sequence to obtain a third mining path sequence and a fourth mining path sequence, and may further include: and updating the global excavation path in the first big data excavation path sequence and the global excavation path in the second big data excavation path sequence to minimize the sum of a first global loss and a second global loss, wherein the first global loss is the sum of losses between the unit excavation path obtained by each excavation service associated attribute in the second excavation path relationship network and the unit excavation path recalculated by each excavation service associated attribute, each unit excavation path recalculated by each excavation service associated attribute is a unit excavation path calculated by the global excavation path on the excavation service unit communicated with each excavation service associated attribute, and the second global loss is the sum of losses between each prior excavation path in the first group of prior excavation paths and the global excavation path on the corresponding excavation service unit.

In an independently contemplated embodiment, a prior mining path may be determined according to a mining path mined by a first big data mining model, and a global mining path on a mining service unit corresponding to the first big data mining model is determined in an auxiliary manner.

Fig. 3 illustrates a hardware structure of the cloud computing system 100 for implementing the artificial intelligence and user portrait based information pushing method, as shown in fig. 3, the cloud computing system 100 may include a processor 110, a machine-readable storage medium 120, a bus 130, and a communication unit 140.

In a specific implementation process, at least one processor 110 executes computer-executable instructions stored in the machine-readable storage medium 120, so that the processor 110 may execute the information pushing method based on artificial intelligence and user representation according to the above method embodiment, the processor 110, the machine-readable storage medium 120, and the communication unit 140 are connected through the bus 130, and the processor 110 may be configured to control the transceiving action of the communication unit 140, so as to perform data transceiving with the service using apparatus 200.

For a specific implementation process of the processor 110, reference may be made to the above-mentioned method embodiments executed by the cloud computing system 100, which implement principles and technical effects are similar, and details of this embodiment are not described herein again.

In addition, the embodiment of the disclosure also provides a readable storage medium, in which computer-executable instructions are preset, and when a processor executes the computer-executable instructions, the information push method based on artificial intelligence and user portrait is implemented.

Finally, it should be understood that the embodiments described herein are merely illustrative of the principles of the embodiments described herein. Other variations are also possible within the scope of this description. Accordingly, by way of example, and not limitation, alternative configurations of the embodiments of the specification can be seen as matching the teachings of the specification. Accordingly, the embodiments of the present description are not limited to only those embodiments explicitly described and depicted herein.

Claims (10)

1. An information pushing method based on artificial intelligence and user portrait is applied to a cloud computing system, the cloud computing system is in communication connection with a plurality of service using devices, and the method comprises the following steps:

determining a basic user portrait corresponding to target user interest data according to a pre-trained user portrait classification model;

acquiring dynamic expansion data related to the target user interest data, and determining a dynamic expansion portrait corresponding to the dynamic expansion data based on the user portrait classification model;

determining a target user portrait corresponding to a service user according to the basic user portrait and the dynamic extension portrait;

and pushing corresponding personalized service item link information to the service using equipment in real time based on the target user portrait corresponding to the service user.

2. The method for pushing information based on artificial intelligence and user portrait according to claim 1, wherein the step of obtaining dynamic extension data related to the interest data of the target user comprises:

acquiring the extension session behavior data of an extension session flow between the current interest-concerned node of the interest data concerned by the target user and the interest-concerned node of the cooperative interaction of the current interest-concerned node and the interest-concerned node;

determining reference persistence behavior data for the extended session flow based on the extended session behavior data;

screening the interaction behavior instances of the extended session flow within a preset time sequence range through the reference continuous behavior data to obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity within the preset time sequence range;

acquiring the behavior data of the interest ending activity, determining the interest ending activity of which the behavior data accords with a first screening condition as an interest triggering activity, and loading the interest triggering activity to the interest triggering log;

and determining dynamic expansion data of the expansion session flow within the preset time sequence range through the interest trigger log.

3. The artificial intelligence and user representation based information push method of claim 2, wherein the reference apersistence behavior data includes common behavior data of apersistence behavior data and low frequency behavior data of apersistence behavior data;

the step of screening the interaction behavior instances of the extended session flow within a preset time sequence range by using the reference persistence behavior data to obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity within the preset time sequence range includes:

screening the interaction behavior instances of the extended session flow in a preset time sequence range through the common behavior data of the continuous behavior data to obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity in the preset time sequence range;

before determining, by the interest trigger log, dynamic extension data of the extended session flow within the preset time sequence range, the method further includes:

and eliminating the interest triggering activities which do not meet the second screening condition in the interest triggering log through the low-frequency behavior data of the continuous behavior data.

4. The information pushing method based on artificial intelligence and user portrait according to claim 3, wherein the extended session behavior data of the extended session flow includes a first connection flow node where an interactive behavior instance connects the current interest-interested node from the extended session flow and a second connection flow node where the interactive behavior instance connects the collaborative interaction interest-interested node, and the preset timing range includes a first preset timing range between a service delivery start node and a service delivery end node and a second preset timing range between a subsequent service delivery start node and a subsequent service delivery end node after the service delivery end node;

the determining reference persistence behavior data for the extended session flow based on the extended session behavior data comprises:

determining a first activation node of an interactive behavior instance for activating the current interest-concerned node from the extended session flow and a second activation node of the interactive interest-concerned node for activating the collaborative interaction by the interactive behavior instance based on the flow node range of the first connection flow node and the second connection flow node;

summarizing behavior data of all interactive behavior instances of the extended session flow through the second activation node and the first connection flow node;

analyzing the behavior data of all interactive behavior instances of the extended session flow to obtain the low-frequency behavior data of the continuous behavior data and the common behavior data of the continuous behavior data;

the step of screening the interaction behavior instances of the extended session flow within a preset time sequence range through the common behavior data of the persistent behavior data to obtain an interest trigger log covering at least one interest trigger activity and an interest end log covering at least one interest end activity within the preset time sequence range includes:

summarizing a first target interaction behavior example of the first activation node in the first preset time sequence range and a second target interaction behavior example of the first activation node in the second preset time sequence range;

screening a first interest trigger log and a first interest end log from the first target interaction behavior instance through comparison information among the first connection process node, the first activation node, the service delivery starting node, the service delivery ending node, the behavior data and common behavior data of the continuous behavior data;

and screening a second interest trigger log and a second interest end log from the second target interaction behavior instance through comparison information among the first connection process node, the first activation node, the latter service delivery starting node, the latter service delivery ending node, the behavior data and the common behavior data of the continuous behavior data;

the acquiring the behavior data of the interest ending activity, and determining the interest ending activity of which the behavior data meets the first screening condition as the interest triggering activity and loading the interest triggering activity into the interest triggering log comprises the following steps:

acquiring the behavior data of the interest ending activities in the first interest ending log, determining the interest ending activities of which the behavior data meet first screening conditions as interest triggering activities, and loading the interest ending activities into the first interest triggering log;

acquiring the behavior data of the interest ending activities in the second interest ending log, determining the interest ending activities of which the behavior data accord with the first screening condition as interest triggering activities, and loading the interest ending activities into the second interest triggering log;

the removing, by the low-frequency behavior data of the persistent behavior data, the interest trigger activities in the interest trigger log that do not meet the second filtering condition includes:

removing the interest trigger activities which do not accord with the second screening condition in the first interest trigger log through the low-frequency behavior data of the continuous behavior data, and removing the interest trigger activities which do not accord with the second screening condition in the second interest trigger log;

the determining, by the interest trigger log, dynamic extension data of the extended session flow within the preset time sequence range includes:

and determining dynamic expansion data of the expansion session flow in the first preset time sequence range through a first interest trigger log after the removing processing, and determining dynamic expansion data of the expansion session flow in the second preset time sequence range through a second interest trigger log after the removing processing.

5. The information pushing method based on artificial intelligence and user portrait according to claim 4, wherein the analyzing the behavior data of all interactive behavior instances of the extended session flow, and the obtaining the low frequency behavior data of the persistent behavior data and the common behavior data of the persistent behavior data includes:

sequencing and sorting the behavior data of all the interactive behavior instances to obtain a behavior data sequence;

acquiring behavior data in a preset frequency range in the behavior data sequence, and considering the behavior data in the preset frequency range as low-frequency behavior data of the continuous behavior data;

and screening the behavior data within the range of the set behavior tag in the behavior data sequence, and regarding the shared behavior data part of the screened behavior data as the common behavior data of the continuous behavior data.

6. The method of claim 4, wherein the determining a first activation node of an interactive behavior instance to activate the current interest-of-interest node from the extended session flow and a second activation node of the collaborative interactive interest-of-interest node based on a range of flow nodes where the first connection flow node and the second connection flow node are located comprises:

if the first connection process node is in the range of the activated process node, the first connection process node is regarded as the first activated node;

if the first connection process node is in the range of the inactive process node, the service delivery starting node after the range of the inactive process node is finished is regarded as the first active node;

if the second connection process node is in the range of the activation process node, the second connection process node is regarded as the second activation node;

and if the second connection process node is in the range of the inactive process node, the service delivery starting node after the range of the inactive process node is finished is regarded as the second active node.

7. The method of claim 4, wherein the screening out a first interest trigger log and a first interest end log from the first target interaction behavior instance through comparison information between common behavior data of the first connection process node, the first activation node, the service delivery start node, the service delivery end node, the behavior data and the persistent behavior data comprises:

sequencing the first target interaction behavior instances according to the sequence of the first connection process nodes;

judging whether a first interactive behavior example in the sorted first target interactive behavior examples accords with a first preset activity characteristic or not;

if yes, all the first target interaction behavior instances are determined as non-interest triggering activities;

if not, the first target interaction behavior instances after being managed are sequentially analyzed, and the first target interaction behavior instances are loaded to the first interest trigger log or the first interest end log based on the analysis result.

8. The information pushing method based on artificial intelligence and user portrait according to claim 7, wherein the method includes, sequentially parsing the sorted first target interaction behavior instances, and loading the first target interaction behavior instances to the first interest trigger log or the first interest end log based on a parsing result:

default configuring the first end information as a first metric parameter;

the analyzing the sorted first target interaction behavior instances in sequence, and the loading the first target interaction behavior instances to the first interest trigger log or the first interest end log based on the analysis result includes:

if the second activation node of the current first target interactive behavior instance is prior to the service delivery starting node, the current first target interactive behavior instance and the subsequent first target interactive behavior instance are determined as the non-interest triggering activities;

if the behavior engagement degree between the first activation node and the first connection process node of the current first target interaction behavior instance is within the first target behavior engagement degree, and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interaction behavior instance to the first interest trigger log;

if the behavior engagement degree between the first activation node and the first connection process node of the current first target interaction behavior instance is within the first target behavior engagement degree, and the ending metric parameter of the first ending information is not the first metric parameter, loading the current first target interaction behavior instance to the first interest ending log;

if the first connection process node of the current first target interactive behavior instance is smaller than the service delivery starting node and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interactive behavior instance to the first interest trigger log;

if the first connection process node of the current first target interaction behavior instance is smaller than the service launching starting node and the ending measurement parameter of the first ending information is not the first measurement parameter, loading the current first target interaction behavior instance to the first interest ending log;

if the behavior engagement degree between the first activation node of the current first target interaction behavior instance and the first activation node of the previous first target interaction behavior instance conforms to a second preset activity characteristic, the current first target interaction behavior instance and the subsequent first target interaction behavior instance are considered as non-interest triggering activities;

if the behavior engagement degree of the current first target interaction behavior instance is equal to zero and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interaction behavior instance to the first interest trigger log;

if the behavior engagement degree of the current first target interaction behavior instance is equal to zero and the end measurement parameter of the first end information is not the first measurement parameter, loading the current first target interaction behavior instance to the first interest end log;

if the behavior engagement degree of the current first target interaction behavior example is greater than or equal to the first ratio routine of the common behavior data of the continuous behavior data and is the engagement degree, and the ending measurement parameter of the first ending information is the first measurement parameter, loading the current first target interaction behavior example to the first interest trigger log;

if the behavior engagement degree of the current first target interactive behavior instance is greater than or equal to the first comparison routine of the common behavior data of the continuous behavior data and is the engagement degree, and the ending metric parameter of the first ending information is not the first metric parameter, loading the current first target interactive behavior instance to the first interest ending log;

if the behavior engagement degree of the current first target interaction behavior instance is smaller than the second proportion behavior engagement degree of the common behavior data of the continuous behavior data, the current first target interaction behavior instance and the subsequent first target interaction behavior instances are determined as the non-interest triggering activities;

and if quantitative comparison information between the behavior data of the current first target interaction behavior instance and the common behavior data of the persistent behavior data is between the first proportional routine degree and the second proportional behavior engagement degree, setting the first end information as a second metric parameter, and loading the current first target interaction behavior instance into a first interest end log.

9. The artificial intelligence and user representation based information push method of claim 8, wherein the first ratio is a degree of engagement greater than the second ratio is a degree of engagement;

and/or, the first preset activity characteristic comprises any one of:

the behavior connection degree between the first connection process node and the service delivery starting node is greater than a first preset behavior connection degree;

the behavior connection degree between the first connection process node and the service delivery starting node is greater than a second preset behavior connection degree, the behavior connection degree between the first activation node and the first connection process node is less than a third preset behavior connection degree, and the behavior connection degree is less than the first ratio routine of the common behavior data of the continuous behavior data;

the first preset behavior engagement degree is greater than the second preset behavior engagement degree and the third preset behavior engagement degree;

and/or, the second preset activity characteristic comprises any one of:

the behavior connection degree between the first activation node of the current first target interaction behavior instance and the first activation node of the previous first target interaction behavior instance is greater than a fourth preset behavior connection degree, and the first activation node is smaller than the service delivery end node;

the behavior connection degree between a first activation node of a current first target interactive behavior instance and a first activation node of a previous first target interactive behavior instance is greater than a fifth preset behavior connection degree, the first activation node is smaller than the service release end node, the behavior connection degree between the first activation node and the first connection process node is smaller than a sixth preset behavior connection degree, and the behavior connection degree is smaller than the first ratio routine of the common behavior data of the continuous behavior data and is a connection degree;

and the fourth preset behavior engagement degree is greater than the fifth preset behavior engagement degree and the sixth preset behavior engagement degree.

10. A cloud computing system, characterized in that the cloud computing system comprises a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores a computer program, and the computer program is loaded and executed by the processor to implement the artificial intelligence and user representation-based information pushing method according to any one of claims 1 to 9.

Images