CN111915086A - Abnormal user prediction method and equipment - Google Patents
Abnormal user prediction method and equipment Download PDFInfo
- Publication number
- CN111915086A CN111915086A CN202010782777.3A CN202010782777A CN111915086A CN 111915086 A CN111915086 A CN 111915086A CN 202010782777 A CN202010782777 A CN 202010782777A CN 111915086 A CN111915086 A CN 111915086A
- Authority
- CN
- China
- Prior art keywords
- behavior data
- user
- behavior
- content
- data set
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 230000002159 abnormal effect Effects 0.000 title claims abstract description 106
- 238000000034 method Methods 0.000 title claims abstract description 62
- 238000000605 extraction Methods 0.000 claims abstract description 39
- 238000012163 sequencing technique Methods 0.000 claims abstract description 8
- 238000012549 training Methods 0.000 claims description 22
- 238000004590 computer program Methods 0.000 claims description 10
- 238000013528 artificial neural network Methods 0.000 claims description 5
- 230000000306 recurrent effect Effects 0.000 claims description 4
- 230000003542 behavioural effect Effects 0.000 claims 4
- 230000006399 behavior Effects 0.000 description 279
- 230000008569 process Effects 0.000 description 13
- 206010000117 Abnormal behaviour Diseases 0.000 description 11
- 238000010586 diagram Methods 0.000 description 11
- 238000004891 communication Methods 0.000 description 8
- 230000006870 function Effects 0.000 description 8
- 239000013598 vector Substances 0.000 description 6
- 230000015654 memory Effects 0.000 description 5
- 230000003287 optical effect Effects 0.000 description 4
- 238000012545 processing Methods 0.000 description 4
- 238000012552 review Methods 0.000 description 4
- 230000002547 anomalous effect Effects 0.000 description 2
- 230000002457 bidirectional effect Effects 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 239000011159 matrix material Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- 239000004065 semiconductor Substances 0.000 description 2
- 208000037170 Delayed Emergence from Anesthesia Diseases 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000008034 disappearance Effects 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000001902 propagating effect Effects 0.000 description 1
- 230000006403 short-term memory Effects 0.000 description 1
- 230000011273 social behavior Effects 0.000 description 1
- 230000003997 social interaction Effects 0.000 description 1
- 238000013526 transfer learning Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3438—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment monitoring of user actions
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/01—Social networking
Abstract
The embodiment of the application discloses a method and equipment for predicting an abnormal user. One specific implementation of the abnormal user prediction method includes: acquiring a behavior data set of a user and a user associated with the user on social software; performing feature extraction on the behavior data set to generate a behavior feature set; generating a directed graph triple set based on the behavior feature set; sequencing the behavior characteristics of the users with the directed graph triple in the behavior characteristic set according to a time sequence to generate time sequence characteristics; and inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user. The embodiment predicts whether the user is an abnormal user according to the behavior context of the user and the associated user using the social software, so that the missing rate of the abnormal user is reduced.
Description
Technical Field
The embodiment of the application relates to the technical field of computers, in particular to a method and equipment for predicting abnormal users.
Background
Social, i.e. social interactions. The software that accomplishes this through the network is social software. With the rapid development of the mobile internet, various social software layers are in a large number. The phenomenon of propagating violation information through social software is becoming more and more serious. Currently, abnormal users are detected based on their behavior data on social software. Only users that propagate violation information directly on the social software will be detected as anomalous users. In the case of group violation information propagation, the group members can not be detected because they only participate in assistance and do not directly propagate the violation information.
Disclosure of Invention
The embodiment of the application provides a method and equipment for predicting an abnormal user.
In a first aspect, an embodiment of the present application provides an abnormal user prediction method, including: acquiring a behavior data set of a user and a user associated with the user on social software; performing feature extraction on the behavior data set to generate a behavior feature set; generating a directed graph triple set based on the behavior feature set; sequencing the behavior characteristics of the users with the directed graph triple in the behavior characteristic set according to a time sequence to generate time sequence characteristics; and inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user.
In some embodiments, the behavior data in the behavior data set includes operation class behavior data; and performing feature extraction on the behavior data set, including: and performing Multi-Hot coding on the operation class behavior data in the behavior data set to generate operation class behavior characteristics.
In some embodiments, the behavior data in the behavior data set further comprises content class behavior data, the content class behavior data comprising image content class behavior data; and performing feature extraction on the behavior data set, and further comprising: and performing feature extraction on the image content behavior data in the behavior data set by using an image classification network to generate image content behavior features.
In some embodiments, performing feature extraction on the image content class behavior data in the behavior data set by using an image classification network to generate image content class behavior features includes: and inputting the image content behavior data into an image classification network, and outputting the image content behavior characteristics from a full connection layer of the image classification network.
In some embodiments, performing feature extraction on the image content class behavior data in the behavior data set by using an image classification network to generate image content class behavior features includes: inputting the image content behavior data into an image classification network, and outputting an image content behavior characteristic diagram from a characteristic layer of the image classification network; and expanding the content behavior characteristic graph to generate the image content behavior characteristic.
In some embodiments, the image classification network is a ResNet or VGG model.
In some embodiments, the content class behavior data further comprises textual content class behavior data; and performing feature extraction on the behavior data set, and further comprising: and inputting the text content behavior data in the behavior data set to a pre-training model, and outputting the text content behavior characteristics.
In some embodiments, the pre-training model is a BERT semantic network.
In some embodiments, the content class behavior data further comprises voice content class behavior data; and performing feature extraction on the behavior data set, and further comprising: performing voice recognition on voice content behavior data in the behavior data set to generate character content behavior data; and performing text coding on the character content behavior data to generate voice content behavior characteristics.
In some embodiments, generating a set of directed graph triples based on the set of behavior features includes: and generating a directed graph triple set based on the entity incidence relation of the operation type behavior data in the behavior data set, wherein the directed graph triple in the directed graph triple set comprises the entity and the incidence relation of the operation type behavior data.
In some embodiments, the abnormal user prediction model is trained by: acquiring a positive sample behavior data set of the abnormal user and the associated user of the abnormal user on the social software, and a negative sample behavior data set of the normal user and the associated user of the normal user on the social software; respectively generating a positive sample time sequence feature and a negative sample time sequence feature based on the positive sample behavior data set and the negative sample behavior data set; and training the time sequence network by utilizing the positive sample time sequence characteristics and the negative sample time sequence characteristics to obtain an abnormal user prediction model.
In some embodiments, the timing network is a recurrent neural network.
In some embodiments, obtaining a set of behavior data of a user and associated users of the user on social software comprises: identifying that social software is installed on terminal equipment of a user, and acquiring an address list of the user on the social software; acquiring operation data and content data of various operations executed by a user and an associated user in an address book on social software, and generating a behavior data set, wherein the operation data comprises operation data for executing at least one operation of adding friends, establishing a group chat session, inviting to join the group chat session, browsing a social space or replying comment information in the social space, and the content data comprises content data for executing at least one operation of publishing chat messages in the chat session, publishing shared information in the social space and modifying personal data; and generating a directed graph triple set based on the behavior feature set, wherein the generation comprises the following steps: and generating a directed graph triple set based on the behavior characteristics of at least one operation of adding friends, inviting to join a group chat session and replying comment information on the social space.
In a second aspect, an embodiment of the present application provides an abnormal user prediction apparatus, including: the acquisition module is configured to acquire a behavior data set of a user and an associated user of the user on the social software; the extraction module is configured to perform feature extraction on the behavior data set to generate a behavior feature set; a generating module configured to generate a set of directed graph triples based on the set of behavior features; the sequencing module is configured to sequence the behavior characteristics of the users with the directed graph triples in the behavior characteristic set according to a time sequence to generate time sequence characteristics; and the prediction module is configured to input the time sequence characteristics into a pre-trained abnormal user prediction model and predict whether the user is an abnormal user.
In some embodiments, the behavior data in the behavior data set includes operation class behavior data; and the extraction module comprises: and the coding submodule is configured to perform Multi-Hot coding on the operation class behavior data in the behavior data set to generate operation class behavior characteristics.
In some embodiments, the behavior data in the behavior data set further comprises content class behavior data, the content class behavior data comprising image content class behavior data; and the extraction module further comprises: and the extraction sub-module is configured to perform feature extraction on the image content behavior data in the behavior data set by using an image classification network to generate image content behavior features.
In some embodiments, the extraction sub-module comprises: and the first output unit is configured to input the image content behavior data into the image classification network and output the image content behavior characteristics from the full connection layer of the image classification network.
In some embodiments, the extraction sub-module comprises: a second output unit configured to input the image content class behavior data to the image classification network, and output an image content class behavior feature map from a feature layer of the image classification network; and the expansion unit is configured to expand the content behavior feature graph to generate the image content behavior feature.
In some embodiments, the image classification network is a ResNet or VGG model.
In some embodiments, the content class behavior data further comprises textual content class behavior data; and the extraction module further comprises: and the output sub-module is configured to input the text content behavior data in the behavior data set to the pre-training model and output the text content behavior characteristics.
In some embodiments, the pre-training model is a BERT semantic network.
In some embodiments, the content class behavior data further comprises voice content class behavior data; and the extraction module further comprises: the recognition submodule is configured to perform voice recognition on voice content behavior data in the behavior data set to generate character content behavior data; and the coding submodule is configured to perform text coding on the character content behavior data to generate voice content behavior characteristics.
In some embodiments, the generation module is further configured to: and generating a directed graph triple set based on the entity incidence relation of the operation type behavior data in the behavior data set, wherein the directed graph triple in the directed graph triple set comprises the entity and the incidence relation of the operation type behavior data.
In some embodiments, the abnormal user prediction model is trained by: acquiring a positive sample behavior data set of the abnormal user and the associated user of the abnormal user on the social software, and a negative sample behavior data set of the normal user and the associated user of the normal user on the social software; respectively generating a positive sample time sequence feature and a negative sample time sequence feature based on the positive sample behavior data set and the negative sample behavior data set; and training the time sequence network by utilizing the positive sample time sequence characteristics and the negative sample time sequence characteristics to obtain an abnormal user prediction model.
In some embodiments, the timing network is a recurrent neural network.
In some embodiments, the acquisition module is further configured to: identifying that social software is installed on terminal equipment of a user, and acquiring an address list of the user on the social software; acquiring operation data and content data of various operations executed by a user and an associated user in an address book on social software, and generating a behavior data set, wherein the operation data comprises operation data for executing at least one operation of adding friends, establishing a group chat session, inviting to join the group chat session, browsing a social space or replying comment information in the social space, and the content data comprises content data for executing at least one operation of publishing chat messages in the chat session, publishing shared information in the social space and modifying personal data; and the generation module is further configured to: and generating a directed graph triple set based on the behavior characteristics of at least one operation of adding friends, inviting to join a group chat session and replying comment information on the social space.
In a third aspect, an embodiment of the present application provides a computer device, including: one or more processors; a storage device having one or more programs stored thereon; when the one or more programs are executed by the one or more processors, the one or more processors are caused to implement the method as described in any implementation of the first aspect.
In a fourth aspect, the present application provides a computer-readable medium, on which a computer program is stored, which, when executed by a processor, implements the method as described in any implementation manner of the first aspect.
According to the abnormal user prediction method and the abnormal user prediction equipment, firstly, a behavior data set of a user and a user associated with the user on social software is obtained; then, performing feature extraction on the behavior data set to generate a behavior feature set; then generating a directed graph triple set based on the behavior feature set; then, the behavior characteristics of the users with the directed graph triple in the behavior characteristic set are sequenced according to time sequence, and time sequence characteristics are generated; and finally, inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user. Whether the user is an abnormal user or not is predicted according to the behavior context of the user and the associated user using the social software, so that the missing rate of the abnormal user is reduced.
Drawings
Other features, objects and advantages of the present application will become more apparent upon reading of the following detailed description of non-limiting embodiments thereof, made with reference to the accompanying drawings in which:
FIG. 1 is an exemplary system architecture to which the present application may be applied;
FIG. 2 is a flow diagram of one embodiment of an abnormal user prediction method according to the present application;
FIG. 3 is a flow diagram of yet another embodiment of an abnormal user prediction method according to the present application;
FIG. 4 is a flow diagram of one embodiment of an abnormal user prediction model training method according to the present application;
FIG. 5 is a schematic block diagram of a computer system suitable for use in implementing the computer device of an embodiment of the present application.
Detailed Description
The present application will be described in further detail with reference to the following drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the relevant invention and not restrictive of the invention. It should be noted that, for convenience of description, only the portions related to the related invention are shown in the drawings.
It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict. The present application will be described in detail below with reference to the embodiments with reference to the attached drawings.
FIG. 1 illustrates an exemplary system architecture 100 to which embodiments of the anomalous user prediction methods of the present application may be applied.
As shown in fig. 1, devices 101, 102, 103, 104 and network 105 may be included in system architecture 100. The network 105 is the medium by which communication links are provided between the devices 101, 102, 103, 104. Network 105 may include various connection types, such as wired, wireless communication links, or fiber optic cables, to name a few.
The devices 101, 102, 103, 104 may be hardware devices or software that support network connectivity to provide various network services. When the device is hardware, it can be a variety of electronic devices including, but not limited to, smart phones, tablets, laptop portable computers, desktop computers, servers, and the like. In this case, the hardware device may be implemented as a distributed device group including a plurality of devices, or may be implemented as a single device. When the device is software, the software can be installed in the electronic devices listed above. At this time, as software, it may be implemented as a plurality of software or software modules for providing a distributed service, for example, or as a single software or software module. And is not particularly limited herein.
In practice, a device may provide a respective network service by installing a respective client application or server application. After the device has installed the client application, it may be embodied as a client in network communications. Accordingly, after the server application is installed, it may be embodied as a server in network communications.
As an example, in fig. 1, the devices 101, 102, 103 are embodied as clients and the device 104 is embodied as a server. For example, devices 101, 102, 103 may be clients of social software and device 104 may be a server of social software.
It should be noted that the abnormal user prediction method provided in the embodiment of the present application may be executed by the device 104.
It should be understood that the number of networks and devices in fig. 1 is merely illustrative. There may be any number of networks and devices, as desired for implementation.
With continued reference to FIG. 2, a flow 200 of one embodiment of an abnormal user prediction method according to the present application is shown. The abnormal user prediction method comprises the following steps:
In this embodiment, an executing subject (e.g., the device 104 shown in fig. 1) of the abnormal user prediction method may obtain a behavior data set of the user and its associated user on the social software.
In practice, the user and its associated users are users who are registered with the social software. The user and the associated user have a direct or indirect association relationship on the social software. For example, there may be a friend relationship on social software or a direct association between users belonging to the same group chat session. There are indirect associations between users who have common friends on social software.
The behavior data set may include behavior data generated by a user and its associated user performing various behaviors on the social software. The behavior data may be operation data and content data that perform various operations on the social software. The operation data may be used to record operation processes for performing operations on the social software, such as an operation process for adding a friend, an operation process for establishing a group chat session, an operation process for inviting to join a group chat session, an operation process for browsing a social space, an operation process for publishing a chat message in a chat session, an operation process for publishing shared information in a social space, and the like. In addition, the operational data may also include statistical-type data, such as how often a user logs into the social software, how often a user sends chat messages, how often a user establishes a group chat session, how often a user refreshes social space, and so forth. The content data may be used to record content resulting from operations performed on the social software, such as chat messages posted in a chat session, shared information posted on a social space, and modified profiles, among others.
Typically, the executing body is a background server of the social software, and behavior data of all users registered with the social software is stored on the background server. And the behavior data of different users correspond to different user identifications. Therefore, the execution subject can locally search the behavior data set of the user and the associated user on the social software. In addition, the terminal device of the user who registers the social software can also store the behavior data of the user who registers the social software. Therefore, the executing body may also obtain behavior data of the user from a terminal device of the user (e.g., device 101 shown in fig. 1), and obtain behavior data of its associated user from a terminal device of its associated user (e.g., devices 102 and 103 shown in fig. 1), so as to obtain a behavior data set.
In a practical application scenario, the behavior data included in the behavior data set is generally generated within a recent period of time, for example, behavior data generated by a user and its associated user performing various behaviors on the social software within three days.
In this embodiment, for behavior data in the behavior data set, the execution subject may perform feature extraction on the behavior data to generate corresponding behavior features. The behavior features can be used to describe characteristics of behaviors in the behavior data, and are usually expressed in a vector form.
And step 203, generating a directed graph triple set based on the behavior feature set.
In this embodiment, for a behavior feature in the behavior feature set, the execution subject may generate a corresponding triple of the directed graph based on the behavior feature.
It should be noted that the behavior characteristics capable of describing two entities and the association relationship between the two entities can generate the corresponding triple of the directed graph. The directed graph triple comprises two entities and an incidence relation between the two entities. The two entities are a subject and an object, respectively, and the direction is from the subject to the object. For example, the user a adds the behavior that the user B is a friend, the corresponding triple is (user a, add friend, user B), and the direction is pointed to the entity "user B" by the entity "user a". As another example, user A modifies the career in the profile to the engineer's behavior, which corresponds to the triplet (user A, career, engineer) and the direction pointed to by entity "user A" to entity "engineer".
And 204, sequencing the behavior characteristics of the users with the directed graph triple in the behavior characteristic set according to a time sequence to generate time sequence characteristics.
In this embodiment, the execution subject may select behavior features of users having a triple of the directed graph from the behavior feature set, and perform sorting and combining on the selected behavior features according to a time sequence, so as to obtain a time sequence feature. For example, the behavior features of users with the presence of the triple of the directed graph are sorted from front to back in the order of time from morning to evening.
The users having the directed graph triples may be users corresponding to entities in the directed graph triples. For example, for a triple of the directed graph (user a, add a friend, user B), the users who exist the triple of the directed graph are user a and user B, so all the behavior characteristics of user a and user B are selected from the behavior characteristic set. For another example, for a triple of a directed graph (user a, profession, engineer), the user who exists the triple of the directed graph is user a, so all the behavior features of user a are selected from the behavior feature set.
In this embodiment, the execution subject may input the timing characteristics to an abnormal user prediction model trained in advance, and predict whether the user is an abnormal user.
Wherein, the abnormal user prediction model can be used for predicting whether the user is an abnormal user. In general, when analyzing the behavior features in the time sequence features, the abnormal user prediction model may analyze the behavior features in combination with the context behavior features thereof, and output a probability that the user is an abnormal user.
The abnormal user can be a user who performs the abnormal behavior on the social software, and the abnormal user comprises a user who directly performs the abnormal behavior and a user who indirectly performs the abnormal behavior. The directly performed abnormal behavior may be, for example, a user who directly posts violation information (such as yellow-related information, black-related information, fraud information, etc.) on the social software. The indirectly performed abnormal behavior may be that an associated user of the abnormal user directly performs the abnormal behavior on the social software, for example, the associated user with close relationship directly posts violation information on the social software. In this case, the associated user of the abnormal user is also the abnormal user.
For example, user a establishes a group chat session on social software and publishes a two-dimensional code of the group chat session on a social space; then, the user A adds the user B as a friend, and the user B is added into the group chat session through the two-dimensional code of the group chat session issued by the user A; then user C joins the group chat session by the invitation of user B; finally, user C publishes fraud information in the group chat session. The user C is a user who directly executes an abnormal behavior, and is an abnormal user. And the user A and the user B are users who indirectly execute abnormal behaviors, and may also be abnormal users.
In addition, in the case where the abnormal user prediction model predicts that the user is an abnormal user, it indicates that the user has a high possibility of being an abnormal user, and cannot be directly determined as the abnormal user. In order to reduce the false detection rate, the abnormal user prediction model can also mark abnormal behavior characteristics in the time sequence characteristics and send the context related information to manual review, so that the review efficiency can be improved. The manner in which the abnormal behavior feature is marked may be, for example, a highlight mark.
For convenience of understanding, the following provides a practical application scenario in which the abnormal user prediction method according to the embodiment of the present application may be implemented, specifically as follows:
firstly, installing social software on terminal equipment of a user, acquiring an address book of the user on the social software, acquiring operation data and content data of various operations executed on the social software by the user and associated users on the address book, and generating a behavior data set.
The operation data may include operation data for performing at least one of adding a friend, establishing a group chat session, inviting to join a group chat session, browsing a social space, or replying to comment information on the social space. The content data may include content data that performs at least one of publishing chat messages in a chat session, publishing shared information on a social space, and modifying a profile.
And then, performing feature extraction on the behavior data set to generate a behavior feature set, and generating a directed graph triple set based on the behavior features of at least one operation of adding friends, inviting to join a group chat session and replying comment information on a social space.
And then, sequencing the behavior characteristics of the users with the triple of the directed graph according to a time sequence to generate time sequence characteristics.
And finally, inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user. And when the abnormal user prediction model predicts that the user is the abnormal user, sending the highlight mark to manual review, prompting high risk and sending the context related information to review. The abnormal user prediction method provided by the embodiment of the application comprises the steps of firstly, acquiring a behavior data set of a user and a user associated with the user on social software; then, performing feature extraction on the behavior data set to generate a behavior feature set; then generating a directed graph triple set based on the behavior feature set; then, the behavior characteristics of the users with the directed graph triple in the behavior characteristic set are sequenced according to time sequence, and time sequence characteristics are generated; and finally, inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user. Whether the user is an abnormal user or not is predicted according to the behavior context of the user and the associated user using the social software, so that the missing rate of the abnormal user is reduced.
With further reference to fig. 3, illustrated is a flow 300 that is yet another embodiment of an abnormal user prediction method according to the present application. The abnormal user prediction method comprises the following steps:
In this embodiment, an executing subject (e.g., the device 104 shown in fig. 1) of the abnormal user prediction method may obtain a behavior data set of the user and its associated user on the social software.
Here, the behavior data set may include behavior data generated by a user and its associated users performing various behaviors on the social software. Due to different recorded information, the behavior data can be divided into operation type behavior data and content type behavior data. The operation class behavior data can be used for recording an operation process of executing the operation on the social software. The content class behavior data may be used to record content resulting from performing operations on the social software. The content type behavior data records can be divided into image content type behavior data, text content type behavior data and voice content type behavior data due to different content formats. Different types of behavior data adopt different feature extraction modes. Specifically, if the operation type behavior data exists in the behavior data set, the execution subject may perform step 302 a; if the behavior data set includes image content behavior data, the executing entity may execute step 302 b; if the behavior data set contains the text content type behavior data, the executing entity may execute step 302 c; if there is voice content type behavior data in the behavior data set, the executing entity may execute step 302 b.
In this embodiment, if there is operation class behavior data in the behavior data set, the execution main body may perform Multi-Hot encoding on the operation class behavior data to generate corresponding operation class behavior characteristics.
Generally, there are multiple values under the classification characteristics corresponding to the operation class behavior data, so Multi-Hot coding is required. Specifically, for a classification feature corresponding to certain operation class behavior data, each value under the classification feature is converted into a One-Hot coding form to construct a dictionary matrix, a key is a coordinate, and a value is 1 or 0, which indicates whether a value exists or not. For example, the operation-type behavior data is used to describe whether the user uses the pan-and-pan function, whether the user refreshes a friend circle, whether the user sends a chat message, whether the user publishes shared information, and whether the user adds a friend, so each operation needs to be encoded separately. The operation behavior feature obtained by encoding may be, for example, [1,1,0,0,1], which indicates that the user uses a shake-and-shake function, refreshes a friend circle, adds friends, but does not send chat messages, and does not publish shared information.
And step 302b, performing feature extraction on the image content behavior data in the behavior data set by using an image classification network to generate image content behavior features.
In this embodiment, if there is image content behavior data in the behavior data set, the execution subject may perform feature extraction on the image content behavior data by using an image classification network to generate corresponding image content behavior features.
Generally, an image classification network can be used to classify images, and the intermediate data generated in the image classification process is used as the image content class behavior characteristics. In a practical application scenario, the image classification Network may be, for example, a ResNet (Residual Network) or a vgg (visual Geometry group) model. Among other things, ResNet is characterized by ease of optimization and can increase accuracy by adding comparable depth. The inner residual block uses jump connection, and the problem of gradient disappearance caused by depth increase in a deep neural network is relieved. The VGG model has better performance in a plurality of transfer learning tasks, and has the following characteristics: small convolution kernels, small pooling kernels, wider feature maps with deeper layers, full connectivity deconvolution, and so on.
In some embodiments, the execution subject may input the image content class behavior data into an image classification network, and output the image content class behavior features from a fully connected layer of the image classification network. In the image classification process, the full connection layer of the image classification network outputs the classification result of the image. And the classification result is used as the behavior characteristic of the image content class, and the image detail characteristic is lost.
In some embodiments, the executing subject may first input the image content behavior data into the image classification network, and output the image content behavior feature map from the feature layer of the image classification network; and then expanding the content behavior characteristic graph to generate the image content behavior characteristic. In the image classification process, the intermediate feature layer of the image classification network outputs a feature map, and the image detail features are reserved. Usually, the feature map exists in the form of a matrix, where the feature map can be expanded and spliced into a multi-dimensional vector. For example, for 8-by-8 feature maps, a 64-dimensional vector may be expanded.
And step 302c, inputting the text content behavior data in the behavior data set into a pre-training model, and outputting the text content behavior characteristics.
In this embodiment, if the behavior data set includes text content behavior data, the execution subject may input the text content behavior data to a pre-training model, and output the text content behavior feature.
Wherein the pre-training model may learn the knowledge representation of each word in the text according to a context of each word. Since the learned knowledge representation of each word can be combined with its contextual information, the meaning of each word can be better expressed. In a practical application scenario, the pre-trained model may be, for example, a BERT (Bidirectional Encoder Representation based on Transformers) semantic network. BERT is a deep bidirectional representation pre-training model, and can extract semantic information of texts more deeply.
And step 302d, performing voice recognition on the voice content behavior data in the behavior data set to generate character content behavior data.
In this embodiment, if there is speech content behavior data in the behavior data set, the execution main body may perform speech recognition on the speech content behavior data to generate corresponding text content behavior data. The speech recognition is a technology for converting speech into corresponding characters.
In this embodiment, the execution main body may perform text encoding on the text content behavior data to generate corresponding voice content behavior characteristics.
The step 302c may be referred to as a method for text coding of the text content behavior data, which is not described herein again.
And step 304, generating a directed graph triple set based on the entity incidence relation of the operation class behavior data in the behavior data set.
In this embodiment, the execution subject may generate a directed graph triple set based on an entity association relationship of operation class behavior data in the behavior data set. And the directed graph triples in the directed graph triple set can comprise entities and incidence relations of the operation class behavior data.
The operation class behavior data is generally used to describe two entities and the association relationship between the two entities. Therefore, the execution subject may generate the corresponding triple of the directed graph based on the entity association relationship of the operation class behavior data.
And 305, sequencing the behavior characteristics of the users with the directed graph triple in the behavior characteristic set according to a time sequence to generate time sequence characteristics.
In the present embodiment, the specific operations of step 305-306 have been described in detail in step 204-205 in the embodiment shown in fig. 2, and are not described herein again.
As can be seen from fig. 3, compared with the embodiment corresponding to fig. 2, the flow 300 of the abnormal user prediction method in the present embodiment highlights the step of extracting the behavior feature. Therefore, the scheme described in this embodiment adopts different feature extraction modes for different types of behavior data, realizes the targeted extraction of behavior features, and can be applied to various practical application scenarios.
With continued reference to FIG. 4, illustrated is a flow diagram 400 of one embodiment of an abnormal user prediction model training method according to the present application. The abnormal user prediction model training method comprises the following steps:
step 401, acquiring a positive sample behavior data set of the abnormal user and the associated user of the abnormal user on the social software, and a negative sample behavior data set of the normal user and the associated user of the normal user on the social software.
In this embodiment, an executing subject (e.g., the device 104 shown in fig. 1) of the abnormal user prediction model training method may obtain a positive sample behavior data set of the abnormal user and an associated user of the abnormal user on the social software, and a negative sample behavior data set of the normal user and an associated user of the normal user on the social software.
The abnormal user may be an abnormal user determined by other manners, such as an abnormal user determined by manual analysis, an abnormal user predicted by other models, an abnormal user reported by other users, and the like. The abnormal behavior data exists in the behavior data set of the abnormal user and the associated user of the abnormal user on the social software, and is a positive sample behavior data set. Similarly, the behavior data sets of the normal user and the associated user of the normal user on the social software all include normal behavior data and are negative sample behavior data sets.
Step 402, respectively generating a positive sample time sequence feature and a negative sample time sequence feature based on the positive sample behavior data set and the negative sample behavior data set.
In this embodiment, the execution subject may generate the positive sample timing characteristics based on the positive sample behavior data set, and generate the negative sample timing characteristics based on the negative sample behavior data set. The generation methods of the positive sample timing characteristics and the negative sample timing characteristics may refer to the generation methods of the timing characteristics in fig. 2 and fig. 3, and are not described herein again.
And 403, training the time sequence network by using the positive sample time sequence characteristics and the negative sample time sequence characteristics to obtain an abnormal user prediction model.
In this embodiment, the execution subject may train the time-series network by using the positive sample time-series characteristic and the negative sample time-series characteristic, so as to obtain the abnormal user prediction model.
Where a positive exemplar timing feature may mark a positive exemplar label (the label value may be, for example, 1), and a negative exemplar timing feature may mark a negative exemplar label (the label value may be, for example, 0). The execution main body can respectively take the positive sample time sequence characteristics and the negative sample time sequence characteristics as input, respectively take the positive sample labels and the negative sample labels as supervision, and carry out supervised training on the time sequence network, so that the abnormal user prediction model can be obtained. Generally, the number of negative examples is much larger than the number of positive examples, so as to simulate the mathematical distribution that normal social behaviors occupy most in practical application scenarios.
The timing network is used for processing timing characteristics, and various major classes of RNNs (Recurrent Neural Networks) may be used, including but not limited to LSTM (Long Short-Term Memory) or BERT semantic Networks. The behavior feature of each time node in the time sequence feature can be used as a word vector. Since the word vectors of different behavior features are different in length, the length of the longest word vector in all samples is taken as a reference, and the deficiency is supplemented completely (e.g. zero padding).
The abnormal user prediction model training method provided by the embodiment of the application comprises the steps of firstly, acquiring a positive sample behavior data set of an abnormal user and an associated user of the abnormal user on social software, and acquiring a negative sample behavior data set of the normal user and the associated user of the normal user on the social software; then respectively generating a positive sample time sequence feature and a negative sample time sequence feature based on the positive sample behavior data set and the negative sample behavior data set; and finally, training the time sequence network by using the positive sample time sequence characteristics and the negative sample time sequence characteristics to obtain an abnormal user prediction model, thereby training the abnormal user prediction model capable of predicting whether the user is an abnormal user.
Referring now to FIG. 5, a block diagram of a computer system 500 suitable for use in implementing a computing device (e.g., device 104 shown in FIG. 1) of an embodiment of the present application is shown. The computer device shown in fig. 5 is only an example, and should not bring any limitation to the function and the scope of use of the embodiments of the present application.
As shown in fig. 5, the computer system 500 includes a Central Processing Unit (CPU)501 that can perform various appropriate actions and processes according to a program stored in a Read Only Memory (ROM)502 or a program loaded from a storage section 508 into a Random Access Memory (RAM) 503. In the RAM 503, various programs and data necessary for the operation of the system 500 are also stored. The CPU 501, ROM 502, and RAM 503 are connected to each other via a bus 504. An input/output (I/O) interface 505 is also connected to bus 504.
The following components are connected to the I/O interface 505: an input portion 506 including a keyboard, a mouse, and the like; an output portion 507 including a display such as a Cathode Ray Tube (CRT), a Liquid Crystal Display (LCD), and the like, and a speaker; a storage portion 508 including a hard disk and the like; and a communication section 509 including a network interface card such as a LAN card, a modem, or the like. The communication section 509 performs communication processing via a network such as the internet. The driver 510 is also connected to the I/O interface 505 as necessary. A removable medium 511 such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, or the like is mounted on the drive 510 as necessary, so that a computer program read out therefrom is mounted into the storage section 508 as necessary.
In particular, according to an embodiment of the present disclosure, the processes described above with reference to the flowcharts may be implemented as computer software programs. For example, embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method illustrated in the flow chart. In such an embodiment, the computer program may be downloaded and installed from a network through the communication section 509, and/or installed from the removable medium 511. The computer program performs the above-described functions defined in the method of the present application when executed by the Central Processing Unit (CPU) 501.
It should be noted that the computer readable medium described herein can be a computer readable signal medium or a computer readable storage medium or any combination of the two. A computer readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples of the computer readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In the present application, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In this application, however, a computer readable signal medium may include a propagated data signal with computer readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, fiber optic cable, RF, etc., or any suitable combination of the foregoing.
Computer program code for carrying out operations for aspects of the present application may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or electronic device. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider).
The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units described in the embodiments of the present application may be implemented by software or hardware. The described units may also be provided in a processor, and may be described as: a processor includes an acquisition module, an extraction module, a generation module, a ranking module, and a prediction module. Where the names of these elements do not constitute a limitation on the elements themselves in this case, for example, the obtaining module may also be described as an "element that obtains a set of behavior data of the user and associated ones of the users on the social software".
As another aspect, the present application also provides a computer-readable medium, which may be contained in the computer device described in the above embodiments; or may exist separately and not be incorporated into the computer device. The computer readable medium carries one or more programs which, when executed by the computing device, cause the computing device to: acquiring a behavior data set of a user and a user associated with the user on social software; performing feature extraction on the behavior data set to generate a behavior feature set; generating a directed graph triple set based on the behavior feature set; sequencing the behavior characteristics of the users with the directed graph triple in the behavior characteristic set according to a time sequence to generate time sequence characteristics; and inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user.
The above description is only a preferred embodiment of the application and is illustrative of the principles of the technology employed. It will be appreciated by those skilled in the art that the scope of the invention herein disclosed is not limited to the particular combination of features described above, but also encompasses other arrangements formed by any combination of the above features or their equivalents without departing from the spirit of the invention. For example, the above features may be replaced with (but not limited to) features having similar functions disclosed in the present application.
Claims (15)
1. An abnormal user prediction method, comprising:
acquiring a behavior data set of a user and a user associated with the user on social software;
performing feature extraction on the behavior data set to generate a behavior feature set;
generating a directed graph triple set based on the behavior feature set;
sequencing the behavior characteristics of the users with the directed graph triple in the behavior characteristic set according to a time sequence to generate time sequence characteristics;
and inputting the time sequence characteristics into a pre-trained abnormal user prediction model to predict whether the user is an abnormal user.
2. The method of claim 1, wherein the behavioral data in the behavioral data set comprises operation class behavioral data; and
the performing feature extraction on the behavior data set comprises:
and performing Multi-Hot coding on the operation class behavior data in the behavior data set to generate operation class behavior characteristics.
3. The method of claim 2, the behavior data in the set of behavior data further comprising content class behavior data, the content class behavior data comprising image content class behavior data; and
the performing feature extraction on the behavior data set further includes:
and performing feature extraction on the image content behavior data in the behavior data set by using an image classification network to generate image content behavior features.
4. The method of claim 3, wherein the performing feature extraction on the image content class behavior data in the behavior data set by using an image classification network to generate image content class behavior features comprises:
and inputting the image content behavior data into the image classification network, and outputting image content behavior characteristics from a full connection layer of the image classification network.
5. The method of claim 3, wherein the performing feature extraction on the image content class behavior data in the behavior data set by using an image classification network to generate image content class behavior features comprises:
inputting the image content behavior data into the image classification network, and outputting an image content behavior feature map from a feature layer of the image classification network;
and expanding the content behavior characteristic graph to generate image content behavior characteristics.
6. The method of claim 3, wherein the image classification network is a ResNet or VGG model.
7. The method of claim 3, wherein the content class behavior data further comprises textual content class behavior data; and
the performing feature extraction on the behavior data set further includes:
and inputting the text content behavior data in the behavior data set to a pre-training model, and outputting the text content behavior characteristics.
8. The method of claim 7, wherein the pre-training model is a BERT semantic network.
9. The method of claim 7, wherein the content class behavior data further comprises voice content class behavior data; and
the performing feature extraction on the behavior data set further includes:
performing voice recognition on the voice content behavior data in the behavior data set to generate character content behavior data;
and performing text coding on the character content behavior data to generate the voice content behavior characteristics.
10. The method of claim 2, wherein generating a set of directed graph triples based on the set of behavioral features comprises:
and generating the directed graph triple set based on the entity incidence relation of the operation class behavior data in the behavior data set, wherein the directed graph triple in the directed graph triple set comprises the entity and the incidence relation of the operation class behavior data.
11. The method of any of claims 1-10, wherein the abnormal user prediction model is trained by:
acquiring a positive sample behavior data set of an abnormal user and an associated user of the abnormal user on the social software, and a negative sample behavior data set of a normal user and an associated user of the normal user on the social software;
respectively generating a positive sample time sequence feature and a negative sample time sequence feature based on the positive sample behavior data set and the negative sample behavior data set;
and training a time sequence network by using the positive sample time sequence characteristics and the negative sample time sequence characteristics to obtain the abnormal user prediction model.
12. The method of claim 11, wherein the timing network is a recurrent neural network.
13. The method of any one of claims 1-10, wherein the obtaining a set of behavior data of the user and associated ones of the user on social software comprises:
identifying that the social software is installed on the terminal equipment of the user, and acquiring an address list of the user on the social software;
acquiring operation data and content data of the user and associated users in the address book for executing various operations on the social software, and generating the behavior data set, wherein the operation data comprises operation data for executing at least one operation of adding friends, establishing a group chat session, inviting to join the group chat session, browsing a social space or replying comment information in the social space, and the content data comprises content data for executing at least one operation of publishing chat messages in the chat session, publishing shared information in the social space and modifying personal profiles; and
generating a set of directed graph triples based on the set of behavior features, including:
and generating the directed graph triple set based on the behavior characteristics of at least one operation of adding friends, inviting to join a group chat session and replying comment information on a social space.
14. A computer device, comprising:
one or more processors;
a storage device on which one or more programs are stored;
when executed by the one or more processors, cause the one or more processors to implement the method of any one of claims 1-13.
15. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method according to any one of claims 1-13.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010782777.3A CN111915086A (en) | 2020-08-06 | 2020-08-06 | Abnormal user prediction method and equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010782777.3A CN111915086A (en) | 2020-08-06 | 2020-08-06 | Abnormal user prediction method and equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111915086A true CN111915086A (en) | 2020-11-10 |
Family
ID=73286546
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010782777.3A Pending CN111915086A (en) | 2020-08-06 | 2020-08-06 | Abnormal user prediction method and equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111915086A (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112699217A (en) * | 2020-12-29 | 2021-04-23 | 西安九索数据技术股份有限公司 | Behavior abnormal user identification method based on user text data and communication data |
| CN112836874A (en) * | 2021-02-02 | 2021-05-25 | 北京嘀嘀无限科技发展有限公司 | Behavior prediction method, behavior prediction apparatus, electronic device, storage medium, and program product |
| CN113240510A (en) * | 2021-05-28 | 2021-08-10 | 平安科技（深圳）有限公司 | Abnormal user prediction method, device, equipment and storage medium |
| CN113283822A (en) * | 2021-07-23 | 2021-08-20 | 支付宝(杭州)信息技术有限公司 | Feature processing method and device |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103853841A (en) * | 2014-03-19 | 2014-06-11 | 北京邮电大学 | Method for analyzing abnormal behavior of user in social networking site |
| CN109191281A (en) * | 2018-08-21 | 2019-01-11 | 重庆富民银行股份有限公司 | A kind of group's fraud identifying system of knowledge based map |
| CN109818942A (en) * | 2019-01-07 | 2019-05-28 | 微梦创科网络科技（中国）有限公司 | A kind of user account number method for detecting abnormality and device based on temporal aspect |
| CN109829089A (en) * | 2018-12-12 | 2019-05-31 | 中国科学院计算技术研究所 | Social network user method for detecting abnormality and system based on association map |
| CN110136019A (en) * | 2019-04-22 | 2019-08-16 | 哈尔滨英赛克信息技术有限公司 | A kind of social media Anomaly groups user's detection method to be developed based on relationship |
-
2020
- 2020-08-06 CN CN202010782777.3A patent/CN111915086A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103853841A (en) * | 2014-03-19 | 2014-06-11 | 北京邮电大学 | Method for analyzing abnormal behavior of user in social networking site |
| CN109191281A (en) * | 2018-08-21 | 2019-01-11 | 重庆富民银行股份有限公司 | A kind of group's fraud identifying system of knowledge based map |
| CN109829089A (en) * | 2018-12-12 | 2019-05-31 | 中国科学院计算技术研究所 | Social network user method for detecting abnormality and system based on association map |
| CN109818942A (en) * | 2019-01-07 | 2019-05-28 | 微梦创科网络科技（中国）有限公司 | A kind of user account number method for detecting abnormality and device based on temporal aspect |
| CN110136019A (en) * | 2019-04-22 | 2019-08-16 | 哈尔滨英赛克信息技术有限公司 | A kind of social media Anomaly groups user's detection method to be developed based on relationship |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112699217A (en) * | 2020-12-29 | 2021-04-23 | 西安九索数据技术股份有限公司 | Behavior abnormal user identification method based on user text data and communication data |
| CN112699217B (en) * | 2020-12-29 | 2023-04-18 | 西安九索数据技术股份有限公司 | Behavior abnormal user identification method based on user text data and communication data |
| CN112836874A (en) * | 2021-02-02 | 2021-05-25 | 北京嘀嘀无限科技发展有限公司 | Behavior prediction method, behavior prediction apparatus, electronic device, storage medium, and program product |
| CN113240510A (en) * | 2021-05-28 | 2021-08-10 | 平安科技（深圳）有限公司 | Abnormal user prediction method, device, equipment and storage medium |
| CN113240510B (en) * | 2021-05-28 | 2023-08-01 | 平安科技（深圳）有限公司 | Abnormal user prediction method, device, equipment and storage medium |
| CN113283822A (en) * | 2021-07-23 | 2021-08-20 | 支付宝(杭州)信息技术有限公司 | Feature processing method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11620321B2 (en) | Artificial intelligence based method and apparatus for processing information | |
| CN111274815B (en) | Method and device for mining entity focus point in text | |
| CN111915086A (en) | Abnormal user prediction method and equipment | |
| CN109976997B (en) | Test method and device | |
| CN109145828B (en) | Method and apparatus for generating video category detection model | |
| CN112164391A (en) | Statement processing method and device, electronic equipment and storage medium | |
| CN111523640B (en) | Training method and device for neural network model | |
| CN111382228B (en) | Method and device for outputting information | |
| CN111666416B (en) | Method and device for generating semantic matching model | |
| WO2020182123A1 (en) | Method and device for pushing statement | |
| CN112650841A (en) | Information processing method and device and electronic equipment | |
| CN110633423A (en) | Target account identification method, device, equipment and storage medium | |
| Windiatmoko et al. | Developing FB chatbot based on deep learning using RASA framework for university enquiries | |
| CN111897950A (en) | Method and apparatus for generating information | |
| CN112182255A (en) | Method and apparatus for storing media files and for retrieving media files | |
| CN113033707B (en) | Video classification method and device, readable medium and electronic equipment | |
| CN111090740B (en) | Knowledge graph generation method for dialogue system | |
| CN116127925B (en) | Text data enhancement method and device based on destruction processing of text | |
| US20230367972A1 (en) | Method and apparatus for processing model data, electronic device, and computer readable medium | |
| CN111914535B (en) | Word recognition method and device, computer equipment and storage medium | |
| CN112149426B (en) | Reading task processing method and related equipment | |
| CN111199421B (en) | Social relationship-based user recommendation method and device and electronic equipment | |
| CN111767290B (en) | Method and apparatus for updating user portraits | |
| CN112446738A (en) | Advertisement data processing method, device, medium and electronic equipment | |
| CN111723188A (en) | Sentence display method and electronic equipment based on artificial intelligence for question-answering system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| TA01 | Transfer of patent application right | ||
| TA01 | Transfer of patent application right |
Effective date of registration: 20211223 Address after: 200131 Zone E, 9th floor, No.1 Lane 666, zhangheng Road, Pudong New Area pilot Free Trade Zone, Shanghai Applicant after: Shanghai Shangxiang Network Technology Co.,Ltd. Address before: 201306 N2025 room 24, 2 New Town Road, mud town, Pudong New Area, Shanghai Applicant before: SHANGHAI LIANSHANG NETWORK TECHNOLOGY Co.,Ltd. |