CN116028319A - Prediction method and device based on user interaction behavior and storage medium - Google Patents
Prediction method and device based on user interaction behavior and storage medium Download PDFInfo
- Publication number
- CN116028319A CN116028319A CN202211649054.1A CN202211649054A CN116028319A CN 116028319 A CN116028319 A CN 116028319A CN 202211649054 A CN202211649054 A CN 202211649054A CN 116028319 A CN116028319 A CN 116028319A
- Authority
- CN
- China
- Prior art keywords
- interaction behavior
- sequence
- target
- interaction
- item
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses a prediction method and device based on user interaction behavior and a storage medium. Wherein the method comprises the following steps: acquiring historical interaction behavior data of a target object, wherein the historical interaction behavior data comprises: an interaction behavior set and an item set; carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; and analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item. The recommendation method and the recommendation system solve the technical problem that the recommendation system in the related art does not consider the fact that the recommendation result is inaccurate due to the fact that the preference of the user is possibly changed along with the time change, and user experience is affected.
Description
Technical Field
The present application relates to the field of recommendation, and in particular, to a user interaction behavior-based prediction method, apparatus, and storage medium.
Background
With the rapid development of the internet, mobile applications are increasing, and they provide massive services (shopping, video, news, etc.), and various services also include massive contents, so how to allow users to see the contents of interest for the first time is a difficult problem in front of enterprises. With the development of machine learning methods, the development of recommendation systems enables enterprises to meet the needs of users, and existing recommendation systems generally predict interests of users, such as age, preference, etc., of users by using some information related to users. Most of these recommendation systems do not take into account inaccuracy in the recommendation due to the long-term change in preferences of the user.
In view of the above problems, no effective solution has been proposed at present.
Disclosure of Invention
The embodiment of the application provides a prediction method, a device and a storage medium based on user interaction behavior, which at least solve the technical problem that a recommendation system in the related art does not consider the fact that the recommendation result is inaccurate and the user experience is affected because the preference of a user is possibly changed along with the change of time.
According to an aspect of the embodiments of the present application, there is provided a prediction method based on user interaction behavior, including: acquiring historical interaction behavior data of a target object, wherein the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; and analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item.
Optionally, analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a predicted result includes: mapping the target normalized sequence into a first embedded vector based on an embedded layer of the pre-trained language characterization model; marking interaction behaviors and items in the first embedded vector based on the parity position parameters to obtain a second embedded vector, wherein the odd positions are interaction behaviors and the even positions are items; a prediction result is determined based on the second embedded vector.
Optionally, determining the prediction result based on the second embedded vector includes: the coding layer based on the pre-trained language characterization model calculates a second embedded vector to obtain a target vector; and analyzing the target vector based on the output layer of the pre-trained language characterization model to obtain a prediction result.
Optionally, the coding layer includes: n encoders, the target vector is determined by: and inputting the second embedded vector into a first encoder of the encoding layer, wherein the input of each subsequent encoder is the output obtained after the previous layer passes through a point-by-point feedforward mechanism, and finally, deep learning of interactive behavior data is completed by stacking n encoders, so that a target vector is obtained.
Optionally, the output layer includes: a fully-connected layer, a dependency score enhancer, and a Softmax layer; the full-connection layer is a two-layer full-connection neural network, the dependency score enhancer is used for determining the degree of interdependence among all items, the Softmax layer calculates corresponding interaction behaviors and items according to vectors passing through the dependency score enhancer, and the type with the highest corresponding probability is selected as a prediction result.
Optionally, before the data preprocessing is performed on the interaction behavior set and the item set to obtain the target normalized sequence, the method further comprises: and respectively converting the interaction behavior set and the item set into a first initial specification sequence and a second initial specification sequence, wherein the elements in the first initial specification sequence are indexes of the elements in the interaction behavior set, and the elements in the second initial specification sequence are indexes of the elements in the item set.
Optionally, performing data preprocessing on the interaction behavior set and the item set to obtain a target normalized sequence, including: ordering the first initial canonical sequence based on the interaction time order; and correlating each interaction behavior in the ordered first initial canonical sequence with each corresponding item in the second initial canonical sequence to obtain a target canonical sequence.
According to another aspect of the embodiments of the present application, there is also provided a prediction apparatus based on user interaction behavior, including: the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring historical interaction behavior data of a target object, and the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; the processing module is used for carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; the analysis module is used for analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item.
According to another aspect of the embodiments of the present application, there is further provided a non-volatile storage medium, where the storage medium includes a stored program, and when the program runs, the device in which the storage medium is controlled to execute any one of the prediction methods based on the user interaction behavior.
According to another aspect of the embodiments of the present application, there is also provided an electronic device, including: a processor; a memory for storing processor-executable instructions; the processor is configured to execute instructions to implement any of a variety of user interaction behavior-based prediction methods.
In this embodiment of the present application, a manner of predicting a next interaction behavior of a user by modeling different types of interaction behaviors in user history information and learning a correlation between the different interaction behaviors is specifically obtained by obtaining history interaction behavior data of a target object, where the history interaction behavior data includes: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; the target normalized sequence is analyzed based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating items corresponding to interaction behaviors in a target period or the interaction behaviors corresponding to each item, so that the technical effects of accurately predicting the interaction behaviors of a user and the interested items and further improving the retention rate of the user are achieved, and the technical problems that recommendation results are inaccurate and user experience is affected due to the fact that a recommendation system does not consider that the preference of the user is possibly changed along with time change in the related art are solved.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application. In the drawings:
FIG. 1 is a flow chart of an alternative method of predicting based on user interaction behavior according to an embodiment of the present application;
FIG. 2 is a schematic architecture diagram of a prediction method in some embodiments of the present application;
FIG. 3 is a flow chart illustrating an implementation of a prediction method according to an embodiment of the present application;
FIG. 4 is a schematic diagram of a case where there are multiple obscured symbols at the output layer in an embodiment of the present application;
fig. 5 is a schematic structural diagram of an alternative prediction apparatus based on user interaction behavior according to an embodiment of the present application.
Detailed Description
In order to make the present application solution better understood by those skilled in the art, the following description will be made in detail and with reference to the accompanying drawings in the embodiments of the present application, it is apparent that the described embodiments are only some embodiments of the present application, not all embodiments. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, shall fall within the scope of the present application.
It should be noted that the terms "first," "second," and the like in the description and claims of the present application and the above figures are used for distinguishing between similar objects and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used may be interchanged where appropriate such that embodiments of the present application described herein may be implemented in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.
The recommendation method in the related art mainly includes a method based on item similarity, which recommends a most similar item list for a user by calculating similarity between items; there is a collaborative filtering-based method that recommends items of interest to a neighbor user to the user by finding the neighbor user most similar to the target user; there are matrix factorization-based methods that use modeling items in a different matrix than items to predict items of interest to a user. With the development of machine learning methods, there is a recommendation method based on Transformer, BERT, which can effectively learn the user interaction history. Compared with the recommendation method based on Transformer, BERT, the method can model different types of interaction through the designed conversion rule, and can effectively cope with heterogeneous scenes formed by different types of interaction. In addition, by introducing the parity position parameters, the interaction behavior pair parameters and the dependency score enhancer into the BERT to learn the potential semantics in the conversion rules proposed by the application more effectively, the future interaction behavior of the user can be predicted more accurately. Because the invention considers different types of interaction behaviors, the invention can model the user negative behaviors neglected in the prior art, such as negative evaluation and other contents, and can expand the application scene of the method.
According to embodiments of the present application, there is provided an embodiment of a method of predicting user interaction behavior, it being noted that the steps illustrated in the flowcharts of the figures may be performed in a computer system, such as a set of computer executable instructions, and that although a logical order is illustrated in the flowcharts, in some cases the steps illustrated or described may be performed in an order other than that illustrated herein.
Fig. 1 is a prediction method based on user interaction behavior according to an embodiment of the present application, as shown in fig. 1, the method includes the following steps:
step S102, historical interaction behavior data of a target object is obtained, wherein the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set;
step S104, carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence;
and S106, analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item.
In the method, a mode of predicting the next interactive behavior of a user by modeling different types of interactive behaviors in user history information and learning correlations among different interactive behaviors is specifically realized by acquiring history interactive behavior data of a target object, wherein the history interactive behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; the target normalized sequence is analyzed based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating items corresponding to interaction behaviors in a target period or the interaction behaviors corresponding to each item, so that the technical effects of accurately predicting the interaction behaviors of a user and the interested items and further improving the retention rate of the user are achieved, and the technical problems that recommendation results are inaccurate and user experience is affected due to the fact that a recommendation system does not consider that the preference of the user is possibly changed along with time change in the related art are solved.
In some embodiments of the present application, the target normalized sequence is analyzed based on the pre-trained language characterization model to obtain the predicted result, which may be implemented by the following steps: mapping the target normalized sequence into a first embedded vector based on an embedded layer of the pre-trained language characterization model; marking interaction behaviors and items in the first embedded vector based on the parity position parameters to obtain a second embedded vector, wherein the odd positions are interaction behaviors and the even positions are items; a prediction result is determined based on the second embedded vector.
Optionally, determining the prediction result based on the second embedded vector includes: the coding layer based on the pre-trained language characterization model calculates a second embedded vector to obtain a target vector; and analyzing the target vector based on the output layer of the pre-trained language characterization model to obtain a prediction result. It should be noted that the pre-trained language characterization model may be abbreviated as BERT.
It should be noted that the above coding layer includes: n encoders, the target vector may be determined as follows: and inputting the second embedded vector into a first encoder of the encoding layer, wherein the input of each subsequent encoder is the output obtained after the previous layer passes through a point-by-point feedforward mechanism, and finally, deep learning of interactive behavior data is completed by stacking n encoders, so that a target vector is obtained.
FIG. 2 is a schematic architecture diagram of a prediction method in some embodiments of the present application, as shown in FIG. 2, an output layer, including: a fully-connected layer, a dependency score enhancer, and a Softmax layer; the full-connection layer is a two-layer full-connection neural network, the dependency score enhancer is used for determining the degree of interdependence among all items, the Softmax layer calculates corresponding interaction behaviors and items according to vectors passing through the dependency score enhancer, and the type with the highest corresponding probability is selected as a prediction result.
In an optional embodiment of the present application, before performing data preprocessing on an interaction behavior set and a project set to obtain a target normalized sequence, the interaction behavior set and the project set may be respectively converted into a first initial normalized sequence and a second initial normalized sequence, where elements in the first initial normalized sequence are indexes of elements in the interaction behavior set, and elements in the second initial normalized sequence are indexes of elements in the project set.
Specifically, the data preprocessing is performed on the interaction behavior set and the project set to obtain a target normalized sequence, which can be realized in the following manner: ordering the first initial canonical sequence based on the interaction time order; and correlating each interaction behavior in the ordered first initial canonical sequence with each corresponding item in the second initial canonical sequence to obtain a target canonical sequence.
FIG. 3 is a schematic diagram of a prediction method according to an embodiment of the present application, as shown in FIG. 3, the method includes: s1, inputting user interaction behavior; s2, modeling user interaction behavior data; s3, inputting data to the BERT network; s4, predicting the future interaction behavior of the user. Specific:
s1, firstly, inputting interactive behavior data related to a user, and in order to improve the accuracy of prediction of the method, introducing historical interactive behavior data of the user to lay a foundation for subsequent modeling of a user interactive behavior sequence. Specifically, the method initializes the input item set it= { IT 1 ,it 2 ,…,it M [ wherein it is M Is item M, interaction behavior set t= { T 1 ,t 2 ,…,t K }, t is K Is the interaction behavior K. For each user, the data entered is { (t, it) 1 ,(t,it) 2 ,…,(t,it) l I.e., historical interaction behavior data, where T e T, IT e IT.
S2, secondly, for inputting user interaction behavior data, in order to enable the BERT model to learn, a conversion rule of normalized input data is designed to model the interaction behavior data, wherein the rule is as follows:
first, this step sets the user interaction behavior t= { T 1 ,t 2 ,…,t K Normalized to e= { E 1 ,e 2 ,…,e K E, where e K Is t K Index number K of (a), the item set it= { IT 1 ,it 2 ,…,it M Normalized i= { I 1 ,i 2 ,…,i M (i) M Is it M Index number M of (c). (i.e., converting the interaction behavior set and the item set into a first initial canonical sequence and a second initial sequence, respectively)
Secondly, for each user's interaction behavior set, the step sorts all the interaction behaviors according to time, and simultaneously normalizes and expresses the interaction behaviors and associated items as e i And i i Finally, for each user' S interaction behavior set, the sequence S (u) = { e can be obtained after normalization 1 ,i 1 ,…,e l ,i l }. (namely, the first initial canonical sequence is ordered based on the interaction time sequence, and each interaction behavior in the ordered first initial canonical sequence is associated with each corresponding item in the second initial canonical sequence to obtain a target canonical sequence.
S3, obtaining a normalized sequence S (u) = { e of each user 1 ,i 1 ,…,e l ,i l After the above steps, the method refers to the idea of the BERT model, and the interactive behavior sequence of the user is better learned by introducing new parameters into the embedded layer. As shown in fig. 2:
(1) Embedding layer
At input S (u) = { e 1 ,i 1 ,…,e l ,i l After the sequence, the embedding layer maps the sequence to an embedded vector (i.e., the embedding layer maps the target normalized sequence to a first embedded vector based on the pre-trained language characterization model). The method uses position embedding (Positional Embedding) to effect the transition of the sequence into an embedded vector for a given item i l The calculation of the embedded layer is shown in equation 1.
h i ＝v i +P pos (i) (1)
Wherein v is i Is i l The method regards each item as a word, and each item can be obtained by using a word2vec methodEmbedding vector v of individual items i 。
Wherein P is pos (i) Is i l As shown in equation 2, the interaction uses cos functions to calculate odd-numbered position embedded vectors and sin functions to calculate even-numbered position embedded vectors.
For a given interaction behavior e l It is also converted into an embedded vector h using the method described above i (second embedded vector).
The method is to input the interaction behavior of the user and the items in pairs, wherein the odd positions are interaction behaviors and the even positions are items, so that the parity position parameter P is introduced type (i) To mark whether embedded vectors in the sequence are interactive behaviors or items, P type (i) The calculation of (1) is shown in equation 3, where the interaction behavior is assigned a value of 1 and the item is assigned a value of 0.
At the same time, the method has the interactive behavior pair parameter P relevance (i) To represent the same pair of interaction behavior and item (for the same pair of interaction behavior (e l ,i l ),P relevance (i) Are identical). P (P) relevance (i) The calculation of (2) is shown in equation 4.
The two parameters can help the model to better learn the relation between the interaction behavior and the item, (namely, the interaction behavior of the first embedded vector and the item are marked based on the parity position parameter to obtain a second embedded vector).
(2) Coding layer
The coding layer consists of n encoders. After obtaining the embedded partOutput vector h of layer i Then, the method refers to the transformation idea to vector h i Encoding is performed to learn more advanced context. For each layer of encoder, the input is
Output is F after stacking intermediate vectors l . Specifically, each encoder is mainly composed of a multi-head attention mechanism and a point-by-point feed-forward mechanism.
Multi-head attention mechanism: the attention mechanism can capture important information in the sequence, and the multi-head attention mechanism can use a plurality of vector spaces to represent the sequence, so that the multi-head attention mechanism is beneficial to learning more information, and is shown in the following formula 5.
Wherein head is i (H) Is the output of each attention head, W MH ,
Is a weight matrix, < >>
Is the input vector of the encoder.
Multi-head point-by-point feed forward mechanism: because the attention mechanism is linear, a nonlinear function needs to be introduced to improve the learning ability of the model, so the method uses a multi-head point-by-point feed-forward mechanism to improve the learning ability of the model. The computation of the multi-headed point-by-point feed forward mechanism is shown in equation 6.
Wherein the method comprises the steps of
Is the output of each point-by-point feed forward head, geLU is the activation functionNumber W (1) ,W (2) ,b (1) ,b (2) Is a weight matrix and a bias value, H l Is an intermediate vector through the multi-headed attention mechanism.
Specifically, for n encoders, the flow of the encoding layer is as in equation 7.
h i ->H 1 ->F 1 ->H 2 ->F 2 ->...->->H n ->T i (7)
Except that the input of the first encoder is the output vector h of the embedded layer i The input of each encoder is the output F obtained by the previous layer after the point-by-point feedforward mechanism l Deep learning of the user interaction sequence is finally completed by stacking n encoders (the number of encoding layers n is manually specified), and the final layer encoder outputs a final vector T i . (i.e., the second embedded vector is computed by the coding layer based on the pre-trained language characterization model to obtain the target vector).
(3) Output layer
The output layer consists of a fully connected layer, a dependency score enhancer and a Softmax layer.
Wherein the full-connection layer is a two-layer full-connection neural network, and outputs a hidden vector L for each mask token in training i 。
Because part of interaction behavior depends on other interaction behaviors, the method designs a dependency score enhancer for optimizing the accuracy of the model. Specifically, if the token being masked in the training is an item i l The method will check the interaction behavior e with which it forms a pair l If the interaction behavior e l Dependent on other interaction behavior e k The dependency score enhancer will then add the item i l The score of (a) is set to 1 (weight can be set manually) and if not, the score is set to 0.1 (weight can be set manually) so that valuable and insignificant items can be distinguished, the calculation is shown in formula 8, where e l ！-e k Representing interaction behavior e l Independent of other interaction behavior e k ，e l -e k Representing interaction behavior e l Dependent on other interaction behavior e k 。
The Softmax layer is based on vector C to be passed through the dependency score enhancer i Calculating the corresponding interaction behavior and items, and selecting the type with the highest corresponding probability as a prediction result Y of the method for each token to be masked in training i As shown in equation 9.
Y i ＝Softmax(C i )(9)
The method adopts the form of training the mask part token of the original user interaction behavior sequence, so that a plurality of mask-added tokens exist in the last output layer in training, as shown in fig. 4. In actual prediction, the method is given a user interaction behavior sequence { (t, it) 1 ,(t,it) 2 ,…,(t,it) l ' and item it l+1 Or interaction behavior t l+1 Thereby predicting the output item it l+1 Corresponding interaction behavior t l+1 Or interaction behavior t l+1 Corresponding item it l+1 . (analysis of the target vector based on the output layer of the pre-trained language characterization model yields the predicted outcome.).
Fig. 5 is a prediction apparatus based on user interaction behavior according to an embodiment of the present application, as shown in fig. 5, the apparatus includes:
the obtaining module 50 is configured to obtain historical interaction behavior data of the target object, where the historical interaction behavior data includes: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set;
the processing module 52 is configured to normalize the interaction behavior set and the item set to obtain a target normalized sequence;
the analysis module 54 is configured to analyze the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, where the prediction result is used to indicate an item corresponding to the interaction behavior or an interaction behavior corresponding to each item in the target period.
In the prediction apparatus, an obtaining module 50 is configured to obtain historical interaction behavior data of a target object, where the historical interaction behavior data includes: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; the processing module 52 is configured to normalize the interaction behavior set and the item set to obtain a target normalized sequence; the analysis module 54 is configured to analyze the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, where the prediction result is used to indicate an item corresponding to the interaction behavior in the target period, or an interaction behavior corresponding to each item, so as to accurately predict the interaction behavior of the user and the item of interest, thereby improving the technical effect of the retention rate of the user, and further solving the technical problem that the recommendation result is inaccurate and the user experience is affected because the recommendation system does not consider that the preference of the user may change along with the time change in the related art.
According to another aspect of the embodiments of the present application, there is further provided a non-volatile storage medium, where the storage medium includes a stored program, and when the program runs, the device in which the storage medium is controlled to execute any one of the prediction methods based on the user interaction behavior.
According to another aspect of the embodiments of the present application, there is also provided an electronic device, including: a processor; a memory for storing processor-executable instructions; the processor is configured to execute instructions to implement any of a variety of user interaction behavior-based prediction methods.
Specifically, the storage medium is configured to store program instructions for the following functions, and implement the following functions:
acquiring historical interaction behavior data of a target object, wherein the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; and analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item.
Alternatively, in the present embodiment, the storage medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a storage medium would include an electrical connection based on 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 an exemplary embodiment of the present application, there is also provided a computer program product comprising a computer program which, when executed by a processor, implements the user interaction behavior based prediction method of any of the above.
Optionally, the computer program may, when executed by a processor, implement the steps of:
acquiring historical interaction behavior data of a target object, wherein the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set; carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence; and analyzing the target normalized sequence based on the pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item.
There is provided, according to an embodiment of the present application, an electronic device including: at least one processor; and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the user interaction behavior based prediction method of any of the above.
Optionally, the electronic device may further include a transmission device and an input/output device, where the transmission device is connected to the processor, and the input device is connected to the processor.
Various implementations of the systems and techniques described here above may be implemented in digital electronic circuitry, integrated circuit systems, field Programmable Gate Arrays (FPGAs), application Specific Integrated Circuits (ASICs), application Specific Standard Products (ASSPs), systems On Chip (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and/or combinations thereof. These various embodiments may include: implemented in one or more computer programs, the one or more computer programs may be executed and/or interpreted on a programmable system including at least one programmable processor, which may be a special purpose or general-purpose programmable processor, that may receive data and instructions from, and transmit data and instructions to, a storage system, at least one input device, and at least one output device.
Program code for carrying out methods of the present application may be written in any combination of one or more programming languages. These program code may be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus such that the program code, when executed by the processor or controller, causes the functions/operations specified in the flowchart and/or block diagram to be implemented. The program code may execute entirely on the machine, partly on the machine, as a stand-alone software package, partly on the machine and partly on a remote machine or entirely on the remote machine or server.
In the context of this application, a machine-readable medium may be a tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. The machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of a machine-readable storage medium would include an electrical connection based on 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.
To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to a user; and a keyboard and pointing device (e.g., a mouse or trackball) by which a user can provide input to the computer. Other kinds of devices may also be used to provide for interaction with a user; for example, feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form, including acoustic input, speech input, or tactile input.
The systems and techniques described here can be implemented in a computing system that includes a background component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front-end component (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such background, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: local Area Networks (LANs), wide Area Networks (WANs), and the internet.
The computer system may include a client and a server. The client and server are typically remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. The server may be a cloud server, a server of a distributed system, or a server incorporating a blockchain.
The foregoing embodiment numbers of the present application are merely for describing, and do not represent advantages or disadvantages of the embodiments.
In the foregoing embodiments of the present application, the descriptions of the embodiments are emphasized, and for a portion of this disclosure that is not described in detail in this embodiment, reference is made to the related descriptions of other embodiments.
In the several embodiments provided in the present application, it should be understood that the disclosed technology content may be implemented in other manners. The above-described embodiments of the apparatus are merely exemplary, and the division of the units, for example, may be a logic function division, and may be implemented in another manner, for example, a plurality of units or components may be combined or may be integrated into another system, or some features may be omitted, or not performed. Alternatively, the coupling or direct coupling or communication connection shown or discussed with each other may be through some interfaces, units or modules, or may be in electrical or other forms.
The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
In addition, each functional unit in each embodiment of the present application may be integrated in one processing unit, or each unit may exist alone physically, or two or more units may be integrated in one unit. The integrated units may be implemented in hardware or in software functional units.
The integrated units, if implemented in the form of software functional units and sold or used as stand-alone products, may be stored in a computer readable storage medium. Based on such understanding, the technical solution of the present application may be embodied in essence or a part contributing to the prior art or all or part of the technical solution in the form of a software product stored in a storage medium, including several instructions to cause a computer device (which may be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a U-disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a removable hard disk, a magnetic disk, or an optical disk, or other various media capable of storing program codes.
The foregoing is merely a preferred embodiment of the present application and it should be noted that modifications and adaptations to those skilled in the art may be made without departing from the principles of the present application and are intended to be comprehended within the scope of the present application.
Claims (10)
1. A method for predicting user interaction behavior, comprising:
acquiring historical interaction behavior data of a target object, wherein the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set;
normalizing the interactive behavior set and the item set to obtain a target normalized sequence;
and analyzing the target normalized sequence based on a pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior or the interaction behavior corresponding to each item in a target period.
2. The prediction method according to claim 1, wherein analyzing the target normalized sequence based on a pre-trained language characterization model to obtain a prediction result comprises:
mapping the target normalized sequence into a first embedded vector based on an embedded layer of a pre-trained language characterization model;
marking interaction behaviors and items in the first embedded vector based on parity position parameters to obtain a second embedded vector, wherein the odd positions are interaction behaviors and the even positions are items; the prediction result is determined based on the second embedded vector.
3. The prediction method of claim 2, wherein determining the prediction result based on the second embedded vector comprises:
calculating the second embedded vector based on the coding layer of the pre-trained language characterization model to obtain a target vector;
and analyzing the target vector based on an output layer of the pre-trained language characterization model to obtain the prediction result.
4. A prediction method according to claim 3, wherein the coding layer comprises: n encoders, the target vector being determined by:
and inputting the second embedded vector into a first encoder of the coding layer, wherein the input of each subsequent encoder is the output obtained after the previous layer passes through a point-by-point feedforward mechanism, and finally, deep learning of the interactive behavior data is completed by stacking the n encoders, so that the target vector is obtained.
5. A prediction method according to claim 3, wherein the output layer comprises: a fully-connected layer, a dependency score enhancer, and a Softmax layer; the full-connection layer is a two-layer full-connection neural network, the dependency score enhancer is used for determining the degree of interdependence among all items, the Softmax layer calculates corresponding interaction behaviors and items according to vectors passing through the dependency score enhancer, and the type with the highest corresponding probability is selected as the prediction result.
6. The method of claim 1, wherein prior to preprocessing the set of interaction behaviors and the set of items to obtain a target normalized sequence, the method further comprises:
and respectively converting the interaction behavior set and the item set into a first initial specification sequence and a second initial specification sequence, wherein elements in the first initial specification sequence are indexes of elements in the interaction behavior set, and elements in the second initial specification sequence are indexes of elements in the item set.
7. The method of claim 6, wherein the preprocessing of the data for the set of interaction behaviors and the set of items to obtain a target normalized sequence comprises:
ordering the first initial canonical sequence based on interaction time order;
and correlating each interaction behavior in the first initial canonical sequence after sequencing with each corresponding item in the second initial canonical sequence to obtain the target canonical sequence.
8. A user interaction behavior-based prediction apparatus, comprising:
the system comprises an acquisition module, a storage module and a processing module, wherein the acquisition module is used for acquiring historical interaction behavior data of a target object, and the historical interaction behavior data comprises: the interactive behavior data in the interactive behavior set and the items in the item set are in one-to-one correspondence, and the arrangement sequence of the interactive behaviors in the interactive behavior set is the same as the arrangement sequence of the items corresponding to the interactive behaviors in the data item set;
the processing module is used for carrying out standardization processing on the interaction behavior set and the item set to obtain a target standardization sequence;
the analysis module is used for analyzing the target normalized sequence based on a pre-trained language characterization model to obtain a prediction result, wherein the prediction result is used for indicating the item corresponding to the interaction behavior in the target period or the interaction behavior corresponding to each item.
9. A non-volatile storage medium, characterized in that the storage medium comprises a stored program, wherein the program, when run, controls a device in which the storage medium is located to perform the prediction method based on user interaction behavior according to any one of claims 1 to 7.
10. An electronic device, comprising:
a processor;
a memory for storing the processor-executable instructions;
wherein the processor is configured to execute the instructions to implement the user interaction behavior based prediction method of any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211649054.1A CN116028319A (en) | 2022-12-21 | 2022-12-21 | Prediction method and device based on user interaction behavior and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211649054.1A CN116028319A (en) | 2022-12-21 | 2022-12-21 | Prediction method and device based on user interaction behavior and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116028319A true CN116028319A (en) | 2023-04-28 |
Family
ID=86071652
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211649054.1A Pending CN116028319A (en) | 2022-12-21 | 2022-12-21 | Prediction method and device based on user interaction behavior and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116028319A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116911912A (en) * | 2023-09-12 | 2023-10-20 | 深圳须弥云图空间科技有限公司 | Method and device for predicting interaction objects and interaction results |
-
2022
- 2022-12-21 CN CN202211649054.1A patent/CN116028319A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116911912A (en) * | 2023-09-12 | 2023-10-20 | 深圳须弥云图空间科技有限公司 | Method and device for predicting interaction objects and interaction results |
| CN116911912B (en) * | 2023-09-12 | 2024-03-15 | 深圳须弥云图空间科技有限公司 | Method and device for predicting interaction objects and interaction results |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2020211566A1 (en) | Method and device for making recommendation to user, computing apparatus, and storage medium | |
| CN109376222B (en) | Question-answer matching degree calculation method, question-answer automatic matching method and device | |
| US8504361B2 (en) | Deep neural networks and methods for using same | |
| US11461638B2 (en) | Figure captioning system and related methods | |
| CN114971748B (en) | Prediction data generation method, model training method, computer device, and storage medium | |
| CN111275514A (en) | Intelligent purchasing method and system, storage medium and electronic device | |
| CN116028319A (en) | Prediction method and device based on user interaction behavior and storage medium | |
| CN116010586A (en) | Method, device, equipment and storage medium for generating health advice | |
| CN111626827A (en) | Method, device, equipment and medium for recommending articles based on sequence recommendation model | |
| CN117633516A (en) | Multi-mode cynics detection method, device, computer equipment and storage medium | |
| CN116342167B (en) | Intelligent cost measurement method and device based on sequence labeling named entity recognition | |
| JP2023533723A (en) | Evaluate interpretation of search queries | |
| CN115794898B (en) | Financial information recommendation method and device, electronic equipment and storage medium | |
| KR102284440B1 (en) | Method to broker deep learning model transactions perfomed by deep learning model transaction brokerage servers | |
| CN115631008B (en) | Commodity recommendation method, device, equipment and medium | |
| CN109119157A (en) | A kind of prediction technique and system of infant development | |
| CN115249016A (en) | Text processing method, device and equipment and readable storage medium | |
| KR102311108B1 (en) | Method to broker deep learning model transactions perfomed by deep learning model transaction brokerage servers | |
| KR102313181B1 (en) | Deep learning solution providing method performed by deep learning platform providing device that provides deep learning solution platform | |
| CN112862007B (en) | Commodity sequence recommendation method and system based on user interest editing | |
| CN113255891B (en) | Method, neural network model and device for processing event characteristics | |
| KR102628947B1 (en) | System for predicting response data and control method thereof | |
| US11983489B1 (en) | Extractive summary generation by abstractive trained model | |
| CN116628179B (en) | User operation data visualization and man-machine interaction recommendation method | |
| CN116029294B (en) | Term pairing method, device and equipment |
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 |