CN111737576B - Application function personalized recommendation method and device - Google Patents
Application function personalized recommendation method and device Download PDFInfo
- Publication number
- CN111737576B CN111737576B CN202010574579.8A CN202010574579A CN111737576B CN 111737576 B CN111737576 B CN 111737576B CN 202010574579 A CN202010574579 A CN 202010574579A CN 111737576 B CN111737576 B CN 111737576B
- Authority
- CN
- China
- Prior art keywords
- user
- data
- function
- determining
- list
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/90—Details of database functions independent of the retrieved data types
- G06F16/95—Retrieval from the web
- G06F16/953—Querying, e.g. by the use of web search engines
- G06F16/9535—Search customisation based on user profiles and personalisation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Abstract
The invention provides a personalized recommendation method and device for application functions, wherein the method comprises the following steps: acquiring user behavior data, user static data and service system function list data; preprocessing user behavior data, user static data and service system function list data, and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data; determining a user behavior-function matrix according to the preprocessed user behavior data; determining a user similarity recommendation list according to the user behavior-function matrix; determining incoming data according to the preprocessed user behavior data and the preprocessed user static data; determining a prediction result matrix according to the modulus-entering data; determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix; and determining a personalized application function list according to the function list data of the preprocessing service system and the function recommendation list. The invention realizes personalized recommendation of the application function in the service system.
Description
Technical Field
The invention relates to the technical field of computer data processing, in particular to an application function personalized recommendation method and device.
Background
This section is intended to provide a background or context to the embodiments of the invention that are recited in the claims. The description herein is not admitted to be prior art by inclusion in this section.
The service application functions (Web applications) of the service in the current service system used by the bank are numerous, and the service condition of the user presents a long tail effect. The business and developer adopted in the prior art make typesetting modes of business application functions of a business system according to business needs and experiences, so that the arrangement of application function modules is preset, the menu display form is single and fixed, subjective factors are easily introduced when the business personnel make function classification and function menu typesetting, the function classification and typesetting layout deviate from actual use preferences of users, and the service quality of the business system is influenced; all users adopt a preset and fixed unified design, and application function menus seen by all users are identical, so that personalized requirements of different users and different posts on functions cannot be met, and user experience is reduced; the service system is often updated and reformed continuously because of the continuous upgrading and the increasing functions, so that many application function paths are deep, and the functions required by users are difficult to find; meanwhile, familiarity degree of each branch line to the service system is inconsistent, so that partial application functions are only used by a few branches.
The above-mentioned multiple conditions can lead to the user to be difficult to find the business function that oneself needs effectively fast, and some business function routes that many times the user needs are very dark in addition, often need click many times just can reach, even some business function users can not find at all and put in what position, can not satisfy the user and use the convenience demand of business function, have restricted the service quality of business system, reduce user experience and office efficiency.
Therefore, how to provide a new solution to the above technical problem is a technical problem to be solved in the art.
Disclosure of Invention
The embodiment of the invention provides an application function personalized recommendation method, which realizes personalized recommendation of service functions, and comprises the following steps:
acquiring user behavior data, user static data and service system function list data;
preprocessing user behavior data, user static data and service system function list data, and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data;
determining a user behavior-function matrix according to the preprocessed user behavior data;
determining a user similarity recommendation list according to the user behavior-function matrix;
Determining incoming data according to the preprocessed user behavior data and the preprocessed user static data;
determining a prediction result matrix according to the modulus-entering data;
determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix;
and determining a personalized application function list according to the function list data of the preprocessing service system and the function recommendation list.
The embodiment of the invention also provides an application function personalized recommendation device, which comprises:
the data acquisition module is used for acquiring user behavior data, user static data and service system function list data;
the preprocessing module is used for preprocessing the user behavior data, the user static data and the service system function list data and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data;
the user behavior-function matrix determining module is used for determining a user behavior-function matrix according to the preprocessed user behavior data;
the user similarity recommendation list determining module is used for determining a user similarity recommendation list according to the user behavior-function matrix;
the incoming data determining module is used for determining incoming data according to the preprocessed user behavior data and the preprocessed user static data;
The prediction result matrix determining module is used for determining a prediction result matrix according to the modulus-entering data;
the function recommendation list determining module is used for determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix;
and the personalized application function list determining module is used for determining a personalized application function list according to the preprocessing service system function list data and the function recommendation list.
The embodiment of the invention also provides computer equipment, which comprises a memory, a processor and a computer program stored on the memory and capable of running on the processor, wherein the processor realizes the application function personalized recommendation method when executing the computer program.
The embodiment of the invention also provides a computer readable storage medium, which stores a computer program for executing the application function personalized recommendation method.
According to the application function personalized recommendation method and device provided by the embodiment of the invention, the historical behavior information and the user personal information of the user are comprehensively considered, the user behavior data and the user static data are utilized for preprocessing, the current demand degree of the user on the function module is predicted through machine learning, a user similarity recommendation list and a prediction result matrix are constructed, the user similarity recommendation list and the prediction result matrix are fused, the function recommendation list is determined, and finally, the personalized application function list is output by combining the preprocessing service system function list data. The embodiment of the invention can individually recommend the service functions in the service system by utilizing the existing client information, provide daily required functions for users and improve office efficiency; the function which is not found by the user is provided, the potential needs of the user are mined, and the office effectiveness and the service quality and efficiency of the system are further improved.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. In the drawings:
fig. 1 is a schematic diagram of an application function personalized recommendation method according to an embodiment of the present invention.
Fig. 2 is a flowchart of an application function personalized recommendation method according to an embodiment of the present invention.
Fig. 3 is a flowchart of determining a user similarity recommendation list according to an embodiment of the present invention.
Fig. 4 is a schematic diagram of a computer device running the method for implementing personalized recommendation of application functions according to the present invention.
Fig. 5 is a schematic diagram of an application function personalized recommendation device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the embodiments of the present invention will be described in further detail with reference to the accompanying drawings. The exemplary embodiments of the present invention and their descriptions herein are for the purpose of explaining the present invention, but are not to be construed as limiting the invention.
As shown in a schematic diagram of an application function personalized recommendation method in the embodiment of the present invention in fig. 1, the embodiment of the present invention provides an application function personalized recommendation method, which implements personalized recommendation of service functions, and the method includes:
step 101: acquiring user behavior data, user static data and service system function list data;
step 102: preprocessing user behavior data, user static data and service system function list data, and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data;
step 103: determining a user behavior-function matrix according to the preprocessed user behavior data;
step 104: determining a user similarity recommendation list according to the user behavior-function matrix;
step 105: determining incoming data according to the preprocessed user behavior data and the preprocessed user static data;
step 106: determining a prediction result matrix according to the modulus-entering data;
step 107: determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix;
step 108: and determining a personalized application function list according to the function list data of the preprocessing service system and the function recommendation list.
According to the application function personalized recommendation method provided by the embodiment of the invention, the historical behavior information and the user personal information of the user are comprehensively considered, the user behavior data and the user static data are utilized for preprocessing, the current demand level of the user on the function module is predicted through machine learning, a user similarity recommendation list and a prediction result matrix are constructed, the user similarity recommendation list and the prediction result matrix are fused, the function recommendation list is determined, and finally, the personalized application function list is output by combining the preprocessing service system function list data. The embodiment of the invention can individually recommend the service functions in the service system by utilizing the existing client information, provide daily required functions for users and improve office efficiency; the function which is not found by the user is provided, the potential needs of the user are mined, and the office effectiveness and the service quality and efficiency of the system are further improved.
The Long Tail Effect is the English name Long Tail Effect; "head" (head) and "tail" (tail) are two statistical terms. The protruding part in the middle of the normal curve is called a head; the relatively gentle portion of the sides is called the "tail". From a human demand perspective, most of the demand will be concentrated on the head, which in turn may be referred to as popularity, while the demands distributed on the tail are personalized, sporadic, and small-volume demands. The differentiated, small number of demands forms a long "tail" over the demand curve, and the so-called long tail effect is in terms of its number, which adds up all non-popular markets to form a market that is larger than popular. Therefore, in order to enhance the attraction of banks to clients distributed in the long-tail effect, personalized function recommendation is required to be carried out on the clients in the long-tail effect so as to realize personalized recommendation of service functions in a banking system, and better serve for 'mainstream' and 'non-mainstream' clients in the bank, so that daily required functions are provided for users, and the office efficiency is improved; the function which is not found by the user is provided, the potential needs of the user are mined, and the office effectiveness and the service quality and efficiency of the system are further improved.
In order to achieve the above functions, as shown in a flowchart of an application function personalized recommendation method in the embodiment of the present invention in fig. 2, the application function personalized recommendation method provided in the embodiment of the present invention is divided into: data acquisition, data preprocessing, matrix construction, feature engineering, model school and fusion and function list output; in specific implementation, the workflow of the six parts may include:
acquiring user behavior data, user static data and service system function list data; preprocessing user behavior data, user static data and service system function list data, and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data; determining a user behavior-function matrix according to the preprocessed user behavior data; determining a user similarity recommendation list according to the user behavior-function matrix; determining incoming data according to the preprocessed user behavior data and the preprocessed user static data; determining a prediction result matrix according to the modulus-entering data; determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix; and determining a personalized application function list according to the function list data of the preprocessing service system and the function recommendation list.
The foregoing obtaining the user behavior data, the user static data, and the service system function list data may include: collecting user behavior data, user static data and service system function list data from a service system log and a database; the service system function list data comprises post information and a service system function list. The user behavior data refers to historical click behavior of a user in a service system and service function using records, and can reflect the use habit of the user in the service system, and the user static data refers to non-user behavior data, namely data related to the user, which generally cannot change in a short time, such as a user address, user personal information, user contact information, user account type and the like after desensitization processing.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the preprocessing is performed on the user behavior data, the user static data and the service system function list data, and the preprocessing of the user behavior data, the preprocessing of the user static data and the preprocessing of the service system function list data is determined, which comprises the following steps:
Preprocessing user behavior data and user static data, removing data with null values and non-normative values, screening data which do not meet modeling requirements according to set screening rules, and determining preprocessed user behavior data and preprocessed user static data;
preprocessing service system function list data to determine preprocessed service system function list data; wherein, the preprocessing service system function list data comprises: a list of hot functions and a list of new functions.
In the embodiment, in the collected user behavior data and user static data, there may be null values, irregular formats and other situations, and the user behavior data and the user static data need to be removed first, and the null values and the irregular data need to be removed, and then, because modeling data need to be high quality data during modeling, a data screening rule needs to be set to meet the modeling requirement, data which does not meet the modeling requirement need to be screened according to the set screening rule, and the preprocessed user behavior data and the preprocessed user static data are determined. In the collected service system function list data, service functions are not orderly arranged, so that post information and service system function lists are required to be classified according to the condition of unified post user classification, and a hot function list and a new function list under the same post user classification are counted to form preprocessing service system function list data. By preprocessing the service system function list data, the new functions of the user service system can be obviously displayed, the new functions are better promoted, and more users try to use the service system; meanwhile, the hot function list can further improve the working efficiency of the service system, so that a user can directly click the hot function on the main page, the searching process is reduced, and the working efficiency is improved.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the determining a user behavior-function matrix according to the preprocessed user behavior data includes:
performing time attenuation on the preprocessed user behavior data by using a time attenuation model, and determining the preprocessed user behavior data after the time attenuation;
preprocessing user behavior data according to time attenuation, and constructing a user behavior-function matrix.
The current behavior of the user can better reflect the current demands of the user, so that the time attenuation model is utilized to process the user behavior data, namely, the influence of the historical behavior data which is closer to the current time on the model is larger, and conversely, the influence is smaller; then constructing a user behavior-function matrix according to the user behavior data after time attenuation;
when the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the user behavior data is preprocessed after time attenuation is determined according to the following mode:
wherein f (x) ij ) Preprocessing user behavior data after time decay; t is a click behavior valued function; t is t 0 The time for the first click action to occur; k is an attenuation factor; t is the current time; t is t ij The time when the j function module was last clicked by the i user.
The foregoing expression of preprocessing the user behavior data after the determined time decay is taken as an example, and it will be understood by those skilled in the art that the foregoing expression may be modified in some manner and other parameters or data may be added according to the need, or other specific formulas may be provided, and these modifications shall fall within the protection scope of the present invention.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, a user behavior-function matrix is constructed according to the following mode:
wherein C is u The function matrix is an m multiplied by n matrix, and each row represents the function module clicked by the user i; v i Representing the ith user; s is(s) i Representing a j-th functional module; y is ij Indicating that the ith user is at time t ij Click on the firstj functional modules; s is(s) ij Representing the sum Σf (x) of the decay of the user click module j ij ) The method comprises the steps of carrying out a first treatment on the surface of the M represents a functional module.
The foregoing expressions for constructing the user behavior-function matrix are given by way of example, and it will be understood by those skilled in the art that the foregoing formulas may be modified and added with other parameters or data in a certain manner or other specific formulas may be provided as needed in the practice of the present invention, and such modifications are intended to fall within the scope of the present invention.
Fig. 3 is a flowchart of determining a user similarity recommendation list according to an application function personalized recommendation method according to an embodiment of the present invention, and as shown in fig. 3, when the application function personalized recommendation method according to the embodiment of the present invention is implemented, in one embodiment, the determining a user similarity recommendation list according to a user behavior-function matrix includes:
step 301: calculating the similarity between every two users according to the user behavior-function matrix through a Pearson correlation coefficient formula, and determining a user similarity matrix;
step 302: ordering the user similarity matrixes according to the similarity from big to small, and determining the arrangement of the user similarity matrixes;
step 303: and selecting the user similarity matrix from the maximum value in the user similarity matrix arrangement according to the set selection quantity, calculating the user function click condition of the selected user similarity matrix, and determining a user similarity recommendation list.
The pearson correlation coefficient is a method which is more complex than Euclidean distance in the prior art and can judge the interest similarity; it will tend to give better results when the data is not very canonical.
Assuming there are two variables X, Y, the user similarity between the two variables can be calculated by the following pearson correlation coefficient formula:
Wherein ρ is X,Y User similarity between variables X, Y; cov is covariance; e is a mathematical expectation; sigma (sigma) X Standard deviation of the X variable; sigma (sigma) Y Standard deviation of Y variable; mu (mu) X Is the mean value of the X variable; mu (mu) Y Is the mean of the Y variables.
The above-mentioned expression of the pearson correlation coefficient formula is given as an example, and those skilled in the art will understand that the above formula may be modified and added with other parameters or data in a certain form or provided with other specific formulas according to the need, and these modifications shall fall within the protection scope of the present invention.
In the embodiment, according to the user behavior-function matrix, the similarity between every two users can be calculated through the pearson correlation coefficient formula, and then the user similarity matrix is determined; further sequencing the user similarity matrixes according to the similarity from large to small, and determining the arrangement of the user similarity matrixes; and finally, selecting the user similarity matrix from the maximum value in the user similarity matrix arrangement according to the set selection quantity, calculating the user function click condition of the selected user similarity matrix, and determining a user similarity recommendation list. In one example, when the set number of choices is N (N is a positive integer), TOP-N user similarity matrices are taken from the maximum value in the user similarity matrix arrangement, and then the user function click condition of the selected TOP-N user similarity matrices is calculated to determine the user similarity recommendation list.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the determining the input data according to the preprocessed user behavior data and the preprocessed user static data includes:
and carrying out characteristic derivation and characteristic engineering by combining business knowledge and engineering experience according to the preprocessed user behavior data and the preprocessed user static data, and determining the input data.
In an embodiment, the foregoing feature engineering, also known as variable selection, attribute selection or variable subset selection in machine learning or statistics, is a process of selecting related features and constructing feature subsets in model construction, and is essentially an engineering activity with the objective of extracting features from raw data to the maximum extent for use by algorithms and models. Collaborative filtering recommendation (Collaborative Filtering recommendation) is further adopted, and is rapidly becoming a popular technology in information filtering and information systems. Collaborative filtering analysis is performed on user interests, similar (interest) users of a specified user are found in a user group, and clicking conditions of the similar users on a certain article are integrated to form a preference degree prediction of the specified user on the article by the system. In the embodiment, feature derivation and feature engineering can be performed according to the preprocessed user behavior data and the preprocessed user static data by combining business knowledge and engineering experience, so as to determine the incoming data.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the determining the prediction result matrix according to the modeling data includes:
and inputting the input data into an XGBOOST+LR algorithm model by using a big data distributed computing framework, and determining a prediction result matrix.
The XGBOOST+LR algorithm model comprises two parts, namely XGBOOST and LR.
Among them, XGBOOST is based on GBDT, strives to exert speed and efficiency extremely, and is therefore called X (Extreme) GBoosted; the GBDT (gbooted, gradient Boosting Decision Tree, gradient lifting decision tree) is used to solve the optimization problem of the general loss function, and the method is to simulate the approximation of the residual error in the regression problem by using the value of the negative gradient of the loss function in the current model, which is a gradient descent tree model.
The LR (Logistic Regression ) is used for solving the regression problem with the dependent variable as the classified variable, usually the two-classification or binomial distribution problem, and also can solve the multi-classification problem, and belongs to a classification method.
The XGBOOST+LR algorithm model combines the XGBOOST and LR, and the performance of the model can be improved by using leaf nodes of the XGBOOST tree model as the input of the LR algorithm. Therefore, the input data are input into the XGBOOST+LR algorithm model by using the big data distributed computing framework, and a prediction result matrix can be determined.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the determining the function recommendation list according to the user similarity recommendation list and the prediction result matrix includes:
training a user similarity recommendation list and a prediction result matrix through an LR model, and determining a weighting coefficient;
weighting and fusing the user similarity recommendation list and the prediction result matrix by using a weighting coefficient to determine a user-function probability matrix;
and sequencing the user-function probability matrix in order from large to small to determine a function recommendation list.
In an embodiment, the user similarity recommendation list and the prediction result matrix are weighted and fused by using a weighting coefficient to determine a user-function probability matrix according to the following manner:
P u ＝αF u +βE u wherein α+β=1
Wherein P is u A user-function probability matrix; f (F) u Recommending a list for the similarity of the users; e (E) u Is a prediction result matrix; alpha and beta are weighting coefficients.
The foregoing expressions for determining the user-function probability matrix are given by way of example, and it will be understood by those skilled in the art that the above-described formulas may be modified and other parameters or data may be added to or provided with other specific formulas as desired in practice, and such modifications are intended to fall within the scope of the present invention.
When the application function personalized recommendation method of the embodiment of the invention is implemented, in one embodiment, the determining a personalized application function list according to the function list data and the function recommendation list of the preprocessing service system includes:
and selecting business functions from the function recommendation list, the hot function list and the new function list according to the set recommendation conditions, and establishing a personalized application function list.
In an embodiment, a set recommendation condition is that TOP-N service functions are respectively taken from a function recommendation list, a hot function list and a new function list, and a personalized application function list of a client is established. Further, another recommended condition is provided in the embodiment of the present invention: assuming that the length of all service system function lists in the service system is L, the TOP N hot function lists are taken, the TOP m new function lists are taken, and the TOP N function recommendation lists are taken, namely, the TOP N function recommendation lists are taken; and (3) eliminating the top L-N-m lists after the hot function list and the new function list, and combining the N function recommendation lists to establish a personalized application function list.
The embodiment of the invention utilizes the click behavior data of the user function to form the user behavior data; forming user static data by using personal information, post information and the like of the user, and automatically learning the demand degree and preference of the user on different functions by combining a function list, a new function list and a hot function list of a service system; learning a recommendation model based on the similarity of the user behaviors by utilizing the click behaviors of the user on the service system functions; recommendation model based on user behavior similarity, XGBOOST+LR prediction model based on user behavior and static information, and fusion scheme thereof; in the internal business system of the bank, a hot function list, a dynamically changeable function recommendation list and a new function recommendation list are combined. Automatically mining personalized requirements of users on internal business systems of banks by utilizing a big data distributed computing framework and a data mining method, and automatically generating a personalized application function list; the embodiment of the invention comprehensively considers the historical behavior information of the user and the personal information of the user, the popular functions of different posts, new functions and other factors, eliminates subjective factors, objectively provides a function list suitable for different users, and obviously provides functions which are not used by a certain user but are frequently used by the user of the same post. And providing a personalized display scheme of the bank internal business system function list. According to the scheme, daily required functions can be obviously provided for users, and office efficiency is improved; the method has the advantages that the function which is not found by the user and is good is provided, the potential needs of the user are mined, and the office availability and the service quality and efficiency of the system are further improved; the new function of the user service system is provided, the user needs are met, and the popularization cost of the new function by the requiring party can be saved.
Fig. 4 is a schematic diagram of a computer device running the method for implementing application function personalized recommendation according to the present invention, and as shown in fig. 4, an embodiment of the present invention further provides a computer device, including a memory, a processor, and a computer program stored on the memory and capable of running on the processor, where the processor implements the method for implementing application function personalized recommendation when executing the computer program.
The embodiment of the invention also provides a computer readable storage medium which stores a computer program for executing the personalized recommendation method for realizing the application function.
The embodiment of the invention also provides an application function personalized recommendation device, which is described in the following embodiment. Because the principle of the device for solving the problems is similar to that of an application function personalized recommendation method, the implementation of the device can refer to the implementation of the application function personalized recommendation method, and the repetition is omitted.
Fig. 5 is a schematic diagram of an application function personalized recommendation device according to an embodiment of the present invention, and as shown in fig. 5, an embodiment of the present invention further provides an application function personalized recommendation device, which may include:
A data acquisition module 501, configured to acquire user behavior data, user static data, and service system function list data;
the preprocessing module 502 is configured to preprocess the user behavior data, the user static data, and the service system function list data, and determine preprocessed user behavior data, preprocessed user static data, and preprocessed service system function list data;
a user behavior-function matrix determining module 503, configured to determine a user behavior-function matrix according to the preprocessed user behavior data;
a user similarity recommendation list determining module 504, configured to determine a user similarity recommendation list according to the user behavior-function matrix;
the incoming data determining module 505 is configured to determine incoming data according to the preprocessed user behavior data and the preprocessed user static data;
a prediction result matrix determining module 506, configured to determine a prediction result matrix according to the modulo data;
the function recommendation list determining module 507 is configured to determine a function recommendation list according to the user similarity recommendation list and the prediction result matrix;
the personalized application function list determining module 508 is configured to determine a personalized application function list according to the preprocessing service system function list data and the function recommendation list.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the preprocessing module is specifically configured to:
preprocessing user behavior data and user static data, removing data with null values and non-normative values, screening data which do not meet modeling requirements according to set screening rules, and determining preprocessed user behavior data and preprocessed user static data;
preprocessing service system function list data to determine preprocessed service system function list data; wherein, the preprocessing service system function list data comprises: a list of hot functions and a list of new functions.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the user behavior-function matrix determining module is specifically configured to:
performing time attenuation on the preprocessed user behavior data by using a time attenuation model, and determining the preprocessed user behavior data after the time attenuation;
preprocessing user behavior data according to time attenuation, and constructing a user behavior-function matrix.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the user behavior-function matrix determining module is further configured to determine the pre-process user behavior data after time attenuation according to the following manner:
Wherein f (x) ij ) Preprocessing user behavior data after time decay; t is a click behavior valued function; t is t 0 The time for the first click action to occur; k is an attenuation factor; t is the current time; t is t ij The time when the j function module was last clicked by the i user.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the user behavior-function matrix determining module is further configured to construct a user behavior-function matrix according to the following manner:
wherein C is u The function matrix is an m multiplied by n matrix, and each row represents the function module clicked by the user i; v i Representing the ith user; s is(s) i Representing a j-th functional module; y is ij Indicating that the ith user is at time t ij Clicking the j-th functional module; s is(s) ij Representing the sum Σf (x) of the decay of the user click module j ij ) The method comprises the steps of carrying out a first treatment on the surface of the M represents a functional module.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the user similarity recommendation list determining module is specifically configured to:
calculating the similarity between every two users according to the user behavior-function matrix through a Pearson correlation coefficient formula, and determining a user similarity matrix;
Ordering the user similarity matrixes according to the similarity from big to small, and determining the arrangement of the user similarity matrixes;
and selecting the user similarity matrix from the maximum value in the user similarity matrix arrangement according to the set selection quantity, calculating the user function click condition of the selected user similarity matrix, and determining a user similarity recommendation list.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the input data determining module is specifically configured to:
and carrying out characteristic derivation and characteristic engineering by combining business knowledge and engineering experience according to the preprocessed user behavior data and the preprocessed user static data, and determining the input data.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the prediction result matrix determining module is specifically configured to:
and inputting the input data into an XGBOOST+LR algorithm model by using a big data distributed computing framework, and determining a prediction result matrix.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the function recommendation list determining module is specifically configured to:
Training a user similarity recommendation list and a prediction result matrix through an LR model, and determining a weighting coefficient;
weighting and fusing the user similarity recommendation list and the prediction result matrix by using a weighting coefficient to determine a user-function probability matrix;
and sequencing the user-function probability matrix in order from large to small to determine a function recommendation list.
When the application function personalized recommendation device of the embodiment of the invention is implemented, in one embodiment, the foregoing personalized application function list determining module is specifically configured to:
and selecting business functions from the function recommendation list, the hot function list and the new function list according to the set recommendation conditions, and establishing a personalized application function list.
In summary, according to the application function personalized recommendation method and device provided by the embodiment of the invention, the historical behavior information and the user personal information of the user are comprehensively considered, the user behavior data and the user static data are utilized for preprocessing, the current demand level of the user on the function module is predicted through machine learning, a user similarity recommendation list and a prediction result matrix are constructed, the user similarity recommendation list and the prediction result matrix are fused, the function recommendation list is determined, finally, the function list data of the preprocessing service system are combined, and a personalized application function list is output. The embodiment of the invention can individually recommend the service functions in the service system by utilizing the existing client information, provide daily required functions for users and improve office efficiency; the function which is not found by the user is provided, the potential needs of the user are mined, and the office effectiveness and the service quality and efficiency of the system are further improved. The embodiment of the invention comprehensively considers the user static data such as user behavior data, user personal information and the like, various factors such as hot functions, new functions and the like of different posts, eliminates subjective factors and objectively provides a function list suitable for different users. The personalized requirements of different users on the function list are considered, a personalized recommendation model of each user is established, the required functions are displayed for each user, the user experience is improved, and the office efficiency is improved; the new functions of the user service system are obviously displayed, the new functions are better promoted, and more users try to use the new functions. And the user behavior data is utilized to combine with a machine learning algorithm to mine the function use preferences of different users at the same post, a function list meeting the user requirements is automatically generated, the function list is provided for the functions used by other people of the user, and the user is promoted to improve the office efficiency.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (9)
1. An application function personalized recommendation method is characterized by comprising the following steps:
acquiring user behavior data, user static data and service system function list data;
preprocessing user behavior data, user static data and service system function list data, and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data;
determining a user behavior-function matrix according to the preprocessed user behavior data;
determining a user similarity recommendation list according to the user behavior-function matrix;
determining incoming data according to the preprocessed user behavior data and the preprocessed user static data;
determining a prediction result matrix according to the modulus-entering data;
determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix;
determining a personalized application function list according to the function list data of the preprocessing service system and the function recommendation list;
determining a user behavior-function matrix from the pre-processed user behavior data, comprising:
performing time attenuation on the preprocessed user behavior data by using a time attenuation model, and determining the preprocessed user behavior data after the time attenuation;
Preprocessing user behavior data according to time attenuation, and constructing a user behavior-function matrix;
preprocessing user behavior data after determining time decay:
wherein f (x) ij ) Preprocessing user behavior data after time decay; t is a click behavior valued function; t is t 0 The time for the first click action to occur; k is an attenuation factor; t is the current time; t is t ij The time of the j function module is clicked by the user last time;
the user behavior-function matrix is constructed as follows:
wherein C is u The function matrix is a matrix of n multiplied by m, each row represents the function modules clicked by the user i, n represents the number of users, and m represents the number of modules; c ij Representing the sum of the user behavior data, Σf (x), preprocessed after each click of module j by user i ij )；
Determining a user similarity recommendation list according to the user behavior-function matrix, including:
calculating the similarity between every two users according to the user behavior-function matrix through a Pearson correlation coefficient formula, and determining a user similarity matrix;
ordering the user similarity matrixes according to the similarity from big to small, and determining the arrangement of the user similarity matrixes;
and selecting the user similarity matrix from the maximum value in the user similarity matrix arrangement according to the set selection quantity, calculating the user function click condition of the selected user similarity matrix, and determining a user similarity recommendation list.
2. The method of claim 1, wherein preprocessing the user behavior data, the user static data, and the business system function list data, determining the preprocessed user behavior data, the preprocessed user static data, and the preprocessed business system function list data, comprises:
preprocessing user behavior data and user static data, removing data with null values and non-normative values, screening data which do not meet modeling requirements according to set screening rules, and determining preprocessed user behavior data and preprocessed user static data;
preprocessing service system function list data to determine preprocessed service system function list data; wherein, the preprocessing service system function list data comprises: a list of hot functions and a list of new functions.
3. The method of claim 1, wherein determining the incoming data based on the pre-processed user behavior data and the pre-processed user static data comprises:
and carrying out characteristic derivation and characteristic engineering by combining business knowledge and engineering experience according to the preprocessed user behavior data and the preprocessed user static data, and determining the input data.
4. The method of claim 1, wherein determining a prediction result matrix from the modulo data comprises:
And inputting the input data into an XGBOOST+LR algorithm model by using a big data distributed computing framework, and determining a prediction result matrix.
5. The method of claim 1, wherein determining a list of functional recommendations based on the list of user similarity recommendations and the prediction result matrix comprises:
training a user similarity recommendation list and a prediction result matrix through an LR model, and determining a weighting coefficient;
weighting and fusing the user similarity recommendation list and the prediction result matrix by using a weighting coefficient to determine a user-function probability matrix;
and sequencing the user-function probability matrix in order from large to small to determine a function recommendation list.
6. The method of claim 2, wherein determining the personalized application function list based on the pre-processing service system function list data and the function recommendation list comprises:
and selecting business functions from the function recommendation list, the hot function list and the new function list according to the set recommendation conditions, and establishing a personalized application function list.
7. An application function personalized recommendation device, comprising:
the data acquisition module is used for acquiring user behavior data, user static data and service system function list data;
The preprocessing module is used for preprocessing the user behavior data, the user static data and the service system function list data and determining preprocessed user behavior data, preprocessed user static data and preprocessed service system function list data;
the user behavior-function matrix determining module is used for determining a user behavior-function matrix according to the preprocessed user behavior data;
the user similarity recommendation list determining module is used for determining a user similarity recommendation list according to the user behavior-function matrix;
the incoming data determining module is used for determining incoming data according to the preprocessed user behavior data and the preprocessed user static data;
the prediction result matrix determining module is used for determining a prediction result matrix according to the modulus-entering data;
the function recommendation list determining module is used for determining a function recommendation list according to the user similarity recommendation list and the prediction result matrix;
the personalized application function list determining module is used for determining a personalized application function list according to the function list data of the preprocessing service system and the function recommendation list;
the user behavior-function matrix determining module is specifically used for:
performing time attenuation on the preprocessed user behavior data by using a time attenuation model, and determining the preprocessed user behavior data after the time attenuation;
Preprocessing user behavior data according to time attenuation, and constructing a user behavior-function matrix;
the user behavior-function matrix determining module is further used for:
the user behavior data is preprocessed after determining the time decay as follows:
wherein f (x) ij ) Preprocessing user behavior data after time decay; t is a click behavior valued function; t is t 0 The time for the first click action to occur; k is an attenuation factor; t is the current time; t is t ij The time of the j function module is clicked by the user last time;
the user behavior-function matrix determining module is further configured to construct a user behavior-function matrix according to the following manner:
wherein C is u The function matrix is a matrix of n multiplied by m, each row represents the function modules clicked by the user i, n represents the number of users, and m represents the number of modules; c ij Representing the sum of the user behavior data, Σf (x), preprocessed after each click of module j by user i ij )；
The function recommendation list determining module is specifically configured to:
calculating the similarity between every two users according to the user behavior-function matrix through a Pearson correlation coefficient formula, and determining a user similarity matrix;
ordering the user similarity matrixes according to the similarity from big to small, and determining the arrangement of the user similarity matrixes;
And selecting the user similarity matrix from the maximum value in the user similarity matrix arrangement according to the set selection quantity, calculating the user function click condition of the selected user similarity matrix, and determining a user similarity recommendation list.
8. A computer device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of claims 1 to 6 when executing the computer program.
9. A computer readable storage medium, characterized in that the computer readable storage medium stores a computer program executable by a computer to implement the method of any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010574579.8A CN111737576B (en) | 2020-06-22 | 2020-06-22 | Application function personalized recommendation method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010574579.8A CN111737576B (en) | 2020-06-22 | 2020-06-22 | Application function personalized recommendation method and device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111737576A CN111737576A (en) | 2020-10-02 |
| CN111737576B true CN111737576B (en) | 2023-09-19 |
Family
ID=72650425
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010574579.8A Active CN111737576B (en) | 2020-06-22 | 2020-06-22 | Application function personalized recommendation method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111737576B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113344744B (en) * | 2021-08-02 | 2022-01-21 | 广东电网有限责任公司中山供电局 | Personalized business function calculation method and device for power system |
| CN115604049B (en) * | 2022-12-14 | 2023-06-13 | 忆月启函(盐城)科技有限公司 | Intelligent communication network system for household multifunctional monitoring |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105893609A (en) * | 2016-04-26 | 2016-08-24 | 南通大学 | Mobile APP recommendation method based on weighted mixing |
| CN105915949A (en) * | 2015-12-23 | 2016-08-31 | 乐视网信息技术（北京）股份有限公司 | Video content recommending method, device and system |
| CN106874355A (en) * | 2016-12-28 | 2017-06-20 | 浙江浙大网新集团有限公司 | The collaborative filtering method of social networks and user's similarity is incorporated simultaneously |
| CN107562818A (en) * | 2017-08-16 | 2018-01-09 | 中国工商银行股份有限公司 | Information recommendation system and method |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US9213729B2 (en) * | 2012-01-04 | 2015-12-15 | Trustgo Mobile, Inc. | Application recommendation system |
| US20180174218A1 (en) * | 2016-12-19 | 2018-06-21 | Sap Se | Recommendation optmization with a dynamic mixture of frequent and occasional recommendations |
-
2020
- 2020-06-22 CN CN202010574579.8A patent/CN111737576B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105915949A (en) * | 2015-12-23 | 2016-08-31 | 乐视网信息技术（北京）股份有限公司 | Video content recommending method, device and system |
| CN105893609A (en) * | 2016-04-26 | 2016-08-24 | 南通大学 | Mobile APP recommendation method based on weighted mixing |
| CN106874355A (en) * | 2016-12-28 | 2017-06-20 | 浙江浙大网新集团有限公司 | The collaborative filtering method of social networks and user's similarity is incorporated simultaneously |
| CN107562818A (en) * | 2017-08-16 | 2018-01-09 | 中国工商银行股份有限公司 | Information recommendation system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111737576A (en) | 2020-10-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Zhang et al. | A gated peripheral-foveal convolutional neural network for unified image aesthetic prediction | |
| US11106999B2 (en) | Automatic segmentation of a collection of user profiles | |
| CN105608477B (en) | Method and system for matching portrait with job position | |
| US8498950B2 (en) | System for training classifiers in multiple categories through active learning | |
| EP2885755B1 (en) | Agent system, agent control method and agent control program with ability of natural conversation with users | |
| CN109033408B (en) | Information pushing method and device, computer readable storage medium and electronic equipment | |
| US20160357845A1 (en) | Method and Apparatus for Classifying Object Based on Social Networking Service, and Storage Medium | |
| CN110110233B (en) | Information processing method, device, medium and computing equipment | |
| CN111400603A (en) | Information pushing method, device and equipment and computer readable storage medium | |
| CN112487291B (en) | Big data-based personalized news recommendation method and device | |
| CN106156163B (en) | Text classification method and device | |
| CN111737576B (en) | Application function personalized recommendation method and device | |
| US20220172260A1 (en) | Method, apparatus, storage medium, and device for generating user profile | |
| CN106557558A (en) | A kind of data analysing method and device | |
| CN112418320B (en) | Enterprise association relation identification method, device and storage medium | |
| CN111625715A (en) | Information extraction method and device, electronic equipment and storage medium | |
| CN115577316A (en) | User personality prediction method based on multi-mode data fusion and application | |
| CN112100221A (en) | Information recommendation method and device, recommendation server and storage medium | |
| CN109829154A (en) | Semantic-based personality prediction technique, user equipment, storage medium and device | |
| CN116468460A (en) | Consumer finance customer image recognition system and method based on artificial intelligence | |
| CN114428910A (en) | Resource recommendation method and device, electronic equipment, product and medium | |
| WO2023087933A1 (en) | Content recommendation method and apparatus, device, storage medium, and program product | |
| CN115048504A (en) | Information pushing method and device, computer equipment and computer readable storage medium | |
| CN114693409A (en) | Product matching method, device, computer equipment, storage medium and program product | |
| CN112507185A (en) | User portrait determination method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |