CN111125521A - Information recommendation method, device, equipment and storage medium - Google Patents
Information recommendation method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN111125521A CN111125521A CN201911280761.6A CN201911280761A CN111125521A CN 111125521 A CN111125521 A CN 111125521A CN 201911280761 A CN201911280761 A CN 201911280761A CN 111125521 A CN111125521 A CN 111125521A
- Authority
- CN
- China
- Prior art keywords
- information
- user
- vector
- model
- training
- 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
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
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/20—Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
- G06F16/24—Querying
- G06F16/245—Query processing
- G06F16/2458—Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
- G06F16/2465—Query processing support for facilitating data mining operations in structured databases
Abstract
The invention discloses an information recommendation method, device and equipment, wherein the method comprises the following steps: the method comprises the steps of obtaining an identification of a target user and information to be recommended from a database, determining the click through rate of the target user to the information to be recommended according to the identification of the target user and a recommendation model containing an interest extraction layer, and recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets a preset condition. By the method, when the click through rate of the user on the information to be recommended is estimated, the relevance between all information in the historical access record of the user and the relevance degree between the information to be recommended can be considered on the basis of the interest extraction layer of the recommendation model, so that the defects of the model in the prior art are avoided, and the click through rate of the information to be recommended is estimated more comprehensively and accurately.
Description
Technical Field
The present invention relates to information processing technologies, and in particular, to an information recommendation method, apparatus, device, and storage medium.
Background
In an information recommendation system, a deep learning model is generally adopted to estimate the click through rate of certain information. In the prior art, when the click through rate of certain information is estimated, a Deep Interest Network (DIN) model is usually adopted to eliminate interference of information irrelevant to the information to be recommended in a user historical access record on the estimated click through rate of the information to be recommended. However, the model can only learn the relevance between each information in the user historical access record and the information to be recommended, and the estimation of the click through rate of the information to be recommended is not comprehensive and accurate enough.
Disclosure of Invention
The invention provides an information recommendation method, device and equipment, which can avoid the defects of a model in the prior art and can more comprehensively and accurately estimate the click through rate of the information to be recommended.
In a first aspect, an embodiment of the present invention provides an information recommendation method, where the method includes:
acquiring the identification and the information to be recommended of a target user from a database;
determining the click through rate of the target user to the information to be recommended according to the identification and the recommendation model of the target user;
wherein, the recommendation model comprises an interest extraction layer;
and recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets the preset condition.
In a second aspect, an embodiment of the present invention further provides an information recommendation apparatus, where the apparatus includes:
the acquisition module is used for acquiring the identification of the target user and the information to be recommended from the database;
the determining module is used for determining the click through rate of the target user to the information to be recommended according to the identification of the target user and the recommendation model;
wherein, the recommendation model comprises an interest extraction layer;
and the recommending module is used for recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets the preset condition.
In a third aspect, an embodiment of the present invention further provides an information recommendation apparatus, where the apparatus includes:
a memory, a processor and a computer program stored on the memory and executable on the processor, which when executed by the processor, implement the information recommendation method as provided by the first aspect of the invention.
The embodiment of the invention provides an information recommendation method, an information recommendation device, information recommendation equipment and a storage medium, wherein an identification of a target user and information to be recommended are obtained from a database, the click through rate of the target user to the information to be recommended is determined according to the identification of the target user and a recommendation model containing an interest extraction layer, and when the click through rate of the information to be recommended meets a preset condition, the information to be recommended is recommended to the target user. By the method, when the click through rate of the user on the information to be recommended is estimated, the relevance between all information in the historical access record of the user and the relevance degree between the information to be recommended can be considered on the basis of the interest extraction layer of the recommendation model, so that the defects of the model in the prior art are avoided, and the click through rate of the information to be recommended is estimated more comprehensively and accurately.
Drawings
FIG. 1 is a flow chart of an information recommendation method in an embodiment of the invention;
FIG. 2 is a flow chart of an information recommendation method in an embodiment of the invention;
FIG. 3 is a schematic diagram of a recommendation model network in an embodiment of the present invention;
FIG. 4 is a flow chart of an information recommendation method in an embodiment of the invention;
FIG. 5 is a schematic structural diagram of an information recommendation apparatus according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of an information recommendation device in an embodiment of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples. It is to be understood that the specific embodiments described herein are merely illustrative of the invention and are not limiting of the invention. It should be further noted that, for the convenience of description, only some of the structures related to the present invention are shown in the drawings, not all of the structures.
In addition, in the embodiments of the present invention, the words "optionally" or "exemplarily" are used for indicating as examples, illustrations or explanations. Any embodiment or design described as "optionally" or "exemplary" in embodiments of the invention is not to be construed as preferred or advantageous over other embodiments or designs. Rather, use of the words "optionally" or "exemplarily" etc. is intended to present the relevant concepts in a concrete fashion.
Fig. 1 is an information recommendation method provided in an embodiment of the present invention, which may be applied to various network platforms to recommend various information that may be of interest to users of the network platforms, as shown in fig. 1, the method includes:
s101, obtaining the identification of the target user and the information to be recommended from the database.
The database in this embodiment may be a background database of each network platform, and the target user may be any user who browses and accesses the network platform, and accordingly, the identifier of the target user may be various identifiers such as a user name registered by the user on the network platform, a terminal device number accessing the network platform, and the like. The information to be recommended can be understood as any information stored in the network platform database.
When a user browses information on a network platform, such as watching a movie, playing music, browsing news, etc., the database of the network platform automatically records the user's identification. Therefore, the identification of the target user and the information to be recommended related to the step can be obtained from the database of the network platform.
S102, determining the click through rate of the target user to the information to be recommended according to the identification of the target user and the recommendation model.
In this embodiment, the recommendation model may include an interest extraction layer, where the interest extraction layer may be configured to obtain a correlation between pieces of information in the history access record of the target user, and a degree of association between the correlation between the pieces of information and the information to be recommended.
Therefore, after the identification and the pseudo-recommendation information of the target user are acquired based on step S101, the click through rate of the user corresponding to the identification to the pseudo-recommendation information can be determined through the recommendation model in this step.
For example, on the premise of acquiring the identifier of the target user, the historical access record of the user corresponding to the identifier of the target user can be automatically acquired from the database, and then the relevance among the information in the historical access record of the target user and the relevance degree between the relevance among the information in the historical access record and the information to be recommended are determined based on the interest extraction layer of the recommendation model, so that the click through rate of the target user to the information to be recommended is determined through the recommendation model.
Therefore, the recommendation model designed in the embodiment can overcome the defect that in the prior art, the model only learns the correlation between each information in the user historical access record and the to-be-recommended information, so that the click through rate of the to-be-recommended information by the user can be estimated more comprehensively and accurately.
S103, recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets a preset condition.
After the click through rate of the target user to the information to be recommended is obtained based on the recommendation model in step S102, the click through rate may be determined. If the click through rate of the target user to the information to be recommended meets the preset condition, the information to be recommended can be recommended to the target user.
For example, if the preset condition is that the click through rate is greater than a certain preset threshold, the click through rate satisfies the preset condition, that is, the click through rate of the target user on the to-be-recommended information is greater than the preset threshold in the preset condition, which indicates that the target user has a higher possibility of clicking to access the to-be-recommended information. Therefore, the information to be recommended can be recommended to the target user.
The embodiment provides an information recommendation method, which includes obtaining an identifier of a target user and information to be recommended from a database, determining a click through rate of the target user to the information to be recommended according to the identifier of the target user and a recommendation model including an interest extraction layer, and recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets a preset condition. By the method, when the click through rate of the user on the information to be recommended is estimated, the relevance between all information in the historical access record of the user and the relevance degree between the information to be recommended can be considered on the basis of the interest extraction layer of the recommendation model, so that the defects of the model in the prior art are avoided, and the click through rate of the information to be recommended is estimated more comprehensively and accurately.
Fig. 2 is a further detailed description of the determination process of the recommendation model in step S102 in the information recommendation method provided in this embodiment, and as shown in fig. 2, the method includes:
s201, constructing a training model.
Illustratively, the training model constructed as described above may include an input layer, a vector layer, an interest extraction layer, an interest layer, a weighted average layer, a concatenation layer, a fully connected layer, and an output layer. The network structure of the training model is shown in fig. 3.
And S202, acquiring training data according to the display click log.
In this embodiment, the presentation click log is used to indicate which presented information the user clicks (or accesses) and which presented information the user does not click.
Illustratively, the presentation click log may be presented in a tabular form, as shown in table 1.
TABLE 1
| User identification | Information identification | Label (R) |
| Xmly1 | a | 1 |
| Xmly2 | b | 0 |
| Xmly3 | c | 0 |
| Xmly4 | c | 1 |
| …… | …… | …… |
The training data obtained from the display click log may include a user identifier, an information identifier, and a tag, where the tag is used to mark whether a user corresponding to the user identifier accesses information corresponding to the information identifier.
For example, in combination with table 1, it is assumed that a tag is 1 to indicate that the user corresponding to the user identifier in the corresponding row has clicked the information corresponding to the information identifier in the corresponding row, and a tag is 0 to indicate that the user corresponding to the user identifier in the corresponding row has not clicked the information corresponding to the new identifier in the corresponding row.
Then, as can be seen from table 1, user Xmly1 clicked to access information a, user Xmly3 did not click to access information c, and user Xmly4 clicked to access information c.
Of course, those skilled in the art may also use other forms to present the relevant content in the display click log, and may also use other forms to indicate whether the user has clicked a certain display, for example, T and F are used to distinguish, and the embodiment does not limit this.
And S203, acquiring user characteristics and information characteristics from the database according to the training data.
Since the training data includes the user identifier, the information identifier and the label, the user characteristics obtained from the database based on the training data may be obtained from the database according to the user identifier, where the obtained user characteristics may include the user personal data and the user historical visit record.
For example, the user personal data may be characteristics of age, sex, taste and the like filled in by the user when the user registers on the network platform, the user historical access record may be a browsing access record automatically recorded and stored by the network platform database when the user accesses the network platform, the browsing access record may be used as a basis for evaluating the interest and preference of the user, and the browsing access record and the user personal data are used as the user characteristics together.
The information features obtained from the database based on the training data may be obtained from the database according to the information identifiers in the training data.
And S204, processing the user characteristics and the information characteristics to generate model input data.
The user characteristics obtained in step S203 include user personal data and user historical access records, and therefore, in this step, processing the user characteristics may be one-hot (one-hot) processing on the user personal data in the user characteristics and encoding the user historical access records in the user characteristics.
The encoding method for the user history access record may adopt any encoding method in the prior art, for example, encoding by using an integer sequence. Assuming that the number of user historical access records requiring coding is 100, the coding in the user historical access records can be 1-100 correspondingly.
Similarly, the information features may be processed by performing unique processing on the information features, and further, model input data is generated based on the processed user features and the information features, that is, the model input data is generated based on the user history access records coded in the user features, the user personal data after the unique processing, and the information features after the unique processing.
And S205, training the training model according to the model input data and the training data to generate a recommendation model.
As shown in fig. 3, the network structure of the training model provided in this embodiment includes an input layer, a vector layer, an interest extraction layer, an interest layer, a weighted average layer, a concatenation layer, a full connection layer, and an output layer, and therefore, a process of training the training model based on the model input data and the training data may be a process of processing information identifiers in the model input data and the training data based on each network layer in the training model.
For example, the training process may be to input the user history access record in the model input data and the information identifier in the training data as input layer data into a vector layer of the training model, and generate a user access information vector sequence and a to-be-processed information vector.
For example, assuming that the user history access record of the target user includes albums 1, 2, and 3, after the albums 1, 2, and 3 and the information identifier are input to the vector layer as input layer data, the album vectors 1, 2, and 3 and the information vector to be processed are correspondingly generated. The album vectors 1, 2 and 3 correspondingly form a user access information vector sequence, and the information identifier is processed by a vector layer to correspondingly generate an information vector to be processed.
It should be noted that the information identifier may be an identifier corresponding to information that is clicked (accessed) or not clicked (not accessed) by the user in the display click log.
And then, processing the user access information vector sequence and the information vector to be processed through an interest extraction layer in the training model to generate an interest vector sequence, and processing the interest vector sequence obtained by the interest extraction layer through the interest layer to generate a similarity sequence. Then, a weighted average layer in the training model processes the similarity sequence to generate a sequence vector, and a splicing layer in the training model splices the sequence vector, the information vector to be processed, the information characteristic in the model input data and the personal data of the user to generate a splicing vector.
And further, generating and predicting the click through rate of the information to be trained by a full-connection layer in the training model according to the splicing vector, outputting the click through rate by an output layer, and training the constructed training model according to the click through rate.
It can be understood that, in the above processing process, since the model input data is generated after processing the user characteristic and the information characteristic, when the concatenation layer concatenates the sequence vector with the information characteristic and the user personal data in the model input data, the information characteristic and the user personal data are corresponding data after being subjected to unique heat processing.
Different model input data and information to be trained are input into the training model, and corresponding processing is carried out on the basis of each network layer in the training model, so that the recommendation model is generated through repeated processing and training processes.
In addition, since the recommended model is obtained by training a training model, the recommended model has the same network structure as the training model.
Fig. 4 is a flowchart illustrating a processing procedure of the interest extraction layer and the interest layer in step S205 in further detail, where as shown in fig. 4, the method includes:
s401, calculating a correlation coefficient between each vector in the user access information vector sequence and the information vector to be processed in the interest extraction layer.
Assuming that the user history access record of the target user includes albums 1, 2, and 3, and correspondingly, the user access information vector sequence includes album vectors 1, 2, and 3, the processing procedure in this step may be to calculate correlation coefficients between the album vectors 1, 2, and 3 and the information vectors to be processed, respectively.
S402, multiplying each vector in the user access information vector sequence by the corresponding correlation coefficient to generate a weighted information vector corresponding to each vector.
For example, it is assumed that the correlation coefficient between the album vector 1 and the information vector to be processed calculated in step S401 is r11, the correlation coefficient between the album vector 2 and the information vector to be processed is r12, and the correlation coefficient between the album vector 3 and the information vector to be processed is r 13.
Then, each vector in the user access information vector sequence is multiplied by the corresponding correlation coefficient, that is, the album vector 1 is multiplied by the coefficient r11, the album vector 2 is multiplied by the coefficient r12, and the album vector 3 is multiplied by the coefficient r13, so that the weighted information vectors corresponding to the respective vectors are generated.
And S403, the interest extraction layer performs interest extraction on each weighted information vector to generate an interest vector sequence.
After the calculation of step S401 and step S402 in the interest extraction layer, the interest extraction is further performed on the obtained weighted information vector to generate an interest vector sequence.
Through the processing of the interest extraction layer, the correlation between the albums 1, 2 and 3 in the user access information vector sequence and the correlation degree between the correlation between the albums 1, 2 and 3 and the information to be processed can be obtained.
S404, the interest layer makes cosine on each vector in the interest vector sequence and the information vector to be processed, and a similarity sequence between the interest vector sequence and the information to be processed is generated.
Based on the processing in steps S401 to S403, the generated interest vector sequence includes three vectors, which are assumed to be vector 1, vector 2, and vector 3, respectively, and then the cosine between each vector in the interest vector sequence and the information vector to be processed is obtained through the interest layer, and three cosine values are also obtained, which are assumed to be r1, r2, and r3, respectively, and then the three cosine values form a similarity sequence [ r1, r2, r3 ] between the interest vector sequence and the information to be processed.
Thus, the interest layer-based processing can further acquire the degree of association between the correlation between the albums 1, 2 and 3 in the user access information vector sequence and the information to be processed.
Fig. 5 is an information recommendation apparatus provided in this embodiment, and as shown in fig. 5, the apparatus includes: an acquisition module 501, a determination module 502 and a recommendation module 503;
the system comprises an acquisition module, a recommendation module and a recommendation module, wherein the acquisition module is used for acquiring the identification of a target user and the information to be recommended from a database;
the determining module is used for determining the click through rate of the target user to the information to be recommended according to the identification of the target user and the recommendation model;
wherein, the recommendation model comprises an interest extraction layer;
and the recommending module is used for recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets the preset condition.
Further, the information recommendation apparatus further includes: the system comprises a construction module, a generation module and a training module;
the building module is used for building a training model;
the acquisition module is also used for acquiring training data according to the display click log and acquiring user characteristics and information characteristics from a database according to the training data;
the training data comprises a user identifier, an information identifier and a label, wherein the label is used for marking whether a user corresponding to the user identifier accesses information corresponding to the information identifier;
the generating module is used for processing the user characteristics and the information characteristics to generate model input data;
and the training model is used for training the training model according to the model input data and the training data to generate a recommendation model.
Further, the obtaining module is specifically configured to obtain user characteristics from the database according to the user identifier, where the user characteristics include user personal data and a user historical access record, and obtain the information characteristics from the database according to the information identifier.
Further, the generating module is specifically configured to perform unique processing on the user personal data and the information features in the user features, encode the user historical access records in the user features, and generate the model input data according to the user features and the information features after the unique processing.
In one example, the training module is specifically configured to input a user history access record in model input data and an information identifier in training data into a vector layer in a training model, and generate a user access information vector sequence and a to-be-processed information vector; processing the user access information vector sequence and the information vector to be processed according to an interest extraction layer in the training model to generate an interest vector sequence; processing the interest vector sequence according to the interest layer of the training model to generate a similarity sequence; processing the similarity sequence according to the weighted average layer of the training model to generate a sequence vector; splicing the sequence vector, the information vector to be processed and the information characteristics in the model input data and the user personal data according to the splicing layer of the training model to generate a splicing vector; generating a click through rate for predicting the information to be trained through a full connection layer of the training model according to the splicing vector; and outputting the click through rate through an output layer of the training model, training the training model according to the click through rate, and generating a recommendation model.
Further, the training module may generate the interest vector sequence by calculating a correlation coefficient between each vector in the user access information vector sequence and the information vector to be processed in the interest extraction layer; multiplying each vector in the user access information vector sequence by the corresponding correlation coefficient to generate a weighted information vector corresponding to each vector; and performing interest extraction on each weighted information vector in an interest extraction layer to generate an interest vector sequence.
Furthermore, the training module may generate the similarity sequence by cosine-dividing each vector in the interest vector sequence and the information vector to be processed in the interest layer, so as to generate the similarity sequence between the interest vector sequence and the information to be processed.
The information recommendation device provided in fig. 5 can execute the information recommendation methods provided in fig. 1, fig. 2, and fig. 4, and has the corresponding functional modules and beneficial effects of the execution methods.
Fig. 6 is a schematic structural diagram of an information recommendation apparatus according to an embodiment of the present invention, as shown in fig. 6, the apparatus includes a processor 601, a memory 602, an input device 603, and an output device 604; the number of processors 601 in the device may be one or more, and one processor 601 is taken as an example in fig. 6; the processor 601, the memory 602, the input device 603 and the output device 604 of the apparatus may be connected by a bus or other means, and fig. 6 illustrates the connection by a bus as an example.
The memory 602 is used as a computer-readable storage medium for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the information recommendation methods in fig. 1-4 according to the embodiments of the present invention (e.g., the obtaining module 501, the determining module 502, and the recommending module 503 in the information recommendation apparatus). The processor 601 executes various functional applications and data processing of the device by running software programs, instructions and modules stored in the memory 602, that is, implements the information recommendation method described above.
The memory 602 may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application program required for at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory 602 may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, the memory 602 may further include memory located remotely from the processor 601, which may be connected to the device/terminal/server via a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device 603 may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the apparatus. The output means 304 may comprise a display device such as a display screen.
Embodiments of the present invention also provide a storage medium containing computer-executable instructions, which when executed by a computer processor, perform a method for information recommendation, the method comprising:
acquiring the identification and the information to be recommended of a target user from a database;
determining the click through rate of the target user to the information to be recommended according to the identification and the recommendation model of the target user;
wherein, the recommendation model comprises an interest extraction layer;
and recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets the preset condition.
Of course, the storage medium provided by the embodiment of the present invention contains computer-executable instructions, and the computer-executable instructions are not limited to the method operations described above, and may also perform related operations in the information recommendation method provided by any embodiment of the present invention.
From the above description of the embodiments, it is obvious for those skilled in the art that the present invention can be implemented by software and necessary general hardware, and certainly, can also be implemented by hardware, but the former is a better embodiment in many cases. Based on such understanding, the technical solutions of the present invention may be embodied in the form of a software product, which may be stored in a computer-readable storage medium, such as a floppy disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a FLASH Memory (FLASH), a hard disk or an optical disk of a computer, and includes several instructions for enabling a computer device (which may be a personal computer, a server, or a network device) to execute the methods according to the embodiments of the present invention.
It should be noted that, in the embodiment of the information recommendation apparatus, the included units and modules are merely divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be implemented; in addition, specific names of the functional units are only for convenience of distinguishing from each other, and are not used for limiting the protection scope of the present invention.
It is to be noted that the foregoing is only illustrative of the preferred embodiments of the present invention and the technical principles employed. It will be understood by those skilled in the art that the present invention is not limited to the particular embodiments described herein, but is capable of various obvious changes, rearrangements and substitutions as will now become apparent to those skilled in the art without departing from the scope of the invention. Therefore, although the present invention has been described in greater detail by the above embodiments, the present invention is not limited to the above embodiments, and may include other equivalent embodiments without departing from the spirit of the present invention, and the scope of the present invention is determined by the scope of the appended claims.
Claims (10)
1. An information recommendation method, comprising:
acquiring the identification and the information to be recommended of a target user from a database;
determining the click through rate of the target user to the information to be recommended according to the identification of the target user and a recommendation model;
wherein, the recommendation model comprises an interest extraction layer;
and recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets a preset condition.
2. The method of claim 1, wherein determining the recommendation model comprises:
constructing a training model;
acquiring training data according to the display click log;
acquiring user characteristics and information characteristics from a database according to the training data;
processing the user characteristics and the information characteristics to generate model input data;
and training the training model according to the model input data and the training data to generate a recommendation model.
3. The method of claim 2, wherein the training data comprises: the system comprises a user identification, an information identification and a label, wherein the label is used for marking whether a user corresponding to the user identification accesses information corresponding to the information identification.
4. The method of claim 3, wherein obtaining user characteristics and information characteristics from a database based on the training data comprises:
acquiring the user characteristics from a database according to the user identification, wherein the user characteristics comprise user personal data and user historical access records;
and acquiring the information characteristics from a database according to the information identifier.
5. The method of any of claims 2-4, wherein processing the user characteristics and information characteristics to generate model input data comprises:
performing one-hot processing on the user personal data and the information characteristics in the user characteristics;
encoding a user historical access record in the user characteristics;
and generating model input data according to the processed user characteristics and information characteristics.
6. The method of claim 5, wherein training the training model based on the model input data and the training data to generate a recommendation model comprises:
inputting the user historical access records in the model input data and the information identification in the training data into a vector layer in the training model to generate a user access information vector sequence and an information vector to be processed;
an interest extraction layer in the training model processes the user access information vector sequence and the information vector to be processed to generate an interest vector sequence;
the interest layer of the training model processes the interest vector sequence to generate a similarity sequence;
processing the similarity sequence by a weighted average layer of the training model to generate a sequence vector;
the splicing layer of the training model splices the sequence vector, the information vector to be processed, the information characteristics in the model input data and the personal data of the user to generate a spliced vector;
the full-connection layer of the training model generates and predicts the click through rate of the information to be trained according to the splicing vector;
and the output layer of the training model outputs the click through rate, trains the training model according to the click through rate and generates a recommendation model.
7. The method of claim 6, wherein an interest extraction layer in the training model processes the sequence of user visit information vectors and information vectors to be processed to generate a sequence of interest vectors, comprising:
calculating a correlation coefficient between each vector in the user access information vector sequence and the information vector to be processed in the interest extraction layer;
multiplying each vector in the user access information vector sequence by the corresponding correlation coefficient to generate a weighted information vector corresponding to each vector;
and the interest extraction layer performs interest extraction on each weighted information vector to generate an interest vector sequence.
8. The method of claim 6 or 7, wherein the interest layer of the training model processes the interest vector sequence to generate a similarity sequence, comprising:
and the interest layer makes cosine on each vector in the interest vector sequence and the information vector to be processed to generate a similarity sequence between the interest vector sequence and the information to be processed.
9. An information recommendation apparatus, comprising:
the acquisition module is used for acquiring the identification of the target user and the information to be recommended from the database;
the determining module is used for determining the click through rate of the target user to the information to be recommended according to the identification of the target user and the recommendation model;
wherein, the recommendation model comprises an interest extraction layer;
and the recommending module is used for recommending the information to be recommended to the target user when the click through rate of the information to be recommended meets a preset condition.
10. An information recommendation device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor implements the information recommendation method of any one of claims 1-8 when executing the computer program.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911280761.6A CN111125521A (en) | 2019-12-13 | 2019-12-13 | Information recommendation method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911280761.6A CN111125521A (en) | 2019-12-13 | 2019-12-13 | Information recommendation method, device, equipment and storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111125521A true CN111125521A (en) | 2020-05-08 |
Family
ID=70500016
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911280761.6A Pending CN111125521A (en) | 2019-12-13 | 2019-12-13 | Information recommendation method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111125521A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111639257A (en) * | 2020-05-09 | 2020-09-08 | 北京三快在线科技有限公司 | Information display method, information display device, storage medium and electronic equipment |
| CN111858688A (en) * | 2020-07-20 | 2020-10-30 | 海尔优家智能科技（北京）有限公司 | Textile material, color chart recommendation method and device and storage medium |
| CN112487513A (en) * | 2020-12-11 | 2021-03-12 | 广东电网有限责任公司 | Method, device and system for design scheme drawing integration optimization |
| CN112800097A (en) * | 2021-01-15 | 2021-05-14 | 稿定（厦门）科技有限公司 | Special topic recommendation method and device based on deep interest network |
| CN113343085A (en) * | 2021-05-27 | 2021-09-03 | 北京三快在线科技有限公司 | Information recommendation method and device, storage medium and electronic equipment |
| CN113743975A (en) * | 2021-01-29 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Advertisement effect processing method and device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913273A (en) * | 2015-12-14 | 2016-08-31 | 乐视网信息技术（北京）股份有限公司 | Intelligent recommendation all-terminal display method and device |
| CN109740068A (en) * | 2019-01-29 | 2019-05-10 | 腾讯科技（北京）有限公司 | Media data recommended method, device and storage medium |
| CN109862432A (en) * | 2019-01-31 | 2019-06-07 | 厦门美图之家科技有限公司 | Clicking rate prediction technique and device |
| CN109960759A (en) * | 2019-03-22 | 2019-07-02 | 中山大学 | Recommender system clicking rate prediction technique based on deep neural network |
| CN110413877A (en) * | 2019-07-02 | 2019-11-05 | 阿里巴巴集团控股有限公司 | A kind of resource recommendation method, device and electronic equipment |
| CN110516162A (en) * | 2019-09-02 | 2019-11-29 | 上海喜马拉雅科技有限公司 | A kind of information recommendation method, device, equipment and storage medium |
-
2019
- 2019-12-13 CN CN201911280761.6A patent/CN111125521A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105913273A (en) * | 2015-12-14 | 2016-08-31 | 乐视网信息技术（北京）股份有限公司 | Intelligent recommendation all-terminal display method and device |
| CN109740068A (en) * | 2019-01-29 | 2019-05-10 | 腾讯科技（北京）有限公司 | Media data recommended method, device and storage medium |
| CN109862432A (en) * | 2019-01-31 | 2019-06-07 | 厦门美图之家科技有限公司 | Clicking rate prediction technique and device |
| CN109960759A (en) * | 2019-03-22 | 2019-07-02 | 中山大学 | Recommender system clicking rate prediction technique based on deep neural network |
| CN110413877A (en) * | 2019-07-02 | 2019-11-05 | 阿里巴巴集团控股有限公司 | A kind of resource recommendation method, device and electronic equipment |
| CN110516162A (en) * | 2019-09-02 | 2019-11-29 | 上海喜马拉雅科技有限公司 | A kind of information recommendation method, device, equipment and storage medium |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111639257A (en) * | 2020-05-09 | 2020-09-08 | 北京三快在线科技有限公司 | Information display method, information display device, storage medium and electronic equipment |
| CN111858688A (en) * | 2020-07-20 | 2020-10-30 | 海尔优家智能科技（北京）有限公司 | Textile material, color chart recommendation method and device and storage medium |
| CN112487513A (en) * | 2020-12-11 | 2021-03-12 | 广东电网有限责任公司 | Method, device and system for design scheme drawing integration optimization |
| CN112800097A (en) * | 2021-01-15 | 2021-05-14 | 稿定（厦门）科技有限公司 | Special topic recommendation method and device based on deep interest network |
| CN113743975A (en) * | 2021-01-29 | 2021-12-03 | 北京沃东天骏信息技术有限公司 | Advertisement effect processing method and device |
| CN113343085A (en) * | 2021-05-27 | 2021-09-03 | 北京三快在线科技有限公司 | Information recommendation method and device, storage medium and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110825957B (en) | Deep learning-based information recommendation method, device, equipment and storage medium | |
| CN111125521A (en) | Information recommendation method, device, equipment and storage medium | |
| CN110149540B (en) | Recommendation processing method and device for multimedia resources, terminal and readable medium | |
| US10747771B2 (en) | Method and apparatus for determining hot event | |
| CN110532469B (en) | Information recommendation method, device, equipment and storage medium | |
| CN110896488B (en) | Recommendation method for live broadcast room and related equipment | |
| CN110598109A (en) | Information recommendation method, device, equipment and storage medium | |
| CN111400586A (en) | Group display method, terminal, server, system and storage medium | |
| CN111046188A (en) | User preference degree determining method and device, electronic equipment and readable storage medium | |
| CN111159563A (en) | Method, device and equipment for determining user interest point information and storage medium | |
| CN110909258B (en) | Information recommendation method, device, equipment and storage medium | |
| CN111460290A (en) | Information recommendation method, device, equipment and storage medium | |
| CN110516162B (en) | Information recommendation method, device, equipment and storage medium | |
| CN110516164B (en) | Information recommendation method, device, equipment and storage medium | |
| CN110765348B (en) | Hot word recommendation method and device, electronic equipment and storage medium | |
| CN111026910B (en) | Video recommendation method, device, electronic equipment and computer readable storage medium | |
| CN107562847A (en) | Information processing method and related product | |
| CN108021713B (en) | Document clustering method and device | |
| CN111368195B (en) | Model training method, device, equipment and storage medium | |
| CN110995681B (en) | User identification method and device, electronic equipment and storage medium | |
| CN107665202B (en) | Method and device for constructing interest model and electronic equipment | |
| CN110213660B (en) | Program distribution method, system, computer device and storage medium | |
| CN111026849A (en) | Data processing method and device | |
| CN111507471B (en) | Model training method, device, equipment and storage medium | |
| CN110059502B (en) | Private data sensing 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 |