An information recommendation method, device, storage medium and electronic device

Abstract

This specification discloses an information recommendation method, apparatus, storage medium, and electronic device. The information recommendation method includes: obtaining information to be recommended; inputting the information to be recommended into a pre-trained information extraction model to extract, through the information extraction model, the information of each key point included in the information to be recommended, where the information of the key point is used to characterize the information content corresponding to the key time node in the information to be recommended; inputting the information of each key point into a pre-trained evaluation model to determine, through the evaluation model, the evaluation value corresponding to each key point information, and for each key point information, the evaluation value corresponding to the key point information is used to characterize the contribution degree of the key point information to the expected browsing situation of the information to be recommended; adjusting the information to be recommended according to the evaluation value, so as to perform information recommendation through the adjusted information to be recommended. This solution optimizes the information to be recommended according to the evaluation value corresponding to each key point, significantly improving the information recommendation effect.

Description

Technical Field

[0001] This specification relates to the field of computer technology, and particularly to an information recommendation method, apparatus, storage medium, and electronic device. Background Art

[0002] As an effective marketing method, information recommendation can, on the one hand, fully meet user needs, and on the other hand, attract users to use products or enjoy services. Therefore, it is widely applied to the promotion and publicity of industrial products, food, games, film and television works, and service items.

[0003] However, currently, the information recommended during information recommendation usually depends on manual production. The understanding degree and production level of the producer severely limit the effect of information recommendation. It is difficult for the information recommended to users to better express the actual content, resulting in low accuracy of content expression and greatly affecting the user experience.

[0004] Therefore, how to improve the information recommendation effect is an urgent problem to be solved. Summary of the Invention

[0005] This specification provides an information recommendation method, apparatus, storage medium, and electronic device to partially solve the above problems existing in the prior art.

[0006] This specification adopts the following technical solutions:

[0007] This specification provides an information recommendation method, including:

[0008] Obtain the information to be recommended;

[0009] Input the information to be recommended into a pre-trained information extraction model, so as to extract, through the information extraction model, the information of each key point contained in the information to be recommended, and the key point information is used to represent the information content corresponding to the key time node in the information to be recommended;

[0010] Input the information of each key point into a pre-trained evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to each key point information. For each key point information, the evaluation value corresponding to the key point information is used to represent the contribution degree of the key point information to the expected browsing situation of the information to be recommended;

[0011] Adjust the information to be recommended according to the evaluation value, so as to perform information recommendation through the adjusted information to be recommended.

[0012] Optionally, before the step of inputting the information to be recommended into the pre-trained information extraction model, the method further includes:

[0013] Partition the information to be recommended to obtain at least one type of sub - data, where the types include: text, image, audio;

[0014] The step of inputting the information to be recommended into a pre - trained information extraction model to extract the information of each key point contained in the information to be recommended through the information extraction model specifically includes:

[0015] Input the at least one type of sub - data into the information extraction model to extract the information of each key point contained in the information to be recommended through the information extraction model.

[0016] Optionally, the step of extracting the information of each key point contained in the information to be recommended through the information extraction model specifically includes:

[0017] For each time node in the information to be recommended, determine the richness value corresponding to each type of sub - data at this time node;

[0018] Determine the information of each key point contained in the information to be recommended according to the richness value corresponding to each type of sub - data at each time node.

[0019] Optionally, the step of determining the information of each key point contained in the information to be recommended according to the richness value corresponding to each type of sub - data at each time node specifically includes:

[0020] For each time node in the information to be recommended, determine the key degree corresponding to this time node according to the richness value corresponding to each type of sub - data at this time node and the preset weight corresponding to each type of sub - data;

[0021] If the key degree corresponding to this time node is greater than the preset key degree, then determine the key point information according to the information content corresponding to this time node.

[0022] Optionally, before the step of inputting the information to be recommended into a pre - trained information extraction model, the method further includes:

[0023] Process the information to be recommended to obtain a data sequence with a specified number of frames;

[0024] The step of inputting the information to be recommended into a pre - trained information extraction model to extract the information of each key point contained in the information to be recommended through the information extraction model specifically includes:

[0025] Input the data sequence with the specified number of frames into the information extraction model to extract the information of each key point contained in the data sequence through the information extraction model.

[0026] Optionally, the step before inputting the information to be recommended into the pre-trained information extraction model further includes:

[0027] Determine the application scenario corresponding to the information content of the information to be recommended;

[0028] The step of inputting the information of each element point into the pre-trained evaluation model to determine the evaluation value corresponding to each element point information through the evaluation model specifically includes:

[0029] Input the information of each element point and the scenario information of the application scenario into the evaluation model to determine the evaluation value corresponding to each element point information in the application scenario through the evaluation model.

[0030] Optionally, the step of optimizing the information to be recommended according to the evaluation value specifically includes:

[0031] For each element point information, if the evaluation value corresponding to the element point information is lower than the preset threshold, then a number of element point information before the time node corresponding to the element point information and a number of element point information after the time node corresponding to the element point information are used as the context information corresponding to the element point information;

[0032] Adjust the information content at the time node corresponding to the element point information according to the context information.

[0033] Optionally, the step of training the information extraction model includes:

[0034] Obtain historical recommended information and determine the browsing records of each user for the historical recommended information, where the browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommended information;

[0035] Input the historical recommended information into the information extraction model to determine the historical element point information included in the historical recommended information through the information extraction model, and determine the key time nodes in the historical recommended information according to the browsing records;

[0036] Train the information extraction model with the optimization goal of minimizing the deviation between the historical element point information and the information content corresponding to the key time nodes in the historical recommended information.

[0037] Optionally, the step of training the evaluation model includes:

[0038] Obtain historical recommendation information and determine the browsing records of each user for the historical recommendation information. The browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommendation information;

[0039] Input the historical recommendation information into a pre-trained information extraction model to extract the information of each key point included in the recommendation information as historical key point information through the information extraction model;

[0040] Input the historical key point information into the evaluation model to determine the evaluation value corresponding to each historical key point information through the evaluation model;

[0041] Train the evaluation model according to the evaluation value corresponding to each historical key point information and the browsing records.

[0042] This specification provides an information recommendation device, including:

[0043] An acquisition module, configured to acquire information to be recommended;

[0044] An extraction module, configured to input the information to be recommended into a pre-trained information extraction model to extract the information of each key point included in the information to be recommended through the information extraction model. The key point information is used to characterize the information content corresponding to the key time node in the information to be recommended;

[0045] An evaluation module, configured to input the information of each key point into a pre-trained evaluation model to determine the evaluation value corresponding to each key point information through the evaluation model. For each key point information, the evaluation value corresponding to the key point information is used to characterize the contribution degree of the key point information to the expected browsing situation of the information to be recommended;

[0046] A recommendation module, configured to adjust the information to be recommended according to the evaluation value to perform information recommendation through the adjusted information to be recommended.

[0047] This specification provides a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the above information recommendation method is implemented.

[0048] This specification provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the above information recommendation method is implemented.

[0049] At least one of the above technical solutions adopted in this specification can achieve the following beneficial effects:

[0050] In the information recommendation method provided in this specification, obtain the information to be recommended; input the information to be recommended into a pre-trained information extraction model to extract, through the information extraction model, the information content of each key point included in the information to be recommended, where the information content of the key point is used to characterize the information content corresponding to the key time node in the information to be recommended; input the information content of each key point into a pre-trained evaluation model to determine, through the evaluation model, the evaluation value corresponding to each key point information. For each key point information, the evaluation value corresponding to the key point information is used to characterize the contribution degree of the key point information to the expected browsing situation of the information to be recommended; adjust the information to be recommended according to the evaluation value, so as to perform information recommendation through the adjusted information to be recommended.

[0051] As can be seen from the above method, in the process of information recommendation for the information to be recommended, this solution can extract the information content of each key point included in the information to be recommended and evaluate each key point information. In this way, the information to be recommended can be optimized based on the evaluation value corresponding to each key point information, so as to make the information to be recommended achieve the best recommendation effect. Brief Description of the Drawings

[0052] The drawings described herein are used to provide a further understanding of this specification and constitute a part of this specification. The illustrative embodiments of this specification and their descriptions are used to explain this specification and do not constitute an improper limitation to this specification. In the drawings:

[0053] Figure 1 It is a schematic flowchart of an information recommendation method provided in this specification;

[0054] Figure 2 It is a schematic diagram of an optimization method for advertising materials provided in this specification;

[0055] Figure 3 It is a schematic diagram of an information recommendation device provided in this specification;

[0056] Figure 4 It is provided in this specification corresponding to Figure 1 a schematic diagram of an electronic device. Detailed Description of the Embodiments

[0057] To make the purpose, technical solutions, and advantages of this specification clearer, the technical solutions of this specification will be clearly and completely described below in conjunction with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this specification, rather than all the embodiments. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of this specification.

[0058] The following will, in conjunction with the accompanying drawings, elaborate in detail on the technical solutions provided in each embodiment of this specification.

[0059] Figure 1 The following is a schematic flowchart of an information recommendation method provided in this specification, including the following steps:

[0060] S101: Obtain the information to be recommended.

[0061] With the booming development of the game industry, advertising and marketing have become an important means to attract players, and the quality of advertising materials directly affects the advertising effect. However, the production of traditional game advertising materials mostly relies on manual work, which is not only time-consuming and laborious, but also difficult to quantify the effect.

[0062] Based on this, this specification provides an information recommendation method, which extracts the key point information in the information to be recommended (advertising materials) through a deep learning model, and optimizes the information to be recommended based on the evaluation results of each key point information, so as to further improve the attractiveness of the information to be recommended to users.

[0063] In this specification, the execution entity for implementing an information recommendation method can be a specified device such as a server. Of course, it can also be a client. For the convenience of description, the following will only take the server as an example of the execution entity to illustrate an information recommendation method provided in this specification.

[0064] Among them, the server can obtain the information to be recommended, and the information to be recommended can include at least one of video, text, and audio. In this specification, the information to be recommended can be advertising materials in a target application scenario.

[0065] The server can determine the application scenario corresponding to the information content of the information to be recommended. In this specification, the application scenario can include: product recommendation, game recommendation, film and television work recommendation, service item recommendation, etc. The scenario information of the application scenario can include the classification information of the application scenario. Taking the game recommendation scenario as an example, its scenario information can include the style types of games, such as action, strategy, cultivation, etc., and taking film and television works as an example, its scenario information can include the types of film and television works, such as martial arts, romance, reasoning, science fiction, etc.

[0066] After obtaining the information to be recommended, the server can preprocess the information to be recommended. Among them, the server can process the information to be recommended to obtain a data sequence of specified frames. For example, for a 30-frame video, the server can extract an image every 3 frames, so as to obtain a data sequence containing 10 frames of images, thereby realizing data compression and improving the overall efficiency.

[0067] In addition, the server can also divide the information to be recommended to obtain at least one type of sub-data. The types of sub-data can include text, images, and audio. Taking the information to be recommended composed of video, text, and audio as an example, the server can divide it into pure image data, pure audio data, and pure text data, and the data sequences of different types of data are aligned with each other in time.

[0068] It should be noted that the server can first determine the data sequence of the specified number of frames, and then divide the data sequence of the specified number of frames into different types of sub-data based on the data sequence of the specified number of frames as the processed data. It can also first divide the information to be recommended into different types of sub-data based on the data sequence of the specified number of frames, and then determine the data sequence of the specified number of frames corresponding to different types of sub-data. Of course, the server can also use any one of the methods to preprocess the information to be recommended to obtain the processed data.

[0069] S102: Input the information to be recommended into a pre-trained information extraction model, so as to extract, through the information extraction model, the information of each key point included in the information to be recommended, where the key point information is used to characterize the information content corresponding to the key time node in the information to be recommended.

[0070] The server can input the above-mentioned processed data into a pre-trained information extraction model, and through this information extraction model, extract the information of each key point included in the processed data.

[0071] In this specification, the time node can refer to a certain frame or a certain moment in the time sequence of the information to be recommended. Of course, it can also refer to a specific time period, and the length of each time period can be determined according to the actual situation. This specification does not make specific limitations on this.

[0072] The key point information is used to characterize the information content corresponding to the key time node in the information to be recommended. This key time node can be understood as the time point or time period when the user's browsing and clicking situations may change significantly, such as the time node when the user's click-through rate and browsing rate change significantly, or the time node when the user stays for a long time.

[0073] In addition, in addition to including image, audio, and text content, the above-mentioned information content can also include special effects added to the video.

[0074] Specifically, when the processed data is different types of sub-data, the server can input at least one type of sub-data into the information extraction model, so as to extract, through the information extraction model, the information of each key point included in the information to be recommended.

[0075] For each time node in the information to be recommended, the information extraction model can determine the richness value corresponding to each type of sub-data at that time node. For text data, its richness can be determined based on the number of words in the text. The more words, the greater the richness value; conversely, the smaller the value. For image data, its richness can be used to express the diversity of elements (such as people, buildings, items, etc.) and the vividness of colors in the image. For audio data, its richness can be used to express the diversity of melodies and pitch changes.

[0076] After that, the information points of each element included in the information to be recommended can be determined according to the richness value corresponding to each type of sub-data at each time node.

[0077] Among them, for each time node in the information to be recommended, the weighted sum can be calculated according to the richness value corresponding to each type of sub-data at that time node and the preset weight corresponding to each type of sub-data, so as to determine the key degree corresponding to that time node.

[0078] If the key degree corresponding to this time node is greater than the preset key degree, it means that this time node is a key time node in the information to be recommended. The server can determine the information points of the element according to the information content corresponding to this time node.

[0079] For example, the server can use the feature vector of the information content corresponding to this time node as the information points of the element. Of course, the information content corresponding to this time node can also be directly used as the information points of the element.

[0080] Of course, the server can also directly input the information to be recommended into the information extraction model without preprocessing the information to be recommended, so as to extract the information points of each element included in the information to be recommended through the information extraction model.

[0081] Before using the above information extraction model, the server can first train the information extraction model.

[0082] Among them, the server can obtain historical recommended information and determine the browsing records of each historical user for the historical recommended information. The browsing record is used to represent the browsing and clicking situations (such as click-through rate, browsing rate, etc.) of each historical user for the information content of each time node in the historical recommended information.

[0083] In other words, the historical recommended information can be a time series of audio-visual data, and each frame of this time series records a browsing record. The time series of the browsing record is aligned with the time series of the audio-visual data on the time scale.

[0084] The server can input the historical recommendation information into the information extraction model to determine the historical key point information contained in the historical information to be recommended through the information extraction model.

[0085] Meanwhile, the server can determine the key time nodes in the historical information to be recommended according to the browsing records.

[0086] Taking the browsing records containing the user's click-through rate and browsing rate as an example, the key time nodes can be the time nodes when the change amount of the browsing rate and / or click-through rate in the browsing records exceeds the preset change amount.

[0087] Of course, the key time nodes can also be the time nodes when the browsing rate and / or click-through rate exceed the first preset threshold (i.e., the time nodes with a relatively high browsing rate and / or click-through rate), and the time nodes when they are lower than the second preset threshold (i.e., the time nodes with a relatively low browsing rate and / or click-through rate). The above preset change amount, the first preset threshold, and the second preset threshold can be set according to the actual situation, and this specification does not make specific limitations on this.

[0088] After determining the key time nodes in the historical recommendation information, the server can determine the information content corresponding to the key time nodes, and take minimizing the deviation between the information content and the historical key point information as the optimization goal to train the information extraction model, so that the key point information extracted by the trained information extraction model is as close as possible to the information content corresponding to the key time nodes that have the greatest impact on the user's browsing situation.

[0089] S103: Input the respective key point information into a pre-trained evaluation model to determine the evaluation value corresponding to each key point information through the evaluation model. For each key point information, the evaluation value corresponding to the key point information is used to represent the contribution degree of the key point information to the expected browsing situation of the information to be recommended.

[0090] After determining the respective key point information contained in the information to be recommended, the server can input the respective key point information into a pre-trained evaluation model, so as to evaluate each key point information through the evaluation model to determine the evaluation values corresponding to the respective key point information. The evaluation value is used to represent the contribution degree of each key point information to the expected browsing situation (such as click-through rate and browsing rate) of the information to be recommended.

[0091] Among them, the greater the contribution degree of the key point information to the browsing situation, the greater the corresponding evaluation value, and vice versa.

[0092] In addition, for different business scenarios, the contribution degrees of different information contents to the browsing situation are different. Take the game scenario as an example. For action games, users may be more inclined to intense and exciting fighting scenes and dynamic music, while for cultivation games, what attracts users may be warm and comfortable game scenes and soothing music.

[0093] Therefore, in the process of evaluating the element point information through the evaluation model, the server can input each element point information and the scenario information of its corresponding application scenario into the evaluation model, so as to determine the evaluation value of each element point information in its corresponding application scenario through the evaluation model, ensuring that the determined evaluation value matches the scenario information of the application scenario.

[0094] Before using the above evaluation model, the server can train the evaluation model.

[0095] Specifically, the server can obtain historical recommendation information and determine the browsing records of each historical user for the historical recommendation information. Then, the server inputs the historical recommendation information into the above information extraction model to extract each element point information contained in the recommendation information through the information extraction model as historical element point information.

[0096] After that, the server can input the historical element point information into the evaluation model to determine the evaluation value corresponding to each historical element point information through the evaluation model. Furthermore, the evaluation model is trained according to the evaluation value corresponding to each historical element point information and the browsing record.

[0097] For example, the server can sort each historical element point information according to the evaluation value corresponding to each element point information to obtain a predicted sorting result. At the same time, based on the actual click-through rate and / or browsing rate at the time node where each historical element point information is located, each historical element point information is sorted to obtain an actual sorting result. Then, the evaluation model is trained with the goal of minimizing the deviation between the predicted sorting result and the actual sorting result.

[0098] For another example, the server can select a specified number of element point information in descending order of the evaluation value as the first element point information, and select the same number of element point information in descending order of the actual click-through rate and / or browsing rate as the second element point information. Then, the evaluation model is trained with the goal of minimizing the deviation between the first element point information and the second element point information.

[0099] S104: Adjust the information to be recommended according to the evaluation value, and perform information recommendation through the adjusted information to be recommended.

[0100] After determining the evaluation values corresponding to each element point information, the server can optimize the information to be recommended according to the evaluation values corresponding to each element point information.

[0101] Specifically, for each element point information, if the evaluation value corresponding to the element point information is lower than the preset threshold, the server can adjust the information content at the time node corresponding to the element point information to optimize the information to be recommended.

[0102] Among them, for each element point information, if the evaluation value corresponding to the element point information is lower than the preset threshold, a number of element point information before the time node corresponding to the element point information and a number of element point information after the time node corresponding to the element point information are used as the context information corresponding to the element point information.

[0103] After that, the information content at the time node corresponding to the element point information can be adjusted according to the context information.

[0104] Of course, the server can also push the element point information with an evaluation value lower than the preset threshold to the production staff in the form of information prompts, so that the production staff can adjust the information to be recommended based on the information prompts.

[0105] In this specification, the information to be recommended can be advertising materials in a specified application scenario. For the convenience of understanding, this specification provides an optimization method for advertising materials, as Figure 2 shown.

[0106] Figure 2 It is a schematic diagram of an optimization method for advertising materials provided in this specification.

[0107] Among them, after the server obtains the advertising materials, it can first preprocess the advertising materials, such as data cleaning and standardization processing, etc. Then, the preprocessed advertising materials are input into the information extraction model to determine the element point information therein. Then, the element point information is input into the evaluation model, and the evaluation model conducts quantitative analysis on each element point information to determine each element point information for quantitative analysis. Then, the advertising materials are optimized based on the evaluation values corresponding to each element point information.

[0108] From the above method, it can be seen that in the process of information recommendation for the information to be recommended in this solution, each element point information included in the information to be recommended can be extracted and evaluated for each element point information. In this way, the information to be recommended can be optimized based on the evaluation values corresponding to each element point information, so that the information to be recommended reaches the best recommendation effect.

[0109] For example, game company A needs to produce a new advertising material, but their team is not clear about which factors may affect the advertising effect. By using this technology, they can input the existing advertising materials into the system, and the system will automatically extract the key points of the advertising materials and conduct quantitative analysis, and finally output the analysis results. Through the analysis results, the team of company A can identify the key factors that affect the advertising effect, so as to optimize these key factors when producing new advertising materials and improve the advertising effect.

[0110] The above is one or more implementation information recommendation methods of this specification. Based on the same idea, this specification also provides a corresponding information recommendation device, such as Figure 3 shown.

[0111] Figure 3 It is a schematic diagram of an information recommendation device provided by this specification, including:

[0112] An acquisition module 301, configured to acquire information to be recommended;

[0113] An extraction module 302, configured to input the information to be recommended into a pre-trained information extraction model, so as to extract, through the information extraction model, each element point information included in the information to be recommended, where the element point information is used to characterize the information content corresponding to the key time nodes in the information to be recommended;

[0114] An evaluation module 303, configured to input the each element point information into a pre-trained evaluation model, so as to determine, through the evaluation model, an evaluation value corresponding to each element point information. For each element point information, the evaluation value corresponding to the element point information is used to characterize the contribution degree of the element point information to the expected browsing situation of the information to be recommended;

[0115] A recommendation module 304, configured to adjust the information to be recommended according to the evaluation value, so as to perform information recommendation through the adjusted information to be recommended.

[0116] Optionally, before the step of inputting the information to be recommended into the pre-trained information extraction model, the acquisition module 301 is further configured to divide the information to be recommended to obtain at least one type of sub-data, and the types include: text, image, audio;

[0117] The extraction module 302 is specifically configured to input the at least one type of sub-data into the information extraction model, so as to extract, through the information extraction model, each element point information included in the information to be recommended.

[0118] Optionally, the extraction module 302 is specifically configured to determine, for each time node in the information to be recommended, the richness value corresponding to each type of sub-data at this time node; and determine the information of each element point included in the information to be recommended according to the richness value corresponding to each type of sub-data at each time node.

[0119] Optionally, the extraction module 302 is specifically configured to determine, for each time node in the information to be recommended, the criticality corresponding to this time node according to the richness value corresponding to each type of sub-data at this time node and the preset weight corresponding to each type of sub-data; if the criticality corresponding to this time node is greater than the preset criticality, then determine the element point information according to the information content corresponding to this time node.

[0120] Optionally, before the step of inputting the information to be recommended into the pre-trained information extraction model, the acquisition module 301 is further configured to process the information to be recommended to obtain a data sequence with a specified number of frames;

[0121] The extraction module 302 is specifically configured to input the data sequence with the specified number of frames into the information extraction model, so as to extract, through the information extraction model, the information of each element point included in the data sequence.

[0122] Optionally, before the step of inputting the information to be recommended into the pre-trained information extraction model, the acquisition module 301 is further configured to determine the application scenario corresponding to the information content of the information to be recommended;

[0123] The evaluation module 303 is specifically configured to input the information of each element point and the scenario information of the application scenario into the evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to each element point information in the application scenario.

[0124] Optionally, the recommendation module 304 is specifically configured to, for each element point information, if the evaluation value corresponding to this element point information is lower than the preset threshold, then use a plurality of element point information before the time node corresponding to this element point information and a plurality of element point information after the time node corresponding to this element point information as the context information corresponding to this element point information; and adjust the information content at the time node corresponding to this element point information according to the context information.

[0125] Optionally, the device further includes:

[0126] A training module 305, configured to obtain historical recommendation information and determine the browsing records of each user for the historical recommendation information in history, where the browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommendation information; input the historical recommendation information into the information extraction model, so as to determine, through the information extraction model, the historical feature point information included in the historical recommendation information, and, according to the browsing records, determine the key time nodes in the historical recommendation information; and train the information extraction model with the optimization goal of minimizing the deviation between the historical feature point information and the information content corresponding to the key time nodes in the historical recommendation information.

[0127] Optionally, the apparatus further includes:

[0128] A training module 305, configured to obtain historical recommendation information and determine the browsing records of each user for the historical recommendation information in history, where the browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommendation information; input the historical recommendation information into a pre-trained information extraction model, so as to extract, through the information extraction model, each piece of feature point information included in the recommendation information as historical feature point information; input the historical feature point information into the evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to each historical feature point information; and train the evaluation model according to the evaluation value corresponding to each historical feature point information and the browsing records.

[0129] This specification also provides a computer-readable storage medium storing a computer program that can be used to execute the above Figure 1 provided information recommendation method.

[0130] This specification also provides Figure 4 a schematic structural diagram of an electronic device corresponding to Figure 1 as shown. As Figure 4 described, at the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile memory. Of course, other hardware required for other services may also be included. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs it to implement the above Figure 1 described information recommendation method. Of course, in addition to the software implementation manner, this specification does not exclude other implementation manners, such as logical devices or a combination of software and hardware, etc. That is to say, the execution subject of the following processing flow is not limited to each logical unit, and may also be hardware or a logical device.

[0131] For an improvement in a technology, it can be clearly distinguished whether it is a hardware improvement (for example, improvements to circuit structures such as diodes, transistors, switches, etc.) or a software improvement (improvements to method processes). However, with the development of technology, many improvements to method processes today can be regarded as direct improvements to hardware circuit structures. Almost all designers obtain the corresponding hardware circuit structure by programming the improved method process into the hardware circuit. Therefore, it cannot be said that an improvement to a method process cannot be implemented with a hardware entity module. For example, a programmable logic device (PLD) (such as a field programmable gate array (FPGA)) is such an integrated circuit whose logical function is determined by the user programming the device. The designer can program by himself to "integrate" a digital system on a PLD, without having to ask a chip manufacturer to design and manufacture a dedicated integrated circuit chip. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, this programming is mostly implemented using "logic compiler" software, which is similar to the software compiler used in program development and writing. The original code before compilation also has to be written in a specific programming language, which is called a hardware description language (HDL), and there is not only one kind of HDL, but many kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), Confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), Lava, Lola, MyHDL, PALASM, RHDL (Ruby Hardware Description Language), etc. Currently, the most commonly used are VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog. Those skilled in the art should also be clear that only by slightly logically programming the method process with the above-mentioned several hardware description languages and programming it into the integrated circuit can the hardware circuit implementing the logical method process be easily obtained.

[0132] The controller can be implemented in any suitable manner. For example, the controller can take the form of, for example, a microprocessor or a processor and a computer-readable medium storing computer-readable program code (such as software or firmware) executable by the (micro)processor, logic gates, switches, an application specific integrated circuit (ASIC), a programmable logic controller, and an embedded microcontroller. Examples of the controller include, but are not limited to, the following microcontrollers: ARC 625D, Atmel AT91SAM, Microchip PIC18F26K20, and Silicone Labs C8051F320. The memory controller can also be implemented as part of the control logic of the memory. Those skilled in the art also know that, in addition to implementing the controller in the form of pure computer-readable program code, it is entirely possible to logically program the method steps to enable the controller to be implemented in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. to achieve the same function. Therefore, such a controller can be considered a hardware component, and the devices included therein for implementing various functions can also be regarded as the structures within the hardware component. Or even, the devices for implementing various functions can be regarded as either software modules for implementing the method or structures within the hardware component.

[0133] The systems, devices, modules, or units illustrated in the above embodiments can be specifically implemented by computer chips or entities, or by products having certain functions. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0134] For the convenience of description, when describing the above devices, they are described separately as various units according to their functions. Of course, when implementing this specification, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0135] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) containing computer-usable program code.

[0136] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to embodiments of the specification. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and combinations of flows and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to the processors of general-purpose computers, special-purpose computers, embedded processors, or other programmable data processing devices to generate a machine, such that the instructions executed by the processors of the computer or other programmable data processing devices produce means for implementing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 one or more of the blocks for implementing the specified functions.

[0137] These computer program instructions can also be stored in a computer-readable memory capable of guiding a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means for implementing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 one or more of the blocks.

[0138] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process, and thus the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more of the flows Figure 1 one or more of the flows and / or blocks Figure 1 one or more of the blocks.

[0139] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0140] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0141] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0142] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0143] Those skilled in the art will appreciate that the embodiments of this specification may be provided as methods, systems or computer program products. Therefore, this specification may take the form of a complete hardware embodiment, a complete software embodiment or an embodiment combining software and hardware. Moreover, this specification may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0144] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0145] The various embodiments in this specification are described in a progressive manner. For the parts that are the same or similar among the various embodiments, reference can be made to each other, and the differences between each embodiment and other embodiments are emphasized. In particular, for system embodiments, since they are basically similar to method embodiments, the description is relatively simple, and reference can be made to the relevant parts of the method embodiments for the relevant parts.

[0146] The above is only the embodiments of this specification and is not intended to limit this specification. For those skilled in the art, various changes and modifications can be made to this specification. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this specification shall be included within the scope of the claims of this specification.

Claims

1. An information recommendation method, characterized in that, Including: Obtain the information to be recommended; Input the information to be recommended into a pre-trained information extraction model, so as to extract, through the information extraction model, each element point information included in the information to be recommended, where the element point information is used to characterize the information content corresponding to the key time node in the information to be recommended, and the key time node is the time node when the change amount of the browsing rate and / or click-through rate exceeds a preset change amount; Input the each element point information into a pre-trained evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to each element point information. For each element point information, the evaluation value corresponding to the element point information is used to characterize the contribution degree of the element point information to the expected browsing situation of the information to be recommended; Adjust the information to be recommended according to the evaluation value, so as to perform information recommendation through the adjusted information to be recommended; Among them, the steps of training the evaluation model include: Obtain historical recommended information and determine the browsing records of each user for the historical recommended information, where the browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommended information; Input the historical recommended information into the information extraction model, so as to extract, through the information extraction model, each element point information included in the recommended information as each historical element point information; Input the each historical element point information into the evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to the each historical element point information; Select a specified number of element point information in descending order of the evaluation values corresponding to the each historical element point information as the first element point information, and select the specified number of element point information in descending order of the actual click-through rate and / or browsing rate as the second element point information, and train the evaluation model with the optimization goal of minimizing the deviation between the first element point information and the second element point information.

2. The method according to claim 1, wherein Before the step of inputting the information to be recommended into the pre-trained information extraction model, the method further includes: Divide the information to be recommended to obtain at least one type of sub-data, and the types include: text, image, audio; The step of inputting the information to be recommended into the pre-trained information extraction model to extract, through the information extraction model, each element point information included in the information to be recommended specifically includes: Input the at least one type of sub-data into the information extraction model, so as to extract, through the information extraction model, each element point information included in the information to be recommended.

3. The method according to claim 2, wherein The step of extracting, through the information extraction model, each element point information included in the information to be recommended specifically includes: For each time node in the information to be recommended, determine the richness value corresponding to each type of sub-data at this time node; Determine each element point information included in the information to be recommended according to the richness value corresponding to each type of sub-data at each time node.

4. The method according to claim 3, characterized in that, The step of determining each element point information included in the information to be recommended according to the richness value corresponding to each type of sub-data at each time node specifically includes: For each time node in the information to be recommended, determine the key degree corresponding to this time node according to the richness value corresponding to each type of sub-data at this time node and the preset weight corresponding to each type of sub-data. If the key degree corresponding to this time node is greater than the preset key degree, determine the element point information according to the information content corresponding to this time node.

5. The method according to claim 1, characterized in that, Before the step of inputting the information to be recommended into a pre-trained information extraction model, the method further includes: Process the information to be recommended to obtain a data sequence with a specified number of frames. The step of inputting the information to be recommended into a pre-trained information extraction model to extract each element point information included in the information to be recommended through the information extraction model specifically includes: Input the data sequence with the specified number of frames into the information extraction model to extract each element point information included in the data sequence through the information extraction model.

6. The method according to claim 1, characterized in that Before the step of inputting the information to be recommended into a pre-trained information extraction model, it further includes: Determine the application scenario corresponding to the information content of the information to be recommended. The step of inputting each element point information into a pre-trained evaluation model to determine the evaluation value corresponding to each element point information through the evaluation model specifically includes: Input each element point information and the scenario information of the application scenario into the evaluation model to determine the evaluation value corresponding to each element point information in the application scenario through the evaluation model.

7. The method according to claim 1, characterized in that The step of optimizing the information to be recommended according to the evaluation value specifically includes: For each element point information, if the evaluation value corresponding to this element point information is lower than the preset threshold, then use several element point information before the time node corresponding to this element point information and several element point information after the time node corresponding to this element point information as the context information corresponding to this element point information. Adjust the information content at the time node corresponding to this element point information according to the context information.

8. The method according to claim 1, characterized in that, The step of training the information extraction model includes: Obtain historical recommended information and determine the browsing records of each user for the historical recommended information. The browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommended information. Input the historical recommended information into the information extraction model to determine the historical element point information included in the historical recommended information through the information extraction model, and determine the key time nodes in the historical recommended information according to the browsing records. Train the information extraction model with the optimization goal of minimizing the deviation between the historical element point information and the information content corresponding to the key time nodes in the historical recommended information.

9. An information recommendation device, characterized in that, Includes: An acquisition module for acquiring the information to be recommended. An extraction module, configured to input the information to be recommended into a pre-trained information extraction model, so as to extract, through the information extraction model, the information of each key point included in the information to be recommended, where the information of the key point is used to characterize the information content corresponding to the key time node in the information to be recommended, and the key time node is a time node when the change amount of the browsing rate and / or click-through rate exceeds a preset change amount; An evaluation module, configured to input the information of each key point into a pre-trained evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to each key point information. For each key point information, the evaluation value corresponding to the key point information is used to characterize the contribution degree of the key point information to the expected browsing situation of the information to be recommended; A recommendation module, configured to adjust the information to be recommended according to the evaluation value, so as to perform information recommendation through the adjusted information to be recommended; Wherein, the step of training the evaluation model includes: Obtaining historical recommended information and determining the browsing records of each user for the historical recommended information, where the browsing records are used to represent the browsing and clicking situations of each user for the information content at each time node in the historical recommended information; Inputting the historical recommended information into the information extraction model, so as to extract, through the information extraction model, the information of each key point included in the recommended information as each historical key point information; Inputting the information of each historical key point into the evaluation model, so as to determine, through the evaluation model, the evaluation value corresponding to the information of each historical key point; Selecting a specified number of key point information in descending order of the evaluation values corresponding to the information of each historical key point as the first key point information, and selecting the specified number of key point information in descending order of the actual click-through rate and / or browsing rate as the second key point information, and training the evaluation model with the optimization objective of minimizing the deviation between the first key point information and the second key point information.

10. A computer-readable storage medium, characterized in that, The storage medium stores a computer program, and when the computer program is executed by a processor, the method according to any one of claims 1 to 8 above is implemented.

11. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, the method according to any one of claims 1 to 8 above is implemented.

Images