Method and apparatus for outputting information

Abstract

The present disclosure provides a method and apparatus for outputting information, relating to the field of artificial intelligence technology, and in particular, to the field of intelligent search. A specific implementation scheme comprises: obtaining a historical set of recommended information, a user's historical click records, and a user profile; selecting recommended information from the recommended information set to form at least one recommended information segment; extracting features of each recommended information segment from the at least one recommended information segment; generating a user representation vector based on the historical click records and user profile; and predicting the click-through rate of each recommended information segment based on the user representation vector and the features of the recommended information segment. This implementation scheme results in richer and more diverse recalled resources.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence technology, in particular to the field of intelligent search, and specifically to a method and device for outputting information. Background Art

[0002] Information push refers to the process of filtering out content that users may be interested in from an information database, and then presenting the filtered content to the user. In the process of information push, it is usually necessary to first recall a large amount of information, and then perform a rough sorting and screening on the recalled information, that is, to preliminarily screen out the information that users may be interested in, and then further perform a fine sorting and screening on the information obtained by the rough sorting and screening, and finally obtain the information set pushed to the user, so that the pushed information set fits the user's interests. In the related art, a single recommended information is recalled according to a unified sorting (rough sorting + fine sorting), and each time the resources of the top interest points are recalled, the head clustering phenomenon is relatively serious, the diversity is poor, and after multi-layer funnel filtering, there are fewer valid candidate sets, which is not conducive to improving the recommendation effect. Summary of the Invention

[0003] The present disclosure provides a method, apparatus, device, storage medium, and computer program product for outputting information.

[0004] According to a first aspect of the present disclosure, a method for outputting information is provided, comprising: obtaining a historical recommendation information set, a user's historical click records, and a user portrait; selecting recommendation information from the recommendation information set to form at least one recommendation information segment; extracting features of each recommendation information segment from the at least one recommendation information segment; generating a user representation vector based on the historical click records and the user portrait; and for each recommendation information segment, predicting the click-through rate of the recommendation information segment based on the user representation vector and the features of the recommendation information segment.

[0005] According to a second aspect of the present disclosure, a device for outputting information is provided, comprising: an acquisition unit configured to acquire a historical recommendation information set, a user's historical click records and a user portrait; a selection unit configured to select recommendation information from the recommendation information set to form at least one recommendation information segment; an extraction unit configured to extract features of each recommendation information segment from the at least one recommendation information segment; a generation unit configured to generate a user representation vector based on the historical click records and the user portrait; and a prediction unit configured to predict, for each recommendation information segment, the click-through rate of the recommendation information segment based on the user representation vector and the features of the recommendation information segment.

[0006] According to a third aspect of the present disclosure, an electronic device is provided, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method described in the first aspect.

[0007] According to a fourth aspect of the present disclosure, a non-transitory computer-readable storage medium storing computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the method described in the first aspect.

[0008] According to a fifth aspect of the present disclosure, a computer program product is provided, comprising a computer program, wherein the computer program implements the method described in the first aspect when executed by a processor.

[0009] The method and apparatus for outputting information provided by the embodiments of the present disclosure consider the relationship between sequences and recall resources by fragments (multiple items (news and other recommended information, such as short videos or pictures and texts)). This can solve the problem of head resource aggregation, play the role of high frequency driving low frequency, and the recalled resources are richer and more diverse.

[0010] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present disclosure, nor are they intended to limit the scope of the present disclosure. Other features of the present disclosure will become readily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The accompanying drawings are provided to facilitate a better understanding of the present invention and do not constitute a limitation of the present disclosure.

[0012] Figure 1 is an exemplary system architecture diagram in which an embodiment of the present disclosure may be applied;

[0013] Figure 2 is a flow chart of an embodiment of a method for outputting information according to the present disclosure;

[0014] Figure 3 is a schematic diagram of an application scenario of the method for outputting information according to the present disclosure;

[0015] Figure 4 is a flowchart of yet another embodiment of a method for outputting information according to the present disclosure;

[0016] Figure 5 is a structural diagram of an embodiment of a device for outputting information according to the present disclosure;

[0017] Figure 6It is a schematic diagram of the structure of a computer system of an electronic device suitable for implementing the embodiments of the present disclosure. DETAILED DESCRIPTION

[0018] The following description of exemplary embodiments of the present disclosure is made in conjunction with the accompanying drawings, including various details of the embodiments of the present disclosure to facilitate understanding. These details should be considered as merely exemplary. Therefore, those skilled in the art will recognize that various changes and modifications may be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. Similarly, for the sake of clarity and conciseness, descriptions of well-known functions and structures are omitted in the following description.

[0019] Figure 1 An exemplary system architecture 100 is shown to which embodiments of the method for outputting information or the apparatus for outputting information disclosed herein may be applied.

[0020] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, 103, a network 104, and a server 105. Network 104 is a medium for providing communication links between terminal devices 101, 102, 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links or fiber optic cables.

[0021] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0022] Terminal devices 101, 102, and 103 can be hardware or software. When terminal devices 101, 102, and 103 are hardware, they can be various electronic devices with display screens and support web browsing, including but not limited to smart phones, tablet computers, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III, Moving Picture Experts Group Audio Layer 3), MP4 (Moving Picture Experts Group Audio Layer IV, Moving Picture Experts Group Audio Layer 4) players, laptop computers, and desktop computers, etc. When terminal devices 101, 102, and 103 are software, they can be installed in the electronic devices listed above. It can be implemented as multiple software or software modules (for example, to provide distributed services), or it can be implemented as a single software or software module. No specific limitation is made here.

[0023] The server 105 may be a server that provides various services, such as a backend web server that supports web pages displayed on the terminal devices 101, 102, and 103. The backend web server may analyze and process received web page requests and other data, and feed back the processing results (e.g., web page data) to the terminal device.

[0024] It should be noted that a server can be either hardware or software. When a server is hardware, it can be implemented as a distributed server cluster consisting of multiple servers, or as a single server. When a server is software, it can be implemented as multiple software programs or software modules (for example, multiple software programs or software modules used to provide distributed services), or as a single software program or software module. This is not specifically limited here. The server can also be a server in a distributed system, or a server integrated with blockchain. The server can also be a cloud server, or an intelligent cloud computing server or intelligent cloud host with artificial intelligence technology.

[0025] It should be noted that the method for outputting information provided in the embodiments of the present disclosure is generally executed by the server 105 , and accordingly, the device for outputting information is generally provided in the server 105 .

[0026] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is merely illustrative. Any number of terminal devices, networks and servers may be provided as required.

[0027] Continue to refer Figure 2 , shows a process 200 of an embodiment of a method for outputting information according to the present disclosure. The method for outputting information includes the following steps:

[0028] Step 201: Obtain a historical recommendation information set, a user's historical click records, and a user portrait.

[0029] In this embodiment, the execution subject of the method for outputting information (eg Figure 1 The server shown in the figure can receive browsing requests from the terminal used by the user to browse the web page through a wired connection or a wireless connection. When the user opens a browser or other application, a browsing request is sent to the server. If the user is logged in, the user account can be obtained and the historical click records and user profile of the user account can be obtained. The user profile can be obtained through the user registration information, such as name, age, occupation, interests, etc. The user profile can also be obtained by analyzing the historical click records. The historical click records may include information such as the subject, tag, click time, etc. of the recommended information clicked by the user.

[0030] The server also retrieves a collection of historical recommendation information, i.e., information that has been previously recommended. This collection of recommendation information can be sorted by click time, for example, with the most recently clicked information at the top. Optionally, the collection of historical recommendation information can be filtered based on click-through rate, filtering out recommendations with a click-through rate below a predetermined threshold.

[0031] Step 202: Select recommended information from the recommended information set to form at least one recommended information segment.

[0032] In this embodiment, multiple recommendation information can be grouped into a recommendation information segment. Several recommendation information can be selected in chronological order to form a recommendation information segment. For example, every three recommendation information can form a recommendation information segment. Different recommendation information segments can contain duplicate recommendation information, or they can contain no duplicate recommendation information.

[0033] Step 203: extract features of each recommendation information segment from at least one recommendation information segment.

[0034] In this embodiment, the features of each recommendation information segment can be extracted through the existing feature extraction network. The input of the feature extraction network is multiple recommendation information, and each recommendation information of a recommendation information segment can be encoded and converted into a sub-feature vector. All the obtained sub-feature vectors are spliced together in the order of the recommendation information, and the splicing result is recorded as the intermediate feature vector. Afterwards, the multi-head self-attention mechanism is used to perform multiple self-attention calculations on the intermediate feature vector. Each self-attention calculation corresponds to a head in the multi-head self-attention mechanism, and an attention parameter is obtained. Therefore, the intermediate feature vector corresponds to multiple attention parameters. The calculated multiple attention parameters are spliced together in sequence to obtain a splicing result, and then the splicing result is multiplied by the overall parameter trained in the multi-self-attention mechanism. The product obtained is the feature of the recommendation information segment.

[0035] Step 204: Generate a user representation vector based on historical click records and user portraits.

[0036] In this embodiment, historical features can be extracted from historical click records using a long short-term memory (LSTM) network. User features can be extracted from user profiles using a convolutional network. The historical features and user features are concatenated to form a user representation vector.

[0037] Step 205 : For each recommended information segment, based on the user representation vector and the features of the recommended information segment, predict the click-through rate of the recommended information segment.

[0038] In this embodiment, the click-through rate (CTR) of the recommended information segment can be predicted using an existing CTR (Click Through Rate) model. CTR is widely used in information recommendation, particularly in advertising recommendations. CTR refers to the click-through rate (CTR) of online ads (such as image ads, text ads, keyword ads, ranked ads, and video ads), which is the actual number of clicks on an ad divided by the number of impressions.

[0039] The CTR model is trained on a large amount of historical data by comprehensively considering various factors and features. Training samples for CTR estimation can be obtained from historical logs and offline feature libraries. Sample labeling is relatively simple: a user click is marked as 1, and no click is marked as 0. Numerous features are considered, such as user representation vectors generated based on historical click records and user profiles, and features of recommended information snippets.

[0040] The method provided by the above embodiment of the present disclosure takes into account diversity factors and adopts a fragment (multiple items) method for recall. The recalled resources are generated by a combination of historical sequences. The idea of sequence modeling in natural language processing is applied to the recommendation system, and the attention mechanism is introduced to learn the relationship between the sequences of resources. Not only the relationship between the user and the recommended information is considered, but also the relationship between the recommended information and the recommended information. The advantage of doing so is that the distribution of recalled resources can be more even, avoiding the phenomenon of head aggregation of points of interest.

[0041] In some optional implementations of this embodiment, recommended information is selected from a recommended information set to form at least one recommended information segment, including: obtaining a predetermined number of recommended information as recommended information segments from the recommended information set through a sliding time window to obtain at least one recommended information segment. A predetermined number of recommended information is obtained each time as a recommended information segment with a predetermined step size and window length. The sliding time window slides to select recommended information in the order in which the recommended information is clicked. If the window length is greater than the step size, adjacent recommended information segments partially overlap. For example, the window length is 5 and the step size is 3, so that the 5 recommended information obtained each time are regarded as a recommended information segment, which partially overlaps with adjacent recommended information segments and can ensure the time order. If adjacent recommended information segments do not need to overlap, the window length can be set to be less than or equal to the step size.

[0042] In some optional implementations of this embodiment, the features of each recommendation information segment are extracted from at least one recommendation information segment, including: for each recommendation information segment in the at least one recommendation information segment, the recommendation information segment and the position offset of each recommendation information in the recommendation information segment are input into a self-attention network to obtain the features of the recommendation information segment. The position offset is the relative position of each recommendation information in a recommendation information segment. For example, there are three recommendation information in the recommendation information segment, and the order of click time is A, B, and C. Then the position offset of A is 0, the position offset of B is 1, and the position offset of C is 2. The position information of each recommendation information is artificially added to the input sequence, so that the same recommendation information at different positions in the input sequence will have syntactic and semantic differences for the self-attention network, and they will produce different outputs. In this way, the relationship between recommended information is learned, playing the role of high frequency driving low frequency.

[0043] In some optional implementations of this embodiment, a user representation vector is generated based on historical click records and user profiles, including: inputting historical click records into a pre-trained gated recurrent unit to obtain historical features; inputting user profiles into a fully connected layer to obtain user features; and fully connecting historical features and user features to obtain a user representation vector. The gated recurrent unit (GRU) is a variant of the long short-term memory (LSTM) network that combines the forget gate and the input gate into a single update gate. The GRU solves the problem of long-term memory and gradients in backpropagation.

[0044] In some optional implementations of this embodiment, the method further includes filtering out recommended information that does not meet the constraints from a predetermined number of recommended information segments with the highest click-through rates, obtaining a first candidate recommended information set for output. The constraints may be, for example, deduplication, diversity, etc., thereby improving the hit rate of recommendations. When combining and issuing lists using segments, considering hard constraints and diversity, some segments that do not meet the constraints will be discarded. For example, if a segment has three items and one of the items does not meet the constraints, one of the items will be discarded and the other two items will be retained.

[0045] Continue to see Figure 3 , Figure 3 FIG. 1 is a schematic diagram of an application scenario of the method for outputting information according to this embodiment. Figure 3In the application scenario, when the server detects that the user has opened the news application, it obtains the user's historical click records news_0, news_1...news_n. It also obtains the user's user model. The historical click records are then encoded into vectors h(1), h(2)...h(n) and input into GRU for feature extraction to obtain the historical feature LAST. The user model is encoded into vector h(0) and fully connected to obtain the user features. The historical features and user features are fully connected to obtain the user representation vector. The database can pre-store the recommended information segments generated by the sliding window combination of the historical sequence. The database also stores the features of the recommended information segments extracted by the self-attention network based on position offset. The user representation vector is sequentially compared with the features of the recommended information segments in the database to predict the user's click-through rate for each recommended information segment. The recommended information in the recommended information segments with a predetermined number of click-through rates is then deduplicated and pushed to the user.

[0046] Further references Figure 4 , which shows a process 400 of another embodiment of a method for outputting information. The process 400 of the method for outputting information includes the following steps:

[0047] Step 401: Obtain a historical recommendation information set, a user's historical click records, and a user portrait.

[0048] Step 402: Select recommended information from the recommended information set to form at least one recommended information segment.

[0049] Step 403: extract features of each recommendation information segment from at least one recommendation information segment.

[0050] Step 404: Generate a user representation vector based on historical click records and user portraits.

[0051] Step 405 : For each recommended information segment, predict the click-through rate of the recommended information segment based on the user representation vector and the features of the recommended information segment.

[0052] Steps 401-405 are substantially the same as steps 201-205, and therefore are not described in detail.

[0053] Step 406 : Filter out recommended information that does not meet the constraint conditions from a predetermined number of recommended information segments with the highest click-through rates to obtain a first candidate recommended information set.

[0054] In this embodiment, constraints can include deduplication and diversity, thereby improving the recommendation hit rate. When combining and delivering lists using segments, considering hard constraints and diversity, some segments that do not meet the constraints will be discarded. For example, if a segment has three items and one of them does not meet the constraints, one of them will be discarded, while the other two resources will be retained. The filtered recommendation information segments can still be used as a whole for subsequent scoring.

[0055] Step 407: Generate a second candidate recommendation information set based on the recommendation information set through the recall model.

[0056] In this embodiment, a common recall model in the prior art (eg, contexDNN (contextual deep learning network), etc.) may be used to recall a second candidate recommendation information set consisting of a single item from the recommendation information set.

[0057] Step 408: Input the first candidate recommendation information set and the second candidate recommendation information set into an evaluation model to obtain an evaluation score for each recommendation information.

[0058] In this embodiment, the evaluation model can be a pre-trained neural network that can score based on various features of the recommended information. The first candidate recommended information set is the combined result of multiple items, and the second candidate recommended information set is the result of a single item. Each recommended information segment in the first candidate recommended information set can be scored as a whole. For example, the average score of each recommended information in the recommended information segment can be used as the evaluation score of the recommended information segment to obtain a segment evaluation score. The recommended information in the second candidate recommended information set is scored separately, and the evaluation score of a single item is obtained.

[0059] Step 409: Output target recommendation information according to the evaluation score.

[0060] In this embodiment, recommendations with evaluation scores above a predetermined threshold can be output, or a specified number of recommendations can be selected in descending order of evaluation scores. Recommendations belonging to the same recommendation information segment can be bundled and output. Alternatively, if recommendations in the second candidate recommendation information set are likely to overlap with recommendations in the first candidate recommendation information set, duplicate removal can be performed before outputting the recommendations.

[0061] from Figure 4 It can be seen that Figure 2Compared to the corresponding embodiment, the process 400 of the method for outputting information in this embodiment embodies the integration of the steps of bundled push of multiple recommended information and independent recommendation of a single recommended information. As a result, the solution described in this embodiment can introduce more effective recommendation information, thereby achieving more comprehensive recommendation information selection and more effective information recommendation.

[0062] Further references Figure 5 As an implementation of the methods shown in the above figures, the present disclosure provides an embodiment of a device for outputting information. Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0063] like Figure 5 As shown, the apparatus 500 for outputting information in this embodiment includes: an acquisition unit 501, a selection unit 502, an extraction unit 503, a generation unit 504, and a prediction unit 505. The acquisition unit 501 is configured to acquire a historical set of recommendation information, a user's historical click records, and a user profile; the selection unit 502 is configured to select recommendation information from the recommendation information set to form at least one recommendation information segment; the extraction unit 503 is configured to extract features of each recommendation information segment from the at least one recommendation information segment; the generation unit 504 is configured to generate a user representation vector based on the historical click records and the user profile; and the prediction unit 505 is configured to predict the click-through rate of each recommendation information segment based on the user representation vector and the features of the recommendation information segment.

[0064] In this embodiment, the specific processing of the acquisition unit 501, the selection unit 502, the extraction unit 503, the generation unit 504 and the prediction unit 505 of the device for outputting information 500 can be referred to. Figure 2 These correspond to step 201, step 202, step 203, step 204 and step 205 in the embodiment.

[0065] In some optional implementations of this embodiment, the selection unit 502 is further configured to: obtain a predetermined number of recommendation information as recommendation information segments from the recommendation information set through a sliding time window to obtain at least one recommendation information segment.

[0066] In some optional implementations of this embodiment, the sliding time window slides to select recommended information in the order in which the recommended information is clicked, and the window length is greater than the step length.

[0067] In some optional implementations of this embodiment, the extraction unit 503 is further configured to: for each recommended information segment in at least one recommended information segment, input the recommended information segment and the position offset of each recommended information in the recommended information segment into the self-attention network to obtain the features of the recommended information segment.

[0068] In some optional implementations of this embodiment, the generation unit 504 is further configured to: input historical click records into a pre-trained gated recurrent unit to obtain historical features; input user portraits into a fully connected layer to obtain user features; and fully connect historical features and user features to obtain a user representation vector.

[0069] In some optional implementations of this embodiment, the device 500 also includes a filtering unit (not shown in the drawings), which is configured to: filter out the recommended information that does not meet the constraint conditions from a predetermined number of recommended information segments with the highest click-through rate, and obtain a first candidate recommended information set for output.

[0070] In some optional implementations of this embodiment, the device 500 also includes a fusion unit (not shown in the drawings), which is configured to: generate a second candidate recommendation information set through a recall model based on the recommendation information set; input the first candidate recommendation information set and the second candidate recommendation information set into the evaluation model to obtain an evaluation score for each recommendation information; and output the target recommendation information according to the evaluation score.

[0071] In the technical solutions disclosed herein, the collection, storage, use, processing, transmission, provision and disclosure of user personal information involved comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0072] According to an embodiment of the present disclosure, the present disclosure also provides an electronic device, a readable storage medium, and a computer program product.

[0073] An electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the method described in process 200 or 400.

[0074] A non-transitory computer-readable storage medium storing computer instructions, wherein the computer instructions are used to cause the computer to execute the method described in process 200 or 400.

[0075] A computer program product includes a computer program, wherein the computer program implements the method described in flow 200 or 400 when executed by a processor.

[0076] Figure 6 A schematic block diagram of an example electronic device 600 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices, and other similar computing devices. The components shown herein, their connections and relationships, and their functions are provided as examples only and are not intended to limit the implementation of the present disclosure described and / or claimed herein.

[0077] like Figure 6 As shown, the device 600 includes a computing unit 601, which can perform various appropriate actions and processes according to a computer program stored in a read-only memory (ROM) 602 or a computer program loaded from a storage unit 608 into a random access memory (RAM) 603. Various programs and data required for the operation of the device 600 can also be stored in the RAM 603. The computing unit 601, the ROM 602, and the RAM 603 are connected to each other via a bus 604. An input / output (I / O) interface 605 is also connected to the bus 604.

[0078] Various components in device 600 are connected to I / O interface 605, including an input unit 606, such as a keyboard, mouse, etc.; an output unit 607, such as various types of displays, speakers, etc.; a storage unit 608, such as a magnetic disk, optical disk, etc.; and a communication unit 609, such as a network card, modem, wireless communication transceiver, etc. The communication unit 609 allows device 600 to exchange information / data with other devices via a computer network such as the Internet and / or various telecommunication networks.

[0079] The computing unit 601 can be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the computing unit 601 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various dedicated artificial intelligence (AI) computing chips, various computing units that run machine learning model algorithms, digital signal processors (DSPs), and any appropriate processors, controllers, microcontrollers, etc. The computing unit 601 performs the various methods and processes described above, such as the method for outputting information. For example, in some embodiments, the method for outputting information can be implemented as a computer software program that is tangibly contained in a machine-readable medium, such as a storage unit 608. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 600 via the ROM 602 and / or the communication unit 609. When the computer program is loaded into the RAM 603 and executed by the computing unit 601, one or more steps of the method for outputting information described above can be performed. Alternatively, in other embodiments, the computing unit 601 can be configured to perform the method for outputting information by any other appropriate means (e.g., by means of firmware).

[0080] Various embodiments of the systems and techniques described herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), system-on-chip systems (SOCs), programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include being implemented in one or more computer programs that are executable and / or interpreted on a programmable system that includes at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0081] The program code for implementing the method of the present disclosure can be written in any combination of one or more programming languages. These program codes can be provided to a processor or controller of a general-purpose computer, a special-purpose computer, or other programmable data processing device so that when the program code is executed by the processor or controller, the functions / operations specified in the flow chart and / or block diagram are implemented. The program code can be executed entirely on the machine, partially on the machine, as a stand-alone software package, partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0082] In the context of the present disclosure, a machine-readable medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or equipment, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium can include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0083] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user can provide input to the computer. Other types of devices can also be used to provide interaction with the user; for example, the feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including acoustic input, voice input, or tactile input).

[0084] The systems and techniques described herein can be implemented in a computing system that includes back-end components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes front-end components (e.g., a user computer having a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such back-end components, middleware components, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), and the Internet.

[0085] A computer system may include a client and a server. The client and server are generally remote from each other and typically interact through a communication network. The client-server relationship arises through computer programs running on the respective computers and having a client-server relationship with each other. The server may be a cloud server, a server in a distributed system, or a server integrated with a blockchain.

[0086] It should be understood that the various forms of the processes shown above can be used to reorder, add, or delete steps. For example, the steps described in this disclosure can be performed in parallel, sequentially, or in a different order, as long as the desired results of the technical solutions disclosed in this disclosure can be achieved. This is not a limitation herein.

[0087] The above specific embodiments do not constitute a limitation on the scope of protection of this disclosure. Those skilled in the art will appreciate that various modifications, combinations, sub-combinations, and substitutions may be made based on design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this disclosure shall be included within the scope of protection of this disclosure.

Claims

1. A method for outputting information, comprising: Obtain historical recommendation information sets, user click history records, and user portraits; Selecting recommended information from the recommended information set to form at least one recommended information segment; Extracting features of each recommendation information segment from the at least one recommendation information segment includes: for each recommendation information segment in the at least one recommendation information segment, inputting the recommendation information segment and the position offset of each recommendation information in the recommendation information segment into a self-attention network to obtain features of the recommendation information segment; Generate a user representation vector based on the historical click records and the user portrait; For each recommended information segment, a click-through rate of the recommended information segment is predicted based on the user representation vector and the features of the recommended information segment.

2. The method according to claim 1, wherein The selecting recommended information from the recommended information set to form at least one recommended information segment includes: A predetermined number of recommendation information is obtained from the recommendation information set through a sliding time window as recommendation information segments, to obtain at least one recommendation information segment.

3. The method according to claim 2, wherein: The sliding time window slides to select recommended information in the order of when the recommended information is clicked, and the window length is greater than the step length.

4. The method according to claim 1, wherein Generating a user representation vector based on the historical click records and the user portrait includes: Inputting the historical click records into a pre-trained gated recurrent unit to obtain historical features; Input the user profile into the fully connected layer to obtain user features; The historical features and the user features are fully connected to obtain a user representation vector.

5. The method according to claim 1, wherein The method further comprises: Recommended information that does not meet the constraint conditions is filtered out from a predetermined number of recommended information segments with the highest click-through rates, and a first candidate recommended information set is obtained and outputted.

6. The method according to claim 5, wherein: The method further comprises: generating a second candidate recommendation information set based on the recommendation information set through a recall model; Inputting the first candidate recommendation information set and the second candidate recommendation information set into an evaluation model to obtain an evaluation score for each recommendation information; Output target recommendation information according to the evaluation score.

7. A device for outputting information, comprising: An acquisition unit configured to acquire a historical set of recommendation information, a user's historical click records, and a user profile; a selection unit, configured to select recommendation information from the recommendation information set to form at least one recommendation information segment; an extraction unit configured to extract features of each recommendation information segment from the at least one recommendation information segment, comprising: for each recommendation information segment in the at least one recommendation information segment, inputting the recommendation information segment and the position offset of each recommendation information in the recommendation information segment into a self-attention network to obtain the features of the recommendation information segment; a generating unit configured to generate a user representation vector based on the historical click records and the user portrait; The prediction unit is configured to predict, for each recommended information segment, a click-through rate of the recommended information segment based on the user representation vector and features of the recommended information segment.

8. The device according to claim 7, wherein The selection unit is further configured to: A predetermined number of recommendation information is obtained from the recommendation information set through a sliding time window as recommendation information segments, to obtain at least one recommendation information segment.

9. The device according to claim 8, wherein The sliding time window slides to select recommended information in the order of when the recommended information is clicked, and the window length is greater than the step length.

10. The device according to claim 7, wherein The generating unit is further configured to: Inputting the historical click records into a pre-trained gated recurrent unit to obtain historical features; Input the user profile into the fully connected layer to obtain user features; The historical features and the user features are fully connected to obtain a user representation vector.

11. The device according to claim 7, wherein The device further comprises a filtering unit configured to: Recommended information that does not meet the constraint conditions is filtered out from a predetermined number of recommended information segments with the highest click-through rates, and a first candidate recommended information set is obtained and outputted.

12. The device according to claim 11, wherein The apparatus further comprises a fusion unit configured to: generating a second candidate recommendation information set based on the recommendation information set through a recall model; Inputting the first candidate recommendation information set and the second candidate recommendation information set into an evaluation model to obtain an evaluation score for each recommendation information; Output target recommendation information according to the evaluation score.

13. An electronic device comprising: at least one processor; as well as a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the method according to any one of claims 1 to 6.

14. A non-transitory computer-readable storage medium storing computer instructions, wherein: The computer instructions are used to cause the computer to execute the method according to any one of claims 1 to 6.

15. A computer program product comprising a computer program, which, when executed by a processor, implements the method according to any one of claims 1 to 6.

Images