Method, electronic device and computer program product for recommending content

Abstract

Embodiments according to this disclosure relate to methods, electronic devices and computer program products for recommending content. A method for recommending content includes determining a similarity between a first recommendation result and a second recommendation result for a content set. The first recommendation result and the second recommendation result are determined based on different recommendation techniques and respectively indicate a degree of recommendation for each content in the content set. The method further includes adjusting the second recommendation result using the similarity. Moreover, the method further includes determining a target recommendation result for the content set based on the first recommendation result and the adjusted second recommendation result. In this way, the accuracy and stability of the finally obtained recommendation result can be improved.

Description

Technical Field

[0001] Embodiments of this disclosure relate to the field of content recommendation technology, and more specifically, to methods, electronic devices, and computer program products for recommending content. Background Technology

[0002] With the development of internet technology, the internet can provide users with an increasing number of online services. For example, users can watch videos, listen to music, read articles, and shop online. Typically, users can use the search function to find the content they need on internet platforms. At the same time, to facilitate users obtaining targeted information, internet platforms can also proactively recommend content to them. With the explosive growth of information on the internet, content recommendation has become a hot topic. Summary of the Invention

[0003] According to an example embodiment of this disclosure, a scheme for recommending content is provided. This scheme determines a final recommendation result based on multiple recommendation results generated by various different recommendation technologies, and utilizes the correlation between these recommendation results by considering their similarity during the determination process. Compared to known conventional schemes, the scheme according to embodiments of this disclosure can strengthen the influence of several recommendation results with high similarity on the final recommendation result, thereby improving the accuracy and stability of the final recommendation result and enhancing its quality.

[0004] In a first aspect of this disclosure, a method for recommending content is provided. The method includes determining a similarity between a first recommendation result and a second recommendation result for a set of content. The first and second recommendation results are determined based on different recommendation techniques and respectively indicate the degree of recommendation for each piece of content in the set. The method also includes adjusting the second recommendation result using the similarity. Furthermore, the method includes determining a target recommendation result for the set of content based on the first recommendation result and the adjusted second recommendation result.

[0005] In a second aspect of this disclosure, an electronic device is provided. The electronic device includes a processor and a memory coupled to the processor, the memory having instructions stored therein, the instructions causing the device to perform actions when executed by the processor. The actions include determining a similarity between a first recommendation result and a second recommendation result for a content set. The first recommendation result and the second recommendation result are determined based on different recommendation techniques and respectively indicate the degree of recommendation for each piece of content in the content set. The actions also include adjusting the second recommendation result using the similarity. Furthermore, the actions include determining a target recommendation result for the content set based on the first recommendation result and the adjusted second recommendation result.

[0006] In a third aspect of this disclosure, a computer program product is provided, which is tangibly stored on a computer-readable medium and includes machine-executable instructions that, when executed, cause a machine to perform the method according to the first aspect.

[0007] The summary section is provided to present the chosen concepts in a simplified form, which will be further described in the detailed description below. The summary section is not intended to identify key or principal features of this disclosure, nor is it intended to limit the scope of this disclosure. Attached Figure Description

[0008] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein:

[0009] Figure 1 A block diagram of an example environment according to some embodiments of the present disclosure is shown;

[0010] Figure 2 A schematic diagram illustrating a process for determining target recommendation results according to some embodiments of the present disclosure is shown;

[0011] Figure 3 A schematic diagram illustrating a process for determining a first optimized recommendation result according to some embodiments of the present disclosure is shown;

[0012] Figure 4 A flowchart of a method for recommending content according to some embodiments of this disclosure is shown; and

[0013] Figure 5 A schematic block diagram of an example device that can be used to implement some embodiments of the present disclosure is shown. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] In the description of embodiments of this disclosure, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "the embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects. Other explicit and implicit definitions may also be included below.

[0016] As mentioned above, content recommendation has become a hot topic in current internet technology. Traditional content recommendation can rely on a specific recommendation technology to obtain recommendations. However, because different conventional recommendation technologies have their own strengths in terms of evaluation metrics such as accuracy and stability, such solutions cannot simultaneously address all evaluation metrics, resulting in mediocre recommendation quality.

[0017] Another known approach uses multiple recommendation techniques to obtain several recommendations, and then simply weights these recommendations to arrive at a final recommendation. Because it only performs a simple weighting without analyzing and utilizing the relationships between the recommendations, the quality of the final recommendation may still be relatively average.

[0018] Embodiments of this disclosure provide a scheme for recommending content. According to various embodiments of this disclosure, multiple recommendation results are integrated by considering the similarity between multiple recommendation results obtained using various different recommendation techniques to determine a final recommendation result.

[0019] As will be understood from the following description, compared with known conventional solutions, the solution according to the embodiments of this disclosure, in the process of integrating multiple recommendation results, utilizes the correlation between the various recommendation results by considering the similarity between these recommendation results to strengthen the influence of several similar recommendation results on the final recommendation result, thereby making the obtained final recommendation result more accurate and more stable.

[0020] The following description will continue with reference to the accompanying drawings, which will provide some exemplary embodiments of this disclosure.

[0021] Figure 1 A block diagram of an example environment 100 according to some embodiments of the present disclosure is shown. Figure 1 As shown, the example environment 100 may generally include an electronic device 120. In some embodiments, the electronic device 120 may be a device with computing capabilities, such as a personal computer, workstation, or server. The scope of this disclosure is not limited in this respect.

[0022] In some embodiments, the electronic device 120 may acquire a content set 130 as input. The content in the content set 130 may be, for example, articles, images, music, or videos. The scope of this disclosure is not limited in this respect.

[0023] In some embodiments, the electronic device 120 receives as input a first recommendation result 110-1, a second recommendation result 110-2, ..., an Nth recommendation result 110-N (collectively referred to as recommendation result 110, either individually or collectively) determined based on various recommendation techniques, where N is an integer greater than 1. Recommendation result 110 indicates the degree of recommendation for each piece of content in the content set 130. Taking article recommendation as an example, various recommendation techniques include, but are not limited to, the BM25 (Best Matching 25) algorithm, the Latent Dirichlet Allocation (LDA) algorithm, the Doc2vec algorithm, and the Paper2Vec algorithm. It should be understood that recommendation result 110 can also be determined by using any other suitable recommendation technique, and the scope of this disclosure is not limited in this respect.

[0024] In some embodiments, the recommendation results 110 may be presented in tabular form. Table 1 below shows an exemplary recommendation result 110.

[0025] Table 1. Exemplary Recommendation Results

[0026] Content Number Recommendation 1 0.37 2 0.21 3 0.05 4 0.11 5 0.26

[0027] The first column of the table, "Content Number," displays the content number corresponding to the content in content set 130, and the second column, "Recommendation Level," shows the degree of recommendation for the corresponding content in numerical form. It should be understood that the recommendation result 110 can also be presented in any other suitable manner, and the degree of recommendation can also be expressed in a non-numerical form; the scope of this disclosure is not limited in this respect.

[0028] Electronic device 120 can integrate multiple recommendation results 110 by considering the similarity between them to obtain a target recommendation result 140 for content set 130. This will be discussed in detail below. Figures 2 to 4 Further detailed description.

[0029] Figure 2 A schematic diagram of a process 200 for determining a target recommendation result 140 according to some embodiments of the present disclosure is shown. Figure 2As shown, the electronic device 120 can generate a first optimized recommendation result 210-1, a second optimized recommendation result 210-2, ..., an Nth optimized recommendation result 210-N (each individually or collectively referred to as an optimized recommendation result 210) based on multiple recommendation results 110, and determine a target recommendation result 140 based on the optimized recommendation results 210.

[0030] In some embodiments, the electronic device 120 may determine an optimized recommendation result 210 by considering the similarity between multiple recommendation results 110. This will be discussed in conjunction with the following. Figure 3 Further detailed description.

[0031] In some embodiments, such as when the recommendation result 110 is in the tabular form described above, the electronic device 120 can directly add the recommendation scores for each item in the optimized recommendation result 210 determined based on the recommendation result 110 to obtain the target recommendation result 140. It should be understood that, depending on the form of the recommendation result, the target recommendation result 140 can also be obtained from the optimized recommendation result 210 in any other suitable manner, and the scope of this disclosure is not limited in this respect.

[0032] In some embodiments, the content set 130 may be a set of candidate articles, and the various recommendation techniques described above may recommend candidate articles from the set of candidate articles based on a predetermined base article (not shown). This base article may, for example, be pre-specified by the user. In this case, the electronic device 120 may additionally determine a set of relevance measures associated with the set of candidate articles. Each relevance measure in this set indicates the degree of relevance between the corresponding candidate article and the base article.

[0033] In some embodiments, for each candidate article in a set of candidate articles, the electronic device 120 may utilize a pre-trained language model to determine the relevance between the references of the candidate article and the title of the base article, as a measure of relevance associated with the candidate article. Language models include, but are not limited to, Bidirectional Encoder Representations from Transformers (BERT) models, Generative Pre-Training (GPT) models, Word2vec models, etc.

[0034] The inventors discovered through research that an article's references often include one or more articles that the author considers relatively relevant to their work. Therefore, the relevance between the references of candidate articles and the titles of base articles, determined using a language model, can accurately reflect the relevance between the candidate and base articles. Thus, by using this relevance metric to adjust the target recommendation result 140, the accuracy of the final recommendation result can be improved, thereby further enhancing the quality of the final recommendation result.

[0035] It should be understood that the relevance between candidate articles and base articles can be determined in any other suitable manner, and the scope of this disclosure is not limited in this respect.

[0036] In some embodiments, for multiple references included in a candidate article, the electronic device 120 can utilize a pre-trained language model to determine the relevance between the title of each reference and the title of the base article, as a measure of the relevance between that reference and the base article. The electronic device 120 can also calculate the average relevance between all references included in a candidate article and the base article as a measure of relevance associated with the candidate article.

[0037] Electronic device 120 can adjust target recommendation result 140 using a determined set of relevance measures to update target recommendation result 140. In some embodiments, for each candidate article in a set of candidate articles, electronic device 120 can use the product of the relevance measure associated with the candidate article and the recommendation degree of the candidate article in target recommendation result 140 as the updated recommendation degree of the candidate article.

[0038] In this way, the electronic device 120 can fine-tune the obtained target recommendation result 140 by taking into account the relevance between the candidate article and the base article, so as to further improve the quality of the final recommendation result.

[0039] Figure 3 A schematic diagram of a process 300 for determining a first optimized recommendation result 210-1 according to some embodiments of the present disclosure is shown. Figure 3As shown, the electronic device 120 can determine a set of similarities 310-1 to 310-N (collectively referred to as similarity 310, individually or collectively) associated with the first recommendation result 110-1 based on the recommendation result 110. Similarity 310 can indicate the degree of similarity between two different recommendation results 110. For example, similarity 310-2 indicates the degree of similarity between the first recommendation result 110-1 and the second recommendation result 110-2, and similarity 310-N indicates the degree of similarity between the first recommendation result 110-1 and the Nth recommendation result 110-N. It should be noted that, in the context of this disclosure, similarity 310 can also include the degree of similarity between a recommendation result 110 and itself. For example, similarity 310-1 indicates the degree of similarity between the first recommendation result 110-1 and itself.

[0040] In the following description, the determination of the similarity 310-2 between the first recommendation result 110-1 and the second recommendation result 110-2 will be used as an example. In some embodiments, the electronic device 120 may determine a first set of instructions based on the first recommendation result 110-1, wherein the first set of instructions represents the degree of recommendation for the corresponding content in the content set 130. The first set of instructions may be represented, for example, as a vector of recommendation degrees. For example, a set of instructions for the recommendation results shown in Table 1 above may be a vector [0.37, 0.21, 0.05, 0.11, 0.26]. Similarly, the electronic device 120 may determine a second set of instructions based on the second recommendation result 110-2, wherein the second set of instructions represents the degree of recommendation for the corresponding content in the content set 130. Then, the electronic device 120 may determine the cosine similarity between the first set of instructions and the second set of instructions as the similarity 310-2 between the first recommendation result 110-1 and the second recommendation result 110-2.

[0041] In this way, compared with conventional simple weighting technical solutions, the solution according to the embodiments of this disclosure can analyze and explore the correlation between various recommendation results, thereby strengthening the proportion of multiple recommendation results with high similarity 310 in the final recommendation result, while taking into account recommendation results with relatively low similarity 310, thereby improving the accuracy and stability of the final recommendation result.

[0042] It should be understood that, depending on the specific form of the first set of instructions and the second set of instructions, the similarity 310 between the two recommendation results 110 can also be determined by any other suitable similarity metric such as Euclidean distance, Hamming distance, etc., and the scope of this disclosure is not limited in this respect.

[0043] The remaining similarities 310 of the electronic device 120 can be determined in a manner similar to that described above with reference to similarity 310-2. Further details are omitted here.

[0044] like Figure 3 As shown, electronic device 120 can adjust the corresponding recommendation result 110 based on the determined similarity 310 using adjusters 320-1 to 320-N (referred to individually or collectively as adjuster 320) to obtain an adjusted recommendation result. In some embodiments, electronic device 120 can determine a scaling factor based on similarity 310 and use the scaling factor to adjust the recommendation result 110. Exemplarily, electronic device 120 can directly determine the value of the cosine similarity described above as the scaling factor and calculate the product of the recommendation result 110 and the scaling factor as the adjusted recommendation result. It should be understood that electronic device 120 can also adjust the recommendation result 110 based on similarity 310 in any other suitable manner, such as by using a mapping table indicating the mapping relationship between similarity 310 and scaling factor to determine the scaling factor based on similarity 310, and the scope of this disclosure is not limited in this respect. By employing this adjustment method, the method according to embodiments of this disclosure can utilize similarity to adjust the influence of each recommendation result 110 on the final recommendation result, thereby strengthening the proportion of recommendation results with higher similarity 310 in the final recommendation result, while also taking into account recommendation results with relatively lower similarity 310. Therefore, the accuracy and stability of the final recommendation result can be improved.

[0045] In some embodiments, since the cosine similarity between a recommendation result 110 and itself is always 1, it is not necessary to calculate the similarity 310-1 for the first recommendation result 110-1, and the adjuster 320-1 can be omitted, thereby saving the computing power of the electronic device 120.

[0046] like Figure 3 As shown, electronic device 120 can determine a first optimized recommendation result 210-1 corresponding to the first recommendation result 110-1 based on the adjusted recommendation result output by adjuster 320. In some embodiments, electronic device 120 can directly add the various adjusted recommendation results to obtain the first optimized recommendation result 210. It should be understood that the optimized recommendation result 210 can also be obtained from the adjusted recommendation results in any other suitable manner, and the scope of this disclosure is not limited in this respect. Through this adjustment method, the method according to the embodiments of this disclosure can use similarity to adjust the various recommendation results 110, so that the target recommendation result 140 obtained based on the optimized recommendation result 210 can reflect the recommendation results with high similarity to a large extent, but at the same time, it can also take into account the recommendation results with relatively low similarity, thereby improving the accuracy and stability of the final recommendation result.

[0047] The electronic device 120 can determine the optimized recommendation result 210 corresponding to the remaining recommendation results 110 in a similar manner to determining the first optimized recommendation result 210. For example, the electronic device 120 can determine the similarity 310 between the second recommendation result 110-2 and each of the recommendation results 110, and use the determined similarity 310 to adjust each recommendation result 110 to obtain the second optimized recommendation result 210-2. Further details are omitted here.

[0048] Figure 4 A flowchart of a method 400 for recommending content according to some embodiments of the present disclosure is shown. For example, method 400 may be provided by, Figure 1 The illustrated electronic device 120 performs this action. It should be understood that method 400 may also include additional boxes not shown, and / or the boxes shown may be omitted. The scope of this disclosure is not limited in this respect. For ease of description, reference is made below. Figures 1 to 3 To describe method 400.

[0049] At box 402, electronic device 120 determines a similarity 310 between a first recommendation result 110-1 and a second recommendation result 110-2 for content set 130, the first recommendation result 110-1 and the second recommendation result 110-2 being determined based on different recommendation techniques and respectively indicating the degree of recommendation for each content in content set 130.

[0050] In some embodiments, determining similarity 310 includes: determining a first set of instructions based on a first recommendation result 110-1, wherein the first set of instructions respectively represent the degree of recommendation for corresponding content in the content set 130; determining a second set of instructions based on a second recommendation result 110-2, wherein the second set of instructions respectively represent the degree of recommendation for corresponding content in the content set 130; and determining the cosine similarity between the first set of instructions and the second set of instructions.

[0051] At box 404, the electronic device 120 uses similarity 310 to adjust the second recommendation result 110-2. In some embodiments, adjusting the second recommendation result 110-2 includes: determining a scaling factor based on similarity 310; and scaling the second recommendation result 110-2 using the scaling factor.

[0052] At box 406, electronic device 120 determines a target recommendation result 140 for content set 130 based on a first recommendation result 110-1 and an adjusted second recommendation result. In some embodiments, determining the target recommendation result 140 includes: using the first recommendation result 110-1 and the adjusted second recommendation result to obtain a first optimized recommendation result 210-1 corresponding to the first recommendation result 110-1; and determining the target recommendation result 140 based on the first optimized recommendation result 210-1.

[0053] In some embodiments, determining the target recommendation result 140 based on the first optimized recommendation result 210-1 includes: adjusting the first recommendation result 110-1 using similarity 310; obtaining a second optimized recommendation result 210-2 corresponding to the second recommendation result 110-2 using the second recommendation result 110-2 and the adjusted first recommendation result; and determining the target recommendation result 140 based on the first optimized recommendation result 210-1 and the second optimized recommendation result 210-2.

[0054] In some embodiments, the content set 130 is a set of candidate articles, and the recommendation technology recommends candidate articles from the set of candidate articles based on predetermined base articles. The method 400 further includes: determining a set of relevance measures associated with the set of candidate articles, each of the relevance measures indicating the degree of relevance between the corresponding candidate article and the base articles; and using the set of relevance measures to adjust the target recommendation result 140 to update the target recommendation result 140.

[0055] In some embodiments, a set of candidate articles includes a first candidate article, and a set of relevance measures includes a first relevance measure corresponding to the first candidate article. Determining a set of relevance measures includes: using a pre-trained language model to determine the degree of relevance between the references of the first candidate article and the title of the base article as the first relevance measure.

[0056] Through the above combination Figures 1 to 4 As can be seen from the description, the method for recommending content according to embodiments of this disclosure can, during the process of integrating multiple recommendation results to obtain a final recommendation result, utilize the correlation between the various recommendation results by considering the similarity between these recommendation results, thereby strengthening the influence of multiple similar recommendation results on the final recommendation result. Therefore, compared with known conventional solutions, the method according to embodiments of this disclosure can strengthen the proportion of several recommendation results with high similarity among multiple recommendation results in the final recommendation result, thereby improving the accuracy and stability of the final recommendation result and enhancing the quality of the final recommendation result.

[0057] Figure 5 A schematic block diagram is shown of an example device 500 that can be used to implement some embodiments according to this disclosure. Figure 5As shown, device 500 includes a central processing unit (CPU) 501, which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) 502 or loaded from storage unit 508 into random access memory (RAM) 503. RAM 503 may also store various programs and data required for the operation of device 500. CPU 501, ROM 502, and RAM 503 are interconnected via bus 504. Input / output (I / O) interface 505 is also connected to bus 504.

[0058] Multiple components in device 500 are connected to I / O interface 505, including: input unit 506, such as keyboard, mouse, etc.; output unit 507, such as various types of monitors, speakers, etc.; storage unit 508, such as disk, optical disk, etc.; and communication unit 509, such as network card, modem, wireless transceiver, etc. Communication unit 509 allows device 500 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0059] The various processes and handling described above, such as method 400, can be executed by processing unit 501. For example, in some embodiments, method 400 can be implemented as a computer software program tangibly contained in a machine-readable medium, such as storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on device 500 via ROM 502 and / or communication unit 509. When the computer program is loaded into RAM 503 and executed by CPU 501, one or more actions of methods 200 and 300 described above can be performed.

[0060] This disclosure can be a method, apparatus, system, and / or computer program product. A computer program product may include a computer-readable storage medium having computer-readable program instructions loaded thereon for performing various aspects of this disclosure.

[0061] Computer-readable storage media can be tangible devices capable of holding and storing instructions for use by an instruction execution device. Computer-readable storage media can be, for example—but not limited to—electrical storage devices, magnetic storage devices, optical storage devices, electromagnetic storage devices, semiconductor storage devices, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of computer-readable storage media include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static random access memory (SRAM), portable compact disc read-only memory (CD-ROM), digital multifunction disc (DVD), memory sticks, floppy disks, mechanical encoding devices, such as punch cards or recessed protrusions storing instructions thereon, and any suitable combination of the foregoing. The computer-readable storage media used herein are not to be construed as transient signals themselves, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., light pulses through fiber optic cables), or electrical signals transmitted through wires.

[0062] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0063] Computer program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, status setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing the status information of the computer-readable program instructions to implement various aspects of this disclosure.

[0064] Various aspects of this disclosure are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this disclosure. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0065] These computer-readable program instructions can be provided to a processing unit of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processing unit of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner. Thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0066] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0067] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction containing one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0068] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments. The terminology used herein is chosen to best explain the principles, practical application, or improvement of the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for recommending content, comprising: The similarity between a first recommendation result and a second recommendation result for a content set is determined. The first recommendation result and the second recommendation result are determined based on different recommendation techniques and respectively indicate the degree of recommendation for each content in the content set. The similarity is used to adjust the second recommendation result; as well as Based on the first recommendation result and the adjusted second recommendation result, a target recommendation result is determined for the content set. Determining the similarity between the first recommendation result and the second recommendation result includes: Based on the first recommendation result, a first vector containing a first set of instructions is determined, wherein the first set of instructions is represented in the form of a vector composed of the recommendation degree of the corresponding content in the content set; Based on the second recommendation result, a second vector containing a second set of indicators is determined, wherein the second set of indicators is represented in the form of a vector consisting of the recommendation degree of the corresponding content in the content set; and The similarity is determined at least in part based on the calculated cosine similarity between the first vector and the second vector; The adjustment of the second recommendation result using the similarity includes: The scaling factor is determined based on the similarity; and The second recommendation result is adjusted based on the scaling factor to adjust its influence in determining the target recommendation result; The adjustment of the second recommendation result based on the scaling factor includes: calculating the updated value of each indicator of the second group of indicators in the second vector based on the product of the value of each indicator of the second group of indicators in the second vector and the factor.

2. The method according to claim 1, wherein determining the target recommendation result includes: The first recommended result and the adjusted second recommended result are used to obtain the first optimized recommended result corresponding to the first recommended result; as well as The target recommendation result is determined based on the first optimized recommendation result.

3. The method according to claim 2, wherein determining the target recommendation result based on the first optimized recommendation result includes: The first recommendation result is adjusted using the similarity. The second recommendation result and the adjusted first recommendation result are used to obtain a second optimized recommendation result corresponding to the second recommendation result; as well as Based on the first optimized recommendation result and the second optimized recommendation result, the target recommendation result is determined.

4. The method according to claim 1, wherein the content set is a set of candidate articles, the recommendation technology recommends candidate articles from the set of candidate articles based on predetermined base articles, and the method further comprises: A set of relevance measures are determined to be associated with the set of candidate articles, each of the relevance measures indicating the degree of relevance between the corresponding candidate article and the base article; as well as The target recommendation results are adjusted using the aforementioned set of relevance measures to update the target recommendation results.

5. The method according to claim 4, wherein the set of candidate articles includes a first candidate article, the set of relevance measures includes a first relevance measure corresponding to the first candidate article, and determining the set of relevance measures includes: A pre-trained language model is used to determine the degree of relevance between the references of the first candidate article and the title of the base article, which is used as the first relevance measure.

6. An electronic device, comprising: processor; as well as A memory coupled to the processor, the memory having instructions stored therein, the instructions causing the device to perform actions when executed by the processor, the actions including: The similarity between a first recommendation result and a second recommendation result for a content set is determined. The first recommendation result and the second recommendation result are determined based on different recommendation techniques and respectively indicate the degree of recommendation for each content in the content set. The similarity is used to adjust the second recommendation result; and Based on the first recommendation result and the adjusted second recommendation result, a target recommendation result is determined for the content set. Determining the similarity between the first recommendation result and the second recommendation result includes: Based on the first recommendation result, a first vector containing a first set of instructions is determined, wherein the first set of instructions is represented in the form of a vector composed of the recommendation degree of the corresponding content in the content set; Based on the second recommendation result, a second vector containing a second set of indicators is determined, wherein the second set of indicators is represented in the form of a vector consisting of the recommendation degree of the corresponding content in the content set; and The similarity is determined at least in part based on the calculated cosine similarity between the first vector and the second vector; The adjustment of the second recommendation result using the similarity includes: The scaling factor is determined based on the similarity; and The second recommendation result is adjusted based on the scaling factor to adjust its influence in determining the target recommendation result; The adjustment of the second recommendation result based on the scaling factor includes: calculating the updated value of each indicator of the second group of indicators in the second vector based on the product of the value of each indicator of the second group of indicators in the second vector and the factor.

7. The electronic device of claim 6, wherein determining the target recommendation result includes: The first recommended result and the adjusted second recommended result are used to obtain the first optimized recommended result corresponding to the first recommended result; as well as The target recommendation result is determined based on the first optimized recommendation result.

8. The electronic device of claim 7, wherein determining the target recommendation result based on the first optimized recommendation result includes: The first recommendation result is adjusted using the similarity. The second recommendation result and the adjusted first recommendation result are used to obtain a second optimized recommendation result corresponding to the second recommendation result; as well as Based on the first optimized recommendation result and the second optimized recommendation result, the target recommendation result is determined.

9. The electronic device of claim 6, wherein the content set is a set of candidate articles, the recommendation technology recommends candidate articles from the set of candidate articles based on predetermined base articles, and the action further includes: A set of relevance measures are determined to be associated with the set of candidate articles, each of the relevance measures indicating the degree of relevance between the corresponding candidate article and the base article; as well as The target recommendation results are adjusted using the aforementioned set of relevance measures to update the target recommendation results.

10. The electronic device of claim 9, wherein the set of candidate articles includes a first candidate article, the set of relevance measures includes a first relevance measure corresponding to the first candidate article, and determining the set of relevance measures includes: A pre-trained language model is used to determine the degree of relevance between the references of the first candidate article and the title of the base article, which is used as the first relevance measure.

11. A computer program product tangibly stored on a computer-readable medium and comprising machine-executable instructions that, when executed, cause a machine to perform the method according to any one of claims 1 to 5.

Images