Method and apparatus for expanding content, device, medium, and program product

Abstract

Embodiments of the present disclosure relate to a method and apparatus for expanding content, a device, a medium, and a program product. The method comprises: determining a set of second search words in a search word set that are associated with a first search word, wherein the quantity of original content associated with the first search word is less than a predetermined quantity threshold. The method further comprises: on the basis of a set of second target content associated with the second search words, determining first target content associated with the first search word, the first target content being used for expanding content associated with the first search word. In the method, one search word can be used to expand content used for recalling and associated with another search word, so that more content can be recalled during the search of search words. Thus, the effect of recommending content for search words can be improved, and the user experience can be improved.

Description

Methods, apparatus, devices, media, and programs for expanding content. Technical Field

[0001] The embodiments of this disclosure generally relate to the field of data processing technology, and more specifically to a method, apparatus, device, medium, and program product for expanding content. Background Technology

[0002] With the popularization of the internet and people's increasing reliance on it, the amount of data on the internet is growing exponentially, gradually forming massive data resources. Processing and classifying this data has become increasingly important. Along with the continuous increase in internet data, various search and recommendation systems have begun to be applied. These systems can recommend content to users based on their selected or entered search terms, enabling users to quickly find the content they need from a vast amount of data.

[0003] Currently, search recommendation systems typically employ a funnel structure, which includes three levels: recall, coarse ranking, and fine ranking. Recall refers to selecting as many results as possible relevant to the search term from a database containing a large amount of information and data. Coarse ranking is the initial screening and sorting of the results obtained from recall. Fine ranking is the precise screening and sorting of the coarse ranking results, selecting the optimal few results.

[0004] Summary of the Invention

[0005] Embodiments of this disclosure provide a method, apparatus, device, medium, and program product for expanding content.

[0006] According to a first aspect of this disclosure, a method for expanding content is provided. The method includes identifying a set of second search terms in a search term set associated with a first search term, wherein the number of original content associated with the first search term is less than a predetermined number threshold. The method further includes determining first target content associated with the first search term based on a set of second target content associated with the second search terms, the first target content being used to expand the content associated with the first search term.

[0007] In a second aspect of this disclosure, an apparatus for expanding content is provided. The apparatus includes: an association module configured to determine a set of search terms in a search term set that are associated with a first search term, wherein the number of original contents associated with the first search term is less than a predetermined quantity threshold; and an expansion module configured to determine first target content associated with the first search term based on a set of target content associated with the set of search terms, wherein the first target content is used to expand the content associated with the first search term.

[0008] In a third aspect of this disclosure, an electronic device is provided, including at least one processor; and a storage device for storing at least one program, which, when executed by the at least one processor, causes the at least one processor to implement the method according to the first aspect of this disclosure.

[0009] In a fourth aspect of this disclosure, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the method according to a first aspect of this disclosure.

[0010] In a fifth aspect of this disclosure, a computer program product is provided, which is tangibly stored on a non-volatile, i.e., computer-readable medium and includes computer-executable instructions that, when executed, cause a computer to perform the method provided according to a first aspect of this disclosure.

[0011] It should be understood that the content described in this section is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0012] The above and other objects, features and advantages of this disclosure will become more apparent from the accompanying drawings, in which like reference numerals generally denote like parts.

[0013] Figure 1 illustrates a schematic diagram of an example environment in which the methods of the embodiments of this disclosure may be implemented;

[0014] Figure 2 illustrates a scenario of recommending content based on search terms in some embodiments of this disclosure;

[0015] Figure 3 illustrates a schematic diagram of a method for expanding content according to some embodiments of the present disclosure;

[0016] Figure 4 illustrates a schematic diagram of a method for determining a second search term associated with a first search term according to some embodiments of the present disclosure;

[0017] Figure 5 illustrates a schematic diagram of a method for determining a first target content according to some embodiments of the present disclosure;

[0018] Figure 6 illustrates a schematic diagram of a method for expanding content according to some embodiments of the present disclosure;

[0019] Figure 7 illustrates a schematic block diagram of an apparatus for expanding content according to some embodiments of the present disclosure; and

[0020] Figure 8 illustrates a schematic block diagram of an example device suitable for implementing embodiments of the present disclosure.

[0021] In the various figures, the same or corresponding reference numerals indicate the same or corresponding parts. Detailed Implementation

[0022] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition or use of the data) shall comply with the requirements of relevant laws, regulations and related provisions. Upon receiving a user's active request, a prompt message is sent to the user to clearly inform the user that the requested operation will require the acquisition and use of the user's personal information. This allows the user to independently choose whether to provide personal information to the electronic device, application, server, or storage medium, or other software or hardware that performs the operation of this disclosed technical solution, based on the prompt message. The user interaction operations or user-content interaction involved in this disclosure, as well as the data related to user operations (including but not limited to data used for analysis, stored data, displayed data, etc.), are all recorded, collected or stored with the user's authorization or full authorization from all parties. Furthermore, the collection, use, and processing of related data must comply with the relevant laws, regulations and standards of the relevant countries and regions, and corresponding operation entry points are provided for the user to choose to authorize or refuse. The collection, storage, use, processing, transmission, provision, and disclosure of user-related information involved in the technical solutions of this disclosure embodiments all comply with the provisions of relevant laws and regulations and do not violate public order and good morals.

[0023] 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.

[0024] 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.

[0025] As mentioned above, in the recall phase of the search recommendation process, the search recommendation system needs to select content related to the search terms from a large amount of content from various information sources. Subsequent coarse and fine ranking are based on the content obtained from the recall. In some cases, for some search terms entered by the user, the amount of content retrieved is relatively small. This will result in too little content being ultimately used to recommend to the user after coarse and fine ranking, which will degrade the user experience.

[0026] To address at least the aforementioned and other potential problems, embodiments of this disclosure propose a method for expanding content. In this method, for a first search term whose recalled content is less than a predetermined threshold, a set of second search terms associated with the first search term can be identified in the search term set, and first target content associated with the first search term can be determined based on a set of second target content associated with the second search terms.

[0027] This method allows for the merging of retrieved content from related search terms into the retrieved content of a single search term, thereby expanding the retrieved results for that term. In this way, the retrieved results for search recommendation systems with limited content can be expanded, ensuring a sufficient number of results for these terms, thus improving recommendation performance and enhancing user experience.

[0028] The embodiments of this disclosure will now be described in further detail with reference to the accompanying drawings. Figure 1 illustrates an example environment 100 in which the methods of the embodiments of this disclosure can be implemented. As shown in Figure 1, environment 100 may include a recommendation device 101 and a user device 102. The recommendation device 101 may include, but is not limited to, a server, a minicomputer, a mainframe computer, a personal computer, a multiprocessor system, a handheld or laptop device, or a mobile device (such as a mobile phone, a personal digital assistant (PDA), a media player, etc.). The recommendation device 101 can obtain search terms input by the user from the user device 102, and can also generate content to recommend to the user based on the search terms, and send the content to the user device 102.

[0029] User device 102 can be any device capable of accepting input information, such as a mobile phone, tablet, laptop, PDA, smart TV, PDA, smart printer, smart home appliance, in-vehicle terminal, wearable device (smartwatch, smart bracelet, smart glasses, etc.), virtual reality (VR) device, augmented reality (AR) device, etc., and this application embodiment does not limit this. User device 102 can receive search terms entered by the user and send the search terms to recommendation device 101. User device 102 can also receive content corresponding to the search terms from recommendation device 101 and present it to the user.

[0030] As an example, user device 102 may have a display interface, which may include a search term input box 103 and a content display area 104. User device 102 can receive search terms entered by the user through the search term input box 103 and display recommended content to the user through the content display area 104. In embodiments of this disclosure, the content recommended to the user may include, but is not limited to, text, audio, video, etc.

[0031] In some embodiments, the recommendation device 101 may, based on search terms obtained from the user device 102, identify content associated with the search terms from a large amount of content through recall, and then determine the content to be recommended to the user from the recalled content through coarse and fine ranking. In some embodiments, the recommendation device 101 may merge the recalled content for one search term into the recalled content for another search term.

[0032] In some embodiments, the recommendation device 101 may locally store content associated with search terms. In some embodiments, the recommendation device 101 may merge locally stored content associated with one search term into content associated with another search term. It should be understood that environment 100 is merely an example of an embodiment of this disclosure and should not be construed as a limitation of this disclosure. In some embodiments, the recommendation device 101 and the user device 102 may be the same device, that is, the user device 102 may implement the functions of the recommendation device 101.

[0033] Figure 2 illustrates a schematic diagram of scenario 200, which recommends content based on search terms, according to an embodiment of this disclosure. Referring to Figure 2, scenario 200 includes a recommendation device 210, which may be, for example, a recommendation device 101 in environment 100. The recommendation device 210 can acquire a search term 220 and generate recommended content 230 corresponding to the search term 220. The recommendation device 210 may include a recall module 211, a filtering module 212, and a storage 213. The recall module 2111 can determine multiple contents related to the search term 220 from all content from the cloud or local storage based on the search term 220. The filtering module 212 can perform coarse and fine filtering on the contents determined by the recall module 211 to ultimately determine the recommended content 230.

[0034] In some embodiments, the recall module 211 can determine thousands of contents from millions of contents using a predefined natural language processing (NLP) model. The filtering module 212 can determine the relevance between the contents determined by the recall module 211 and the search term 220, and determine the contents with a relevance greater than a predefined relevance threshold or the contents with the highest predefined number of relevance as recommended content 230.

[0035] In some embodiments, the recall module 211 can merge the recalled content associated with one search term into the recalled content associated with another search term, and use the merged content as the final recall result for the other search term. In some embodiments, the recommendation device 210 can store the recommended content 230 determined by the filtering module 212 in association with the search term 220 in the memory 213, so that when the search term 220 is obtained again, the recommendation device 210 can perform near-line recall, that is, directly match the obtained search term with the search terms stored in the memory 213 to determine the recommended content 230. In some embodiments, after obtaining the search term 220, the recommendation device 210 can match it with the search terms stored in the memory 213 to determine multiple candidate contents, and the filtering module 212 can further filter the determined multiple candidate contents to determine new recommended content.

[0036] The above description, with reference to Figures 1 and 2, illustrates example environments and scenarios in which the methods of the embodiments of this disclosure can be implemented. The following description, with reference to Figures 3 to 6, describes the methods for expanding content provided by this disclosure. Figure 3 shows a schematic diagram of methods for expanding content according to some embodiments of this disclosure. The methods shown in Figure 3 can be executed on the recommending device 101 shown in Figure 1, the recommending apparatus 200 shown in Figure 2, or any other suitable device. The method 300 provided by the embodiments of this disclosure will now be illustrated using the recommending apparatus as an example. Referring to Figure 3, method 300 may include blocks 302 and 304.

[0037] At box 302, the recommendation device determines a set of second search terms associated with the first search term from the search term set. The number of original contents associated with the first search term is less than a predetermined threshold. In embodiments of this disclosure, search terms (including the first search term and search terms in the search term set) can be text or tags entered by the user for searching content of interest to the user. In some embodiments, search terms can be tags obtained based on content classification from user input or identified through an algorithmic model.

[0038] The search term set may include one or more search terms, and the first search term may also include one or more search terms. Both the first search term and the search terms in the search term set may be associated with one or more pieces of content. In some embodiments, the content associated with a search term may be obtained by the recommendation device through a recall process. In some embodiments, the content associated with a search term may be determined during a search process prior to the search term, and these search terms and content may be stored in association in the memory of the recommendation device, for example, in the following format:

[0039] {query1:[doc1,doc2,doc3…]};

[0040] {query2:[doc2,doc4,doc5…]};

[0041] Wherein, “doc1”, “doc2”, and “doc3” can be content associated with the search term “query1”, and “doc2”, “doc4”, and “doc5” can be content associated with the search term “query2”. In some embodiments, the search term set, the first search term, and the content associated with it can be obtained by the recommendation device from other devices, such as from one or more user devices. In some embodiments, the number of contents associated with the search terms in the search term set is greater than or equal to a predetermined threshold. In some embodiments, the number of contents associated with the first search term can be zero, that is, the first search term may have no associated content.

[0042] In some embodiments, the recommendation device may determine a set of second search terms from the search term set based on the similarity between the first search term and search terms in the search term set. In some embodiments, the recommendation device may determine a set of second search terms from the search term set based on a predefined classification model. The set of second search terms may include one or more search terms.

[0043] At box 304, the recommendation device determines the first target content associated with the first search term based on a set of second target content associated with the second search term. The first target content is used to expand the content associated with the first search term. As described in box 302 above, the second search term is a search term in the search term set, and the second search term is associated with one or more pieces of content, referred to as the second target content for ease of explanation. The recommendation device can determine the first target content from the second target content and use it as new content associated with the first search term. In other words, the recommendation device can use content associated with the second search term to supplement the content associated with the first search term.

[0044] For example, the second search term associated with the first search term "query1" can be "query2", and the second target content associated with the second search term can be "doc2", "doc4", and "doc5". The recommendation device can determine the first target content "doc2" and "doc5". In some embodiments, the recommendation device can determine all content associated with the second search term as the first target content. In some embodiments, the recommendation device can determine the first target content based on the similarity between the second target content and the first search term. In some embodiments, the recommendation device can filter the second target content based on predefined filtering rules and determine the filtered second target content as the first target content.

[0045] Through the aforementioned method 300, the recommendation device can expand the content associated with the first search term by using a second search term associated with it, thus increasing the amount of content associated with the original, less relevant first search term. In this way, even if the results retrieved for the first search term are limited during the search recommendation process, the second search term can be used to expand the retrieved results, thereby increasing the amount of content ultimately available for recommendation to the user. This improves the user experience.

[0046] In some embodiments, the recommendation device may store first target content associated with a first search term. For example, the recommendation device may store the first target content associated with the first search term in local memory. For instance, the first target content may include "doc2" and "doc5", and the first search term may be "query1". The recommendation device may store the first target content associated with the first search term as {query1:[doc2,doc5]}. In some embodiments, the first search term has associated original content, such as "doc1". The recommendation device may store the original content, the first target content, and the first search term associated as {query1:[doc1,doc2,doc5]}. Thus, when the recommendation device receives the search term "query1" again, it can perform near-line recall, directly determining the content associated with the search term using the data stored in local memory. Furthermore, since the content associated with the search term stored in the local cache is expanded, more content associated with the search term can be determined through near-line recall, thereby improving the recommendation effect and user experience.

[0047] In some embodiments, in the aforementioned block 302, the recommendation device may divide multiple search terms into a first search term and search terms in a set of search terms, and on this basis determine a second search term associated with the first search term from the set of search terms. Exemplarily, FIG4 shows a schematic flowchart of a method 400 for determining a second search term associated with a first search term according to some embodiments of the present disclosure. Referring to FIG4, method 400 may include blocks 402 to 408.

[0048] In box 402, the recommendation device acquires multiple search terms. These multiple search terms may be determined based on the historical search records of one or more users. The recommendation device can acquire users' search records from one or more user devices to determine the one or more search terms. In some embodiments, the recommendation device may acquire multiple search terms from other devices configured in the cloud for providing search recommendation services to users.

[0049] In box 404, the recommendation device determines a first search term and a set of search terms based on the frequency with which the search term is searched within a predetermined time period. The predetermined time period may be, for example, one day or one week. The frequency with which the search term is searched may be the frequency with which the search term is searched by all users on the Internet. In some embodiments, the recommendation device is used to provide search recommendation services to a specific user, and the frequency with which the search term is searched may also be the frequency with which the search term is searched by that specific user.

[0050] In box 406, the recommendation device determines the features of the first search term and the features of the search terms in the search term set. Exemplarily, the recommendation device can input the first search term and the search terms in the search term set into a predefined recall model, thereby obtaining the features corresponding to the first search term and the features corresponding to the search terms in the search term set based on the recall model. In some embodiments, the features may be, for example, embeddings. In some embodiments, the recommendation device can determine the features corresponding to the first search term and the features corresponding to the search terms in the search term set based on other types of embedding models.

[0051] In box 408, the recommendation device can determine the second search term associated with the first search term in the search term set based on a predefined classification model. The classification model can include, but is not limited to, the k-nearest neighbor (KNN) algorithm, a self-attention model, or a bidirectional encoder representation from transformers (BERT) model. The recommendation device inputs the features of the first search term and the features of the search terms in the search term set into the classification model to obtain the second search term associated with the first search term. Using this method, if the first search term includes multiple search terms, the second search term associated with each of these multiple search terms can be quickly determined.

[0052] It should be understood that the method 400 shown in Figure 4 is merely an example of an embodiment of this disclosure and should not be construed as a limitation on the embodiments of this disclosure. In some embodiments of this disclosure, the recommendation device may also distinguish between the first search term and the search term set using other methods, and may also determine the second search term set using other methods. For example, in some embodiments, in block 404, the recommendation device may divide multiple search terms into a first search term and search terms in the search term set based on the number of contents associated with the search term. For instance, search terms with a number of associated contents less than a predetermined threshold may be determined as the first search term, and search terms with a number of associated contents greater than or equal to the predetermined threshold may be determined as search terms in the search term set.

[0053] In some embodiments, in block 408, the recommendation device can determine the similarity between the first search term and the search terms in the search term set based on the features of the first search term and the features of the search terms in the search term set, for example, using a dual-tower model. For a specific first search term, the recommendation device can determine the similarity between each of the search terms in the search term set and the first search term, and sort the search terms in the search term set in descending order of similarity. Based on this sorting, the recommendation device can determine the second search term. For example, the recommendation device can determine the top 15 search terms as the second search terms associated with the first search term.

[0054] In some embodiments, in the aforementioned block 302, the second search term identified by the recommendation device as associated with the first search term may include multiple search terms, each of which may be associated with multiple second target contents. The recommendation device may merge these second target contents and determine the first target content based on this. In some embodiments, the recommendation device may filter and select the second target content. For example, the recommendation device may filter out content in the second target content that is identical to the original content. Alternatively, the recommendation device may determine the similarity between the second target content and the first search term, and select the first target content from the second target content based on the similarity.

[0055] For example, FIG5 illustrates a schematic flowchart of a method 500 for determining first target content in some embodiments of the present disclosure. Referring to FIG5, method 500 may include blocks 502 to 508. In block 502, the recommendation device may obtain original content associated with a first search term and second target content associated with a second search term. In some embodiments, the original content associated with the first search term may be determined by the recommendation device from a large amount of content from multiple data sources using a recall model, and the second target content associated with the second search term may also be determined by the recommendation device from a large amount of content using a recall model.

[0056] In some embodiments, the original content associated with the first search term may be determined by a recall model and a filtering model during a previous search recommendation process for the first search term and stored in the memory of the recommendation device. The second search term may be determined by the aforementioned passbox 302 or the aforementioned method 400. The second target content associated with the second search term may be determined by the recommendation device by a recall model and a filtering model during a previous search recommendation process for the second search term and stored in the memory of the recommendation device. The recommendation device can retrieve the original content and the second target content from the memory. In some embodiments, the original content and the second target content may also be obtained by the recommendation device from other recommendation devices or systems.

[0057] In block 504, the recommendation device filters out content in the second target content that is identical to the original content. Since the recommendation device's memory already stores the original content associated with the first search term, or the recommendation device has already determined or obtained the original content associated with the first search term, it can remove content in the second target content that is identical to the original content, processing only content that is different from the original content. This avoids duplicate content in the finally determined content associated with the first search term. In some embodiments, the second target content is derived from content associated with multiple second search terms, where duplicate content exists; the recommendation device can remove the duplicate content in the second target content.

[0058] In box 506, the recommendation device can determine the similarity between the second target content and the first search term based on a predefined recall model. The recommendation device can input the second target content and the first search term together into the predefined recall model, which can determine the features of the second target content and the features of the first search term, and calculate the cosine similarity between the features of the second target content and the features of the first search term.

[0059] In box 508, the recommendation device determines the second target content as the first target content based on its similarity to the first search term. For one first search term, there may be multiple second target contents, and the recommendation device can sort these multiple second target contents in descending order based on their cosine similarity to the first search term. In some embodiments, the recommendation device can determine the second target contents with a cosine similarity greater than a predetermined similarity threshold as the first target content. In some embodiments, the recommendation device can determine a predefined number of second target contents that are ranked first as the first target content.

[0060] Method 500 can remove duplicate content from the second target content and further filter the second target content based on the recall model, so that the identified first target content has a better correlation with the first search term, thereby providing more matching results in subsequent filtering or near-line recall processes and improving the recommendation effect.

[0061] In some embodiments, the recommendation device determines the similarity between the second target content and the first search term through block 506 in method 500, so that the first target content determined from the second target content in block 508 also has a similarity with the first search term. The recommendation device can store the first target content, the first search term, and the similarity between the first target content and the first search term in association. When a user's search request for the first search term is received subsequently, the recommendation device can directly provide the first target content to the user based on the first target content and similarity stored in the memory, in descending order of similarity. In this way, the efficiency of recommending content to the user can be improved.

[0062] In some embodiments, after determining the first target content through method 300, the recommendation device may store the first target content and the first search term in association. Subsequently, if a search request for the first search term is received, the recommendation device can retrieve the first target content from the memory and input it into a predefined fine-ranking model to perform further precise filtering of the first target content, thereby determining content to be recommended that is more closely matched to the first search term.

[0063] In some embodiments, the recommendation device can use the stored first target content and first search term for search recommendations within a predetermined time period. For example, the predetermined time period could be one day. After the recommendation device identifies and stores the first target content associated with the first search term, if it receives a user's search request for the first search term within the following day, it can directly recommend content to the user based on the content associated with the first search term stored in its memory. If more than one day has passed, the recommendation device can execute method 300 again, or use a predefined recall and filtering model to determine the content to recommend to the user from multiple contents stored on the internet or in cloud storage. Thus, even if the content associated with the first search term stored in the recommendation device's memory becomes unsuitable for the user's needs due to excessive time, the recommendation device can adjust accordingly, thereby always recommending content that meets the user's needs.

[0064] Figure 6 illustrates a schematic diagram of a method 600 for expanding content according to some embodiments of this disclosure. The method shown in Figure 6 can be performed by a device for expanding content, such as the recommendation device 101 in scenario 100 or the recommendation device 210 in scenario 200. In method 600, content associated with long-tail search terms can be expanded based on high-frequency search terms. The method 600 will now be illustrated schematically using a recommendation device as an example. Referring to Figure 6, method 600 may include steps 1) to 9).

[0065] In step 1), the recommendation device inputs high-frequency search term 601 and long-tail search term 602 into the recall model 603. High-frequency search terms are those searched at a frequency greater than or equal to a predetermined search threshold within a predetermined time period, while long-tail search terms are those searched at a frequency less than the predetermined search threshold within the predetermined time period. In step 2), the recommendation device obtains features 604 of the high-frequency search term and features 605 of the long-tail search term through the recall model 603. In step 3), the recommendation device inputs features 605 of the long-tail search term and features 604 of the high-frequency search term into the classification model 606. In step 4), the classification model 606 determines a second search term 607 from the high-frequency search terms. In step 5), the recommendation device determines a second target content 608 associated with the second search term 607. In step 6), the recommendation device determines the original content 610 associated with the long-tail search term 602 and removes content in the second target content 608 that overlaps with the original content 610, obtaining the filtered second target content. In step 7), the recommendation device inputs the filtered second target content 609 into the recall model 611, and inputs the features 605 of the long-tail search terms into the recall model 611. The recall model 611 can be the same recall model as the recall model 603, or it can be a different model.

[0066] In step 8), the recall model 611 generates features of the filtered second target content 609, and determines the similarity between the filtered second target content 609 and the long-tail search term 602 based on the features 605 of the long-tail search term. Then, based on the similarity, the first target content 612 is determined from the filtered second target content 609. In step 9), the recommendation device stores the first target content 612, the original content 610, and the long-tail search term 602 in a locally configured cache device. In this way, the recommendation device expands the content associated with the long-tail search term, thereby providing more content associated with the long-tail search term in subsequent near-line recall processes. This improves the effectiveness of search recommendation.

[0067] The above description, in conjunction with Figures 3 to 6, illustrates the method provided in the embodiments of this disclosure. Next, Figures 7 and 8 will be used to describe the apparatus provided in the embodiments of this disclosure. Figure 7 illustrates a schematic diagram of an apparatus 700 for expanding content according to some embodiments of this disclosure. Exemplarily, as shown in Figure 7, apparatus 700 may include: an association module 702, configured to determine a set of search terms in a search term set associated with a first search term, wherein the number of original contents associated with the first search term is less than a predetermined threshold; and an expansion module 704, configured to determine first target content associated with the first search term based on a set of target content associated with the set of search terms, wherein the first target content is used to expand the content associated with the first search term.

[0068] In some embodiments, the first search term is a search term whose search frequency is less than a predetermined search threshold within a predetermined time period, and the search terms in the search term set are search terms whose search frequency is greater than or equal to the predetermined search threshold within a predetermined time period.

[0069] In some embodiments, the association module 702 includes a classification unit configured to determine, based on a predefined classification model, multiple sets of second search terms that are respectively associated with multiple first search terms in a set of search terms.

[0070] In some embodiments, the classification unit includes: a first embedding determination unit configured to determine a first embedding of each of the plurality of first search terms; a second embedding determination unit configured to determine a second embedding of each of the search terms in the search term set; and a first association unit configured to determine multiple sets of second search terms respectively associated with the plurality of first search terms based on the first embedding, the second embedding and a predefined classification model.

[0071] In some embodiments, the association module 702 includes: a first sorting unit configured to determine the order of search terms in the search term set based on the similarity between the first search term and search terms in the search term set; and a second association unit configured to determine a second search term based on the order of search terms in the search term set.

[0072] In some embodiments, the expansion module 704 includes: an acquisition unit configured to acquire a second target content and an original content; a filtering unit configured to filter content in the second target content that is the same as the original content; and a first determination unit configured to determine a first target content based on the filtered second target content.

[0073] In some embodiments, the first determining unit includes: a similarity determining unit configured to determine the similarity between the filtered second target content and the first search term based on a predefined recall model; and a second determining unit configured to determine the first target content based on the similarity between the filtered second target content and the first search term.

[0074] In some embodiments, the apparatus 700 further includes: a first storage module configured to store a first target content, a first search term, and a similarity between the first search term and the first target content in association; and a first recommendation module configured to, in response to receiving a user's search request for the first search term, determine content to recommend to the user based on the similarity between the first target content and the first search term.

[0075] In some embodiments, the apparatus 700 further includes: a second storage module configured to store the first target content in association with the first search term; a second ranking module configured to, in response to receiving a user's search request for the first search term, determine the ranking of the first target content and the original content based on the first target content, the original content, and a predefined ranking model; and a second recommendation module configured to determine content to recommend to the user based on the ranking of the first target content and the original content.

[0076] Figure 8 shows a schematic block diagram of an example device 800 that can be used to implement embodiments of the present disclosure. The recommended device 101 in Figure 1 or the recommended device 210 in Figure 2 can be implemented using device 800. As shown in Figure 8, device 800 includes a central processing unit (CPU) 801, which can perform various appropriate actions and processes according to computer program instructions stored in read-only memory (ROM) 802 or loaded from storage unit 808 into random access memory (RAM) 803. Various programs and data required for the operation of device 800 may also be stored in RAM 803. CPU 801, ROM 802, and RAM 803 are interconnected via bus 804. Input / output (I / O) interface 805 is also connected to bus 804.

[0077] Multiple components in device 800 are connected to I / O interface 805, including: input unit 806, such as keyboard, mouse, etc.; output unit 807, such as various types of monitors, speakers, etc.; storage page 808, such as disk, optical disk, etc.; and communication unit 809, such as network card, modem, wireless transceiver, etc. Communication unit 809 allows device 800 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0078] The various processes and handling described above, such as methods 300, 400, 500, and / or 600, may be executed by processing unit 801. For example, in some embodiments, methods 300, 400, 500, and / or 600 may be implemented as computer software programs tangibly contained in a machine-readable medium, such as storage unit 808. In some embodiments, part or all of the computer program may be loaded and / or installed on device 800 via ROM 802 and / or communication unit 809. When the computer program is loaded into RAM 803 and executed by CPU 801, one or more actions of methods 300, 400, 500, and / or 600 described above may be performed.

[0079] 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.

[0080] 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.

[0081] 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.

[0082] 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.

[0083] 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.

[0084] 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.

[0085] 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.

[0086] 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.

[0087] The various embodiments of this disclosure have been described above. These descriptions are exemplary and not exhaustive, and are not 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 applications, or technical improvements to the technology in the market, or to enable others skilled in the art to understand the embodiments disclosed herein.

Claims

1. A method for expanding content, comprising: Identify a set of second search terms in the search term set that are associated with the first search term, wherein the number of original contents associated with the first search term is less than a predetermined threshold. as well as Based on a set of second target content associated with the second search term, first target content associated with the first search term is determined, and the first target content is used to expand the content associated with the first search term.

2. The method according to claim 1, wherein the first search term is a search term whose search frequency is less than a predetermined search threshold within a predetermined time period, and the search terms in the search term set are search terms whose search frequency is greater than or equal to the predetermined search threshold within the predetermined time period.

3. The method according to claim 1 or 2, wherein determining a set of second search terms associated with the first search term in the search term set includes: Based on a predefined classification model, multiple sets of second search terms are determined from the set of search terms, each associated with a plurality of the first search terms.

4. The method according to claim 3, wherein determining multiple sets of second search terms associated with the multiple first search terms respectively in the search term set based on a predefined classification model includes: Determine the first embedding of each of the multiple first search terms; Determine the second embedding of each search term in the search term set; as well as Based on the first embedding, the second embedding, and the predefined classification model, multiple sets of second search terms are determined that are respectively associated with multiple first search terms.

5. The method according to claim 1 or 2, wherein determining a set of second search terms associated with the first search term in the search term set includes: Based on the similarity between the first search term and the search terms in the search term set, the order of the search terms in the search term set is determined; as well as The second search term is determined based on the order of the search terms in the search term set.

6. The method according to claim 1 or 2, wherein determining the first target content associated with the first search term based on a set of second target content associated with the second search term comprises: Obtain the second target content and the original content; Filter out content in the second target content that is identical to the original content; as well as Based on the filtered second target content, the first target content is determined.

7. The method according to claim 6, wherein determining the first target content based on the filtered second target content comprises: Based on a predefined recall model, the similarity between the filtered second target content and the first search term is determined; as well as The first target content is determined based on the similarity between the filtered second target content and the first search term.

8. The method according to claim 7, further comprising: The first target content, the first search term, and the similarity between the first search term and the first target content are stored in association; as well as In response to receiving a user's search request for the first search term, based on the similarity between the first target content and the first search term, content is determined to be recommended to the user.

9. The method according to claim 1, further comprising: Store the first target content in association with the first search term; In response to receiving a user's search request for the first search term, the ranking of the first target content and the original content is determined based on the first target content, the original content, and a predefined ranking model. as well as Based on the sorting of the first target content and the original content, the content to be recommended to the user is determined.

10. An apparatus for expanding content, comprising: The association module is configured to determine a set of search terms in the search term set that are associated with the first search term, wherein the number of original contents associated with the first search term is less than a predetermined number threshold. as well as The expansion module is configured to determine a first target content associated with the first search term based on a set of target content associated with the first set of search terms. This content is used to expand the content associated with the first search term.

11. An electronic device, comprising: processor; as well as A memory coupled to the processor, the memory having instructions stored therein, which, when executed by the processor, cause the electronic device to perform the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having stored thereon computer-executable instructions, wherein the computer-executable instructions are executed by a processor to implement the method according to any one of claims 1 to 9.

13. A computer program product tangibly stored on a non-volatile computer-readable medium and comprising computer-executable instructions that, when executed, cause a computer to perform the steps of the method according to any one of claims 1 to 9.

Images