Recommendation word determination method and apparatus, device, and storage medium

Abstract

Embodiments of the present disclosure provide a method, device and storage medium for determining recommended words, which comprises: obtaining position information of a terminal, and based on the position information of the terminal, obtaining target historical search behavior data corresponding to the position information; processing and analyzing the target historical search behavior data to extract target keywords with local characteristics; wherein the local characteristics are used to represent characteristics associated with the position information; and fusing the position information and the target keywords to generate a plurality of recommended words. Embodiments of the present disclosure can solve the problem that in the prior art, the determination of a recommended word candidate set requires manual review, and the recommended words in the recommended word candidate set are limited, which cannot meet the search needs of users in time and improve the search efficiency of users.

Description

Technical Field

[0001] This disclosure relates to the field of computer technology, and in particular to a method, apparatus, device, and storage medium for determining recommended terms. Background Technology

[0002] With the development of the Internet and mobile devices, using search functions in various applications or web pages has become increasingly convenient for users.

[0003] Currently, for the search function, a secure set of recommended terms is provided to users beforehand, after filtering and manual review. When a user initiates a search in the search box, to reduce the user's search effort, some recommended terms are often selected from the set of recommended terms to be displayed in the search box.

[0004] However, the current technology requires manual review to determine the candidate set of recommended terms, and the recommended terms in the candidate set are limited, which cannot meet users' search needs in a timely manner and improve users' search efficiency. Summary of the Invention

[0005] This disclosure provides a method, apparatus, device, and storage medium for determining recommended terms, in order to solve the problems in the prior art where the determination of the candidate set of recommended terms requires manual review, and the recommended terms in the candidate set are limited, which cannot meet the user's search needs in a timely manner and improve the user's search efficiency.

[0006] In a first aspect, embodiments of this disclosure provide a method for determining recommendation terms, including:

[0007] Obtain the location information of the terminal, and based on the location information of the terminal, obtain the target historical search behavior data corresponding to the location information;

[0008] The target's historical search behavior data is processed and analyzed to extract target keywords with local features; wherein, the local features are used to represent features associated with the location information;

[0009] The location information and the target keywords are combined to generate multiple recommended words.

[0010] Secondly, embodiments of this disclosure provide a device for determining recommended terms, comprising:

[0011] The first acquisition module is used to acquire the location information of the terminal, and based on the location information of the terminal, acquire the target historical search behavior data corresponding to the location information;

[0012] The second acquisition module is used to process and analyze the target's historical search behavior data and extract target keywords with local features; wherein, the local features are used to represent features associated with the location information;

[0013] The recommendation word generation module is used to merge the location information and the target keywords to generate multiple recommendation words.

[0014] Thirdly, embodiments of this disclosure provide an electronic device, including: at least one processor and a memory;

[0015] The memory stores computer-executed instructions;

[0016] The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the recommendation word determination method as described in the first aspect and various possible designs of the first aspect.

[0017] Fourthly, embodiments of this disclosure provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the recommendation word determination method described in the first aspect and various possible designs of the first aspect.

[0018] The method, apparatus, device, and storage medium for determining recommended terms provided in this disclosure first acquire the location information of a terminal, and then acquire target historical search behavior data corresponding to the location information based on the terminal's location information. Next, the target historical search behavior data is processed and analyzed to extract target keywords with local characteristics. These local characteristics represent features associated with the location information. The location information and the target keywords are then fused to generate multiple recommended terms. Therefore, by freely combining the terminal's location information with target keywords possessing local characteristics from the securely filtered target historical search behavior data, multiple recommended terms are generated. This ensures search security, promptly meets user search needs, improves user search efficiency, and ultimately enhances the user's search experience. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A schematic diagram illustrating a scenario for the method of determining recommended terms provided in this embodiment of the disclosure;

[0021] Figure 2 A flowchart illustrating the method for determining recommendation terms provided in this embodiment of the disclosure;

[0022] Figure 3 A flowchart illustrating a method for determining recommendation terms provided in another embodiment of this disclosure;

[0023] Figure 4 A structural block diagram of the recommendation term determination device provided in the embodiments of this disclosure;

[0024] Figure 5 This is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of this disclosure. Detailed Implementation

[0025] 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.

[0026] It should be understood that the steps described in the method embodiments of this disclosure may be performed in different orders and / or in parallel. Furthermore, the method embodiments may include additional steps and / or omit the steps shown. The scope of this disclosure is not limited in this respect.

[0027] The term "comprising" and its variations as used herein are open-ended inclusions, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". Definitions of other terms will be given in the description below.

[0028] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0029] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0030] The names of messages or information exchanged between multiple devices or modules in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0031] When a user initiates a search in the search box, they may not be able to enter accurate search terms, thus failing to obtain the desired results. To reduce the user's search costs, some search terms are often recommended within the search box. However, in existing technologies, to ensure basic user search experience and search security requirements, a limited set of recommended candidates is provided to the user through multiple layers of filtering and manual review. This significantly limits the recommendation of search terms. Furthermore, since the scale of recommended search terms is closely related to the recommendation effect, the limited set of recommended candidates in existing technologies makes it impossible to accurately recommend search terms to users while ensuring search security requirements. Consequently, it fails to promptly meet the user's search needs and improve search efficiency.

[0032] To address the aforementioned issues, the technical concept disclosed herein is as follows: Local demand features are mined from search behavior obtained with user authorization. Multiple recommended terms are generated by combining the extracted local demand features with the user's geographical location. This achieves a combination of user location information and local demand features, solving the problem of low accuracy in recommended search terms in existing technologies, which fails to promptly meet user search needs and improve search efficiency. It enables precise delivery of personalized local terms to users, better satisfying their personalized search needs while ensuring the security and quality of personalized recommendations.

[0033] The technical solutions of this disclosure will be described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments.

[0034] It should be noted that the user's search information, behavioral data, geographical location, and other information are obtained with the user's authorization, and will not be described again in the following examples.

[0035] See Figure 1 As shown, Figure 1 This is a schematic diagram illustrating a scenario for the recommendation term determination method provided in this embodiment. In practical applications, the entity executing the recommendation term determination method can be a recommendation term determination device. This device can be implemented as software, hardware, or a combination of software and hardware, such as a computer device. The computer device can then execute the recommendation term determination method provided in this embodiment. The computer device may include a server. As an example, the recommendation term determination method provided in this embodiment can be executed by a single server, or multiple servers can collaborate to implement the recommendation term determination method provided in this embodiment.

[0036] For example, taking an application as an example, a user can open the application through terminal 10, or enter a search term in the search input box provided by the application, or click search to trigger an instruction operation to generate suggested search terms. The generated instruction is sent to server 20. After receiving the instruction, the server obtains the location information of the target user who triggered the operation, as well as the target's historical search behavior data associated with the location information. Then, it performs localized feature mining on the target's historical search behavior data to obtain target keywords with local characteristics. Combined with the location information, it generates a set of recommended terms containing a large number of terms to be recommended. The set of recommended terms is then integrated into the recommendation recall set and participates in the recommendation model ranking. This realizes the push of personalized local terms based on the user's location information, better meeting the user's personalized search needs, while ensuring the security and quality of personalized recommendations.

[0037] Specifically, the method for determining recommended keywords can be implemented in the following ways:

[0038] See Figure 2 , Figure 2 This is a flowchart illustrating the method for determining recommended terms provided in this embodiment. The method can be executed by a server, and can be applied to web pages or apps, implemented through interaction between a terminal and the server. The terminal can be a computer or a mobile terminal, which may have a browser or other various apps installed, providing search functionality to the user.

[0039] Combination Figure 2 As shown, the method for determining recommended terms may include:

[0040] Step S101: Obtain the location information of the terminal, and based on the location information of the terminal, obtain the target historical search behavior data corresponding to the location information.

[0041] In this embodiment of the disclosure, when a user performs a search through a terminal, the location information of the terminal used by the target user is first obtained in response to an instruction for indicating recommended search terms. The target user is the user who triggered the operation to generate the instruction.

[0042] In this embodiment of the disclosure, actions such as a user opening an application (e.g., a browser or a mobile app) on a terminal, entering a search term in the search input box provided by the application, or clicking search are all triggering operations (here, the user is the target user, i.e., the object of the recommended search terms). The terminal generates an instruction to indicate the recommended search terms based on the user's triggering operation and sends this instruction to the server. After receiving the instruction, the server can obtain the target user's current geographical location, i.e., location information. This location information can be used to mine local feature information to meet the user's personalized needs.

[0043] Step S102: Process and analyze the target historical search behavior data to extract target keywords with local characteristics.

[0044] The local features are used to represent features associated with the location information.

[0045] In this embodiment, the target historical search behavior data is obtained by performing security screening on the historical search behavior data of each historical user. These historical users may or may not include the target user, depending on the specific application scenario. Mining the target historical search behavior data aims to infer the user's search intent, thereby saving the user's search costs.

[0046] Specifically, the location information here can include geographic location. Based on the acquired location information, secure search behavior data of each historical user within that geographic location is obtained; that is, target historical search behavior data is obtained by performing security screening on the historical search behavior data of each historical user. Therefore, by using geographic location to lock in the historical search behavior data of each historical user, localized search behavior data is provided to the target user, laying the foundation for meeting the user's personalized needs.

[0047] In one embodiment of this disclosure, the target historical search behavior data includes search behavior data corresponding to multiple search behavior sessions. The search behavior data corresponding to each search behavior session includes multiple search terms, interaction data corresponding to each search term, and search order corresponding to each search term. The interaction data and search order corresponding to the search terms are used to characterize the degree to which the search terms meet the search requirements.

[0048] The interaction data here includes the search results of the interaction, and the search behavior data here includes location information.

[0049] The acquisition of the target's historical search behavior data corresponding to the location information can be achieved through the following steps:

[0050] Step a1: Retrieve historical search behavior data of each historical user corresponding to the location information from the database. The historical search behavior data includes search keywords, visited web page information, clicked search results, and the search order of the search keywords.

[0051] Step a2: Cluster the search keywords, visited webpage information, clicked search results, and search order of the search keywords for each historical user.

[0052] Step a3: Filter the sensitive words in the clustering results to obtain the target historical search behavior data.

[0053] In this embodiment of the disclosure, in order to provide users with localized recommended words, the historical search behavior data of each historical user with the same location information can be found from the database first. However, in order to ensure the search security of the target user, sensitive words need to be filtered.

[0054] Specifically, the historical search behavior data here may include search keywords, web page information visited, search results clicked, and the search order of the search keywords. In order to avoid wasting resources or time by filtering each piece of data in the historical search behavior data, the historical search behavior data of each historical user can be clustered first. For example, the web page information visited by the search keywords, the search results clicked, and the search order of the search keywords can be clustered separately.

[0055] Cluster analysis can be performed by first extracting valid behavioral operations (such as browsing, clicking, viewing, and searching) from historical search behavior data to form a feature set. This feature set is then converted into a vector space model. A clustering algorithm is used to cluster the feature items in this model, yielding the clustering results. Each cluster is then treated as a whole and filtered out words containing sensitive terms (i.e., words lacking security). Only safe words are retained as target historical search behavior data, providing secure and localized search behavior data for the target users.

[0056] Step S103: Merge the location information and the target keywords to generate multiple recommended words.

[0057] In one embodiment of this disclosure, the target keywords are multiple, and the location information includes historical location information and current location information; the fusion of the location information and the target keywords to generate multiple recommended words can be achieved through the following steps:

[0058] Step b1: Deduplicate the location information and the target keywords;

[0059] Step b2: Concatenate the keywords in the deduplicated results to generate the multiple recommended words.

[0060] In this embodiment, since location information may contain target keywords, and target keywords may also contain location information, it is necessary to deduplicatize the location information and target keywords. For example, if target keywords include region A, weather, and oil price, and location information includes region A, then region A is deduplicated. Finally, the deduplicated results are concatenated with keywords, such as "region A weather" or "region A oil price" as recommended terms. For instance, target keywords may carry geographic locations, such as "geographic location A weather." Therefore, the location information and target keywords can be preprocessed, that is, each pair of location information and target keywords is deduplicated. For example, if the geographic location is "geographic location A" and the target keyword is "geographic location A weather," then "geographic location A" is deduplicated, and then keywords are concatenated or combined to obtain "geographic location A weather," which is the recommended term.

[0061] Specifically, in order to accurately recommend search terms to users and meet their personalized needs, the existing recommendation candidate set can be expanded. Instead of filtering through layers of machine review, manual review, and online review, which results in a limited recommendation candidate set, location information can be combined with all local demand features to generate a massive set of recommendation terms.

[0062] For example, location information includes geographic location A, geographic location B, ..., geographic location N, and local demand characteristic target keywords include target keyword 1, target keyword 2, ..., target keyword M. Combinations are made in pairs: geographic location A target keyword 1, geographic location A target keyword 2, ..., geographic location A target keyword M, geographic location B target keyword 1, geographic location B target keyword 2, ..., geographic location B target keyword M, ..., geographic location N target keyword 1, geographic location N target keyword 2, ..., geographic location N target keyword, etc., generating multiple candidate words for recommendation, forming a set of recommended keywords. Unlike existing limited candidate recommendation sets, this set contains a massive number of candidate words with localized needs, providing users with a sufficient and accurate resource of recommended search terms, thereby meeting users' personalized needs.

[0063] The recommended term determination method provided in this disclosure obtains the location information of a terminal and, based on the location information, obtains target historical search behavior data corresponding to the location information; then, it processes and analyzes the target historical search behavior data to extract target keywords with local features; wherein, the local features are used to represent features associated with the location information; and the location information and the target keywords are fused to generate multiple recommended terms. Therefore, by freely combining the terminal's location information with target keywords with local features from the securely filtered target historical search behavior data to generate multiple recommended terms, it ensures search security, promptly meets users' search needs, improves user search efficiency, and thus enhances the user's search experience.

[0064] In one embodiment of this disclosure, see Figure 3 As shown, Figure 3 This is a flowchart illustrating a method for determining recommendation terms according to another embodiment of the present disclosure. This embodiment builds upon the above embodiments, for example, in... Figure 2 Based on the aforementioned embodiments, a detailed explanation of how to extract target keywords with local characteristics is provided. This can be achieved through the following steps:

[0065] S201. Based on the multiple search terms corresponding to the search behavior sessions, aggregate each search behavior session to obtain a set of multiple search behavior sessions;

[0066] S202. For each set of search behavior sessions, deduplication of search terms in the set of search behavior sessions is performed to obtain multiple target search terms;

[0067] S203. Based on the interaction data and search order corresponding to each of the search terms, determine the interaction data and search order corresponding to each of the target search terms;

[0068] S204. Based on the interaction data and search order corresponding to each of the target search terms, determine candidate search terms from multiple search behavior session sets;

[0069] S205. Based on the candidate search terms, determine the target keywords, which include local keywords.

[0070] In this embodiment of the disclosure, localized recommendation candidate words, i.e. candidate search terms, can be constructed by utilizing localization requirements. In the process of generating candidate search terms, localized requirements features that conform to localization can be extracted from the target historical search behavior data. For example, a large number of users search for the same keywords with geographical location information (such as weather at geographical location A, weather at geographical location B, weather at geographical location C, etc.). The "weather" corresponding to these geographical location keyword information is a localized requirement.

[0071] Specifically, since the target historical search behavior data is filtered based on location information, it possesses locality. Then, keywords with local characteristics, such as weather, oil prices, and housing prices, are extracted from this localized historical search behavior data as local demand features, i.e., target keywords. Determining target keywords allows for the delivery of localized terms tailored to users' needs, avoiding the push of unrepresentative local terms, thereby stimulating user search demand and improving the overall search performance of the client.

[0072] In one embodiment of this disclosure, determining the target keyword based on the candidate search terms can be achieved through the following steps:

[0073] Step c1: Find search terms related to local content information from the candidate search terms;

[0074] Step c2: Select search terms related to local content information as the target keywords.

[0075] In this embodiment of the disclosure, search terms related to local content information, such as weather and oil prices, are directly searched from the candidate search terms and used as target keywords.

[0076] In addition, by obtaining a recommendation recall set and determining the target keyword based on the candidate search terms and the recommendation recall set, the target keyword can be used as the word to be recommended.

[0077] In this embodiment of the disclosure, a recommended candidate word recall method is adopted. Candidate words, i.e. words to be recommended, constructed through localization requirements and user location information are incorporated into the recommendation recall set and participate in the ranking of the recommendation model. The search terms with the highest ranking and a preset number of searches are taken as target recommended words, and the target recommended words, i.e. personalized local words, are pushed to the user.

[0078] Specifically, the determination of the recommended recall set can be achieved through the following steps:

[0079] Step d1: If the target user is a new user, then obtain the attributes corresponding to the target user. The attributes are generated by pushing information to the target user through cold start.

[0080] Step d2: Through personalized recall, obtain the first keyword that matches the attribute from the target keywords, and obtain the first search result associated with the first keyword.

[0081] Step d3: If the target user is the historical user, then based on search behavior trend statistics, user behavior, geographical location and other information are obtained after user authorization.

[0082] Step d4: Through personalized recall, obtain a second keyword that matches user behavior, geographical location and other information from the target keyword, and obtain a second search result associated with the second keyword.

[0083] Step d5: If the terminal used by the target user is detected to have a trigger operation for performing input events, then after authorization by the target user, the current behavior data of the target user is obtained, and the current behavior data includes context information.

[0084] Step d6: Obtain the target association information associated with the context information.

[0085] Step d7: Generate a recommendation recall set by combining the first keyword, the first search result, the second keyword, the second search result, and the target association information.

[0086] In this embodiment of the disclosure, the retrieval of search-related information can be achieved through personalized retrieval or retrieval based on the context of the user's current search.

[0087] For example, if the target user is a new user, since there is no historical search behavior data for the target user, it is impossible to perform behavioral analysis on the target user itself. However, by clustering the attributes of the target user with historical users, we can find historical users who match the attributes of the target user. We can then use the localized information obtained from the historical behavior data analysis of the historical users who meet the conditions to recall or recall the hot search terms for the current time period as a personalized recall. The recalled information is stored in the recommendation recall set.

[0088] In practical applications, if the target user is a new user, personalized information can be pushed to the user when they register for the app for the first time or open the browser for the first time, allowing the target user to make selections. These can include information such as gender, age, occupation, and preferences. Once the target user is identified, their attributes are generated.

[0089] Specifically, based on the target user's attributes, a personalized recall method is used to find the keyword that matches the attribute from the target keywords, i.e., the first keyword, and obtain the first search result associated with the first keyword. Then, the first keyword and the first search result associated with the first keyword are used as elements in the recommendation recall set.

[0090] This disclosure does not limit the specific scenario; it can also achieve contextual recall by monitoring whether the target user's terminal triggers an operation to execute an input event. Here, the input event can refer to entering a search term in an input box, and the triggering operation can be a search or query.

[0091] Specifically, if a triggering operation for executing an input event is detected on the terminal, the search context information can be obtained. Here, context information can refer to the current search content, indicating that the behavior is valid and can be used as recall information. Based on the context information, target association information related to that context information is retrieved from the database or through a full network search, and this target association information is used as an element in the recommendation recall set. Therefore, a recommendation recall set is generated using the first keyword, the first search result, the second keyword, the second search result, and the target association information. All recommendations in this recall set are valid and can be used as search terms to be recommended. That is, the search terms recommended to the user need to be obtained from the recommendation recall set, and the recommended terms obtained from the recommendation recall set have high accuracy.

[0092] Determining target keywords can also be achieved through the following steps:

[0093] Step e1: If the instruction generated by the terminal is triggered by the target user through inputting search information or by clicking the search button, then obtain the search information.

[0094] Step e2: Identify whether there are geographical location and / or local keywords in the search information.

[0095] Step e3: If the keywords related to the geographic location and / or the locality exist, then search for target keywords associated with the keywords related to the geographic location and / or the locality from the target historical search behavior data.

[0096] Step e4: If the instruction is triggered by the target user by opening the application, then search for target keywords with local characteristics from the searched keywords, the accessed web page information, the clicked search results, and the search order.

[0097] Step e5: Use the target keyword as the local demand feature.

[0098] In practical applications, users can perform searches on web pages or within apps. Taking an app as an example, a user can launch the app from their device. The launch of the app triggers an instruction to recommend search terms. Since this triggering action doesn't involve actual input or searching, target keywords with local characteristics can be found from historical search behavior data. Users can also trigger the instruction to recommend search terms by entering search information or clicking the search button. In this scenario, the server can retrieve the search information from the input box as a basis for extracting target keywords with local characteristics.

[0099] Specifically, the system first identifies whether geographical location and / or local keywords exist in the search information, such as keywords related to weather, oil prices, or location A weather. If so, it searches the target's historical search behavior data for keywords associated with the geographical location and / or local keywords as target keywords. These target keywords are all based on localized needs and specific local characteristics, enabling accurate recommendations to be provided to users. Therefore, personalized recommendations can be made for new users or users who do not need to enter any search terms, or based on the search terms entered by the target user.

[0100] In this embodiment of the disclosure, in order to accurately recommend search terms to users and meet their personalized needs, personalized local terms can be pushed to users by mining local demand data, constructing local recommendation term candidates using local demand, and ranking them in the recommendation recall set.

[0101] Specifically, the process involves mining basic data on localized needs: For search behavior (extracting or acquiring user search keywords, visited web pages, click results, search order, etc.), this information is combined to extract characteristics of users' local needs in search results, thus mining localized needs; Localized recommendation keyword candidates are constructed using these localized needs: The above mining methods are used to map corresponding localized needs across the entire network. First, the pattern roots of localized needs are extracted using the above mining methods, such as weather, email, housing prices, and job postings. Then, these are combined based on the user's geographical location to generate a massive set of recommendation keywords; The candidate keywords constructed using localized needs and user geographical information are integrated into the recommendation candidate recall set, i.e., the recommendation recall set, to participate in the recommendation model ranking, and personalized local keywords are pushed based on the user's geographical information.

[0102] In practical applications, the mining of localized search needs involves extracting the results content corresponding to search keywords, including the title of the results, webpage information, and browsing and clicking behaviors, to extract the characteristics of users' localized needs. For example, if a large number of users search for the same keywords with geographical location information (such as weather in a certain region 1, weather in a certain region 2, weather in a certain region 3), the "weather" corresponding to these geographical location keywords is a localized need. Additionally, back-engineering is performed using queries retrieved from local content information (search results), such as oil price information (diesel, 92-octane gasoline), fresh produce (pork price, chicken price, egg price), and job postings (application for xx position). Time-sensitive keyword combinations are then used: based on extensive search behavior, i.e., mining localized demand characteristics and existing geographic location information across the entire network, time-sensitive candidate keywords are constructed. For example, the "weather" query mentioned above is one feature, and the keywords derived from the local content information results are also a type of feature. Further data cleaning, including deduplication and filtering, is performed, combined with user geographic location to generate a massive online pool of localized recommendation candidates. For example, given a user's geographic information and the localized demand they construct, candidates such as "region 1 weather," "region 2 oil price," and "region 3 secondhand housing information" are generated. Recommendation fusion (based on the time-sensitive keywords and recommendation ranking generated above): using the localized keywords generated in the above steps, the recommendation recall set is integrated, and based on the user's geographic location information, the candidates are assigned specific localized features for personalized delivery to the user.

[0103] In one embodiment of this disclosure, based on the above embodiments, a detailed explanation of how to determine target keywords is provided, which can be achieved in the following manner:

[0104] Step f1: Obtain media resource information;

[0105] Step f2: Extract keywords including geographic location and / or local keywords from the media resource information.

[0106] Step f3: Use the keywords of the geographical location and / or local keywords as the target keywords, wherein the target keywords have local characteristics.

[0107] In this embodiment of the disclosure, the media resource information may include keywords related to geographical location and / or local keywords. Therefore, the media resource information can be obtained, and keywords including keywords related to geographical location and / or local keywords can be extracted from the media resource information as target keywords with local characteristics.

[0108] In one embodiment of this disclosure, the method may further include the following steps:

[0109] A set of recommended words within a preset time period is obtained, and a target recommended word is determined to be recommended to the terminal based on the multiple recommended words and the set of recommended words. The target recommended word is used to represent a personalized local word.

[0110] Obtaining the set of recommended words may include the following steps:

[0111] Step g1: Based on the historical behavior data corresponding to the terminal, determine multiple first search keywords corresponding to the historical behavior data;

[0112] Step g2: In response to the triggering operation of the input event, obtain the current behavior data corresponding to the terminal, and determine multiple second search keywords corresponding to the current behavior data;

[0113] Step g3: Generate the recommended word set based on the first search keyword and the second search keyword.

[0114] In this embodiment of the disclosure, firstly, multiple first search keywords corresponding to the historical behavior data are searched from the historical behavior data of the terminal. When an input event is detected on the terminal, the current behavior data corresponding to the terminal is obtained, and multiple second search keywords corresponding to the current behavior data are selected. Then, the first search keywords and the second search keywords are put into the recommended word set.

[0115] In one embodiment of this disclosure, determining the target recommended words to be recommended to the terminal based on the plurality of recommended words and the set of recommended words can be achieved through the following steps:

[0116] Step m1: Determine multiple recommended words with local features that match the location information from the multiple recommended words;

[0117] Step m2: Add the multiple recommended words with local features to the recommended word set, and obtain the target recommended word set by deduplication;

[0118] Step m3: Sort each word to be recommended in the target recommendation word set using a preset recommendation model to obtain the sorted words to be recommended; the preset recommendation model is a recommendation word recommendation model that is at least related to time and location;

[0119] Step m4: Select the words to be recommended with the preset ranking as the target recommended words.

[0120] In this embodiment of the disclosure, multiple recommended words with local features that match the location information are determined from multiple recommended words and added to the recommended word set. After deduplication, a target recommended word set is obtained. Then, each word to be recommended in the set is sorted to obtain the target recommended words to be recommended to the user.

[0121] Specifically, a subset of multiple recommended terms is fused into the recommendation recall set, allowing multiple terms to participate in the ranking of the recommendation model. Specifically, multiple terms are added to the recommendation recall set and deduplication is performed to obtain the target recommendation recall set. This target recommendation recall set contains a massive number of candidate terms. To ensure the accuracy of search term recommendations and meet users' local personalized needs, the candidate terms in the target recommendation recall set can be input into a preset recommendation model for ranking. Then, the search terms with preset rankings can be selected as target recommended terms and displayed in the input box on the terminal. Multiple target recommended terms can be used. This approach, by guessing the user's search intent, reduces the user's retrieval cost, stimulates the user's search demand, and improves the overall search performance of the client.

[0122] In order to continuously push accurate search terms to users, the recommended term set can be updated in real time. This embodiment, based on the above embodiment, provides a detailed explanation of how to update the recommended term set in real time. This can be achieved through the following steps:

[0123] Step h1: If an instruction to search for a target recommended term is received from the target user, then obtain the target search results corresponding to the target recommended term.

[0124] Step h2: Update the recommended word set according to the target search results.

[0125] In this embodiment of the disclosure, after the target recommendation word is pushed to the user's terminal, the triggering events of the client on the terminal are monitored in real time. If the user clicks the search operation on the client, the client can generate an input command to instruct the server to search for the target recommendation word. The server then searches according to the target recommendation word and obtains the target search result corresponding to the target recommendation word. The target search result is the search result of the valid recommendation word, and the target search result can be added to the recommendation recall set to update the recommendation word set.

[0126] Therefore, this disclosure, based on the personalized limited candidate recommendation service, combines the local needs of search users and utilizes the existing local life service information across the entire network to significantly increase the scale of local demand candidates, better meet users' personalized search needs, and at the same time ensure the security and quality of personalized recommendations. It can accurately recommend search terms to users, thereby stimulating users' search needs, improving the overall search performance of the client, and enhancing the user's search experience.

[0127] Corresponding to the recommendation term determination method disclosed in the above embodiments, Figure 4This is a structural block diagram of a recommendation word determination device provided in an embodiment of this disclosure. The recommendation word determination device can be a function computation platform. For ease of explanation, only the parts relevant to the embodiments of this disclosure are shown. (Refer to...) Figure 4 The recommendation word determination device includes:

[0128] The first acquisition module 401 is used to acquire the location information of the terminal and, based on the location information of the terminal, acquire the target historical search behavior data corresponding to the location information.

[0129] The second acquisition module 402 is used to process and analyze the target historical search behavior data and extract target keywords with local features; wherein, the local features are used to represent features associated with the location information;

[0130] The recommendation word generation module 403 is used to merge the location information and the target keyword to generate multiple recommendation words.

[0131] The first acquisition module 401, the second acquisition module 402, and the recommendation word generation module 403 provided in this embodiment are used to acquire the location information of the terminal, and based on the location information, acquire the target historical search behavior data corresponding to the location information; then, process and analyze the target historical search behavior data to extract target keywords with local features; wherein, the local features are used to represent features associated with the location information; and the location information and the target keywords are fused to generate multiple recommendation words. Therefore, by freely combining the terminal's location information with target keywords with local features from the securely filtered target historical search behavior data to generate multiple recommendation words, both search security and timely satisfaction of user search needs are ensured, user search efficiency is improved, and thus the user search experience is enhanced.

[0132] The apparatus provided in this disclosure can be used to execute the technical solutions of the first aspect and the various possible designs corresponding to the first aspect above. The implementation principle and technical effects are similar, and will not be repeated here.

[0133] In one embodiment of this disclosure, the target historical search behavior data includes search behavior data corresponding to multiple search behavior sessions. The search behavior data corresponding to each search behavior session includes multiple search terms, interaction data corresponding to each search term, and search order corresponding to each search term. The interaction data and search order corresponding to the search terms are used to characterize the degree to which the search terms meet the search requirements.

[0134] In one embodiment of this disclosure, the second acquisition module 402 is described in detail. The second acquisition module 402 is specifically used for:

[0135] Based on the multiple search terms corresponding to the search behavior sessions, the search behavior sessions are aggregated to obtain multiple sets of search behavior sessions;

[0136] For each set of search behavior sessions, the search terms in the set of search behavior sessions are deduplicated to obtain multiple target search terms;

[0137] Based on the interaction data and search order corresponding to each of the search terms, determine the interaction data and search order corresponding to each of the target search terms;

[0138] Based on the interaction data and search order corresponding to each target search term, candidate search terms are determined from multiple search behavior session sets;

[0139] Based on the candidate search terms, the target keywords are determined, including local keywords.

[0140] In one embodiment of this disclosure, based on the above-described embodiments, the second acquisition module 402 is described in detail. The second acquisition module 402 is specifically used for:

[0141] Search for search terms related to local content information from the candidate search terms;

[0142] Search terms related to local content information are used as the target keywords.

[0143] In one embodiment of this disclosure, based on the above-described embodiments, the recommended term determination device is described in detail. The recommended term determination device further includes: a first processing module; the first processing module is configured to:

[0144] Obtain media resource information;

[0145] Extract keywords including geographic location and / or local keywords from the media resource information;

[0146] The target keywords are the keywords of the geographical location and / or local keywords, which have local characteristics.

[0147] In one embodiment of this disclosure, based on the above-described embodiments, the recommendation word generation module 403 is described in detail. The target keywords are multiple, and the location information includes historical location information and current location information. The recommendation word generation module 403 is specifically used for:

[0148] The location information and the target keywords are deduplicated;

[0149] The deduplicated results are then concatenated with keywords to generate the multiple recommended terms.

[0150] In one embodiment of this disclosure, the recommendation word determination module further includes a second processing module; the second processing module is configured to:

[0151] A set of recommended words within a preset time period is obtained, and a target recommended word is determined to be recommended to the terminal based on the multiple recommended words and the set of recommended words. The target recommended word is used to represent a personalized local word.

[0152] In one embodiment of this disclosure, the second processing module is specifically used for:

[0153] Based on the historical behavior data corresponding to the terminal, a plurality of first search keywords corresponding to the historical behavior data are determined;

[0154] In response to the triggering operation of the input event, the terminal obtains the current behavior data corresponding to the terminal and determines multiple second search keywords corresponding to the current behavior data;

[0155] The recommended word set is generated based on the first search keyword and the second search keyword.

[0156] In one embodiment of this disclosure, the second processing module is specifically used for:

[0157] From the plurality of recommended words, determine a plurality of recommended words with local features that match the location information;

[0158] The multiple recommended words with local features are added to the recommended word set, and the target recommended word set is obtained by deduplication.

[0159] The target recommendation word set is sorted by a preset recommendation model to obtain the sorted recommendation words; the preset recommendation model is a recommendation word recommendation model that is at least related to time and location.

[0160] The keywords to be recommended based on the preset ranking are used as the target recommendation keywords.

[0161] refer to Figure 5The diagram illustrates a hardware structure suitable for implementing embodiments of the present disclosure, which may be a terminal device or a server. The terminal device may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, personal digital assistants (PDAs), portable Android devices (PADs), portable media players (PMPs), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. Figure 5 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments disclosed herein.

[0162] like Figure 5 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 505 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device. The processing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0163] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic devices to exchange data via wireless or wired communication with other devices. Although Figure 5 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have instead.

[0164] In particular, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the methods of embodiments of this disclosure.

[0165] It should be noted that the computer-readable medium described in this disclosure can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example,—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this disclosure, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0166] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0167] The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the methods shown in the above-disclosed embodiments.

[0168] Computer program code for performing the operations of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed 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 remote computers, the remote computer can 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 can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0169] 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 this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated 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, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0170] The units described in the embodiments of this disclosure can be implemented in software or in hardware. The name of a unit does not necessarily limit the unit itself; for example, the first acquisition unit can also be described as "a unit that acquires at least two Internet Protocol addresses".

[0171] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0172] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0173] In a first aspect, embodiments of this disclosure provide a method for determining recommendation terms, the method comprising:

[0174] Obtain the location information of the terminal, and based on the location information of the terminal, obtain the target historical search behavior data corresponding to the location information;

[0175] The target's historical search behavior data is processed and analyzed to extract target keywords with local features; wherein, the local features are used to represent features associated with the location information;

[0176] The location information and the target keywords are combined to generate multiple recommended words.

[0177] According to one or more embodiments of this disclosure, the target historical search behavior data includes search behavior data corresponding to multiple search behavior sessions. The search behavior data corresponding to each search behavior session includes multiple search terms, interaction data corresponding to each search term, and search order corresponding to each search term. The interaction data and search order corresponding to the search terms are used to characterize the degree to which the search terms meet the search requirements.

[0178] According to one or more embodiments of this disclosure, the step of processing and analyzing the target historical search behavior data to extract target keywords with local characteristics includes:

[0179] Based on the multiple search terms corresponding to the search behavior sessions, the search behavior sessions are aggregated to obtain multiple sets of search behavior sessions;

[0180] For each set of search behavior sessions, the search terms in the set of search behavior sessions are deduplicated to obtain multiple target search terms;

[0181] Based on the interaction data and search order corresponding to each of the search terms, determine the interaction data and search order corresponding to each of the target search terms;

[0182] Based on the interaction data and search order corresponding to each target search term, candidate search terms are determined from multiple search behavior session sets;

[0183] Based on the candidate search terms, the target keywords are determined, including local keywords.

[0184] According to one or more embodiments of this disclosure, determining the target keyword based on the candidate search terms includes:

[0185] Search for search terms related to local content information from the candidate search terms;

[0186] Search terms related to local content information are used as the target keywords.

[0187] According to one or more embodiments of this disclosure, the method further includes:

[0188] Obtain media resource information;

[0189] Extract keywords including geographic location and / or local keywords from the media resource information;

[0190] The target keywords are the keywords of the geographical location and / or local keywords, which have local characteristics.

[0191] According to one or more embodiments of this disclosure,

[0192] The target keywords are multiple, and the location information includes historical location information and current location information; the location information and the target keywords are fused to generate multiple recommended words, including:

[0193] The location information and the target keywords are deduplicated;

[0194] The deduplicated results are then concatenated with keywords to generate the multiple recommended terms.

[0195] According to one or more embodiments of this disclosure, the method further includes:

[0196] A set of recommended words within a preset time period is obtained, and a target recommended word is determined to be recommended to the terminal based on the multiple recommended words and the set of recommended words. The target recommended word is used to represent a personalized local word.

[0197] According to one or more embodiments of this disclosure, obtaining the set of recommended words includes:

[0198] Based on the historical behavior data corresponding to the terminal, a plurality of first search keywords corresponding to the historical behavior data are determined;

[0199] In response to the triggering operation of the input event, the terminal obtains the current behavior data corresponding to the terminal and determines multiple second search keywords corresponding to the current behavior data;

[0200] The recommended word set is generated based on the first search keyword and the second search keyword.

[0201] According to one or more embodiments of this disclosure, determining the target recommended words to be recommended to the terminal based on the plurality of recommended words and the set of recommended words includes:

[0202] From the plurality of recommended words, determine a plurality of recommended words with local features that match the location information;

[0203] The multiple recommended words with local features are added to the recommended word set, and the target recommended word set is obtained by deduplication.

[0204] The target recommendation word set is sorted by a preset recommendation model to obtain the sorted recommendation words; the preset recommendation model is a recommendation word recommendation model that is at least related to time and location.

[0205] The keywords to be recommended based on the preset ranking are used as the target recommendation keywords.

[0206] Secondly, embodiments of this disclosure provide a device for determining recommended terms, comprising:

[0207] The first acquisition module is used to acquire the location information of the terminal, and based on the location information of the terminal, acquire the target historical search behavior data corresponding to the location information;

[0208] The second acquisition module is used to process and analyze the target's historical search behavior data and extract target keywords with local features; wherein, the local features are used to represent features associated with the location information;

[0209] The recommendation word generation module is used to merge the location information and the target keywords to generate multiple recommendation words.

[0210] According to one or more embodiments of this disclosure, the target historical search behavior data includes search behavior data corresponding to multiple search behavior sessions. The search behavior data corresponding to each search behavior session includes multiple search terms, interaction data corresponding to each search term, and search order corresponding to each search term. The interaction data and search order corresponding to the search terms are used to characterize the degree to which the search terms meet the search requirements.

[0211] According to one or more embodiments of this disclosure, the second acquisition module 402 is specifically used for:

[0212] Based on the multiple search terms corresponding to the search behavior sessions, the search behavior sessions are aggregated to obtain multiple sets of search behavior sessions;

[0213] For each set of search behavior sessions, the search terms in the set of search behavior sessions are deduplicated to obtain multiple target search terms;

[0214] Based on the interaction data and search order corresponding to each of the search terms, determine the interaction data and search order corresponding to each of the target search terms;

[0215] Based on the interaction data and search order corresponding to each target search term, candidate search terms are determined from multiple search behavior session sets;

[0216] Based on the candidate search terms, the target keywords are determined, including local keywords.

[0217] According to one or more embodiments of this disclosure,

[0218] The second acquisition module 402 is specifically used for:

[0219] Search for search terms related to local content information from the candidate search terms;

[0220] Search terms related to local content information are used as the target keywords.

[0221] According to one or more embodiments of this disclosure, the recommendation word determination apparatus further includes: a first processing module; the first processing module is configured to:

[0222] Obtain media resource information;

[0223] Extract keywords including geographic location and / or local keywords from the media resource information;

[0224] The target keywords are the keywords of the geographical location and / or local keywords, which have local characteristics.

[0225] According to one or more embodiments of this disclosure, the target keywords are multiple, and the location information includes historical location information and current location information; the recommendation word generation module 403 is specifically used for:

[0226] The location information and the target keywords are deduplicated;

[0227] The deduplicated results are then concatenated with keywords to generate the multiple recommended terms.

[0228] According to one or more embodiments of this disclosure, the recommendation word determination module further includes a second processing module; the second processing module is configured to:

[0229] A set of recommended words within a preset time period is obtained, and a target recommended word is determined to be recommended to the terminal based on the multiple recommended words and the set of recommended words. The target recommended word is used to represent a personalized local word.

[0230] According to one or more embodiments of this disclosure, the second processing module is specifically used for:

[0231] Based on the historical behavior data corresponding to the terminal, a plurality of first search keywords corresponding to the historical behavior data are determined;

[0232] In response to the triggering operation of the input event, the terminal obtains the current behavior data corresponding to the terminal and determines multiple second search keywords corresponding to the current behavior data;

[0233] The recommended word set is generated based on the first search keyword and the second search keyword.

[0234] According to one or more embodiments of this disclosure, the second processing module is specifically used for:

[0235] From the plurality of recommended words, determine a plurality of recommended words with local features that match the location information;

[0236] The multiple recommended words with local features are added to the recommended word set, and the target recommended word set is obtained by deduplication.

[0237] The target recommendation word set is sorted by a preset recommendation model to obtain the sorted recommendation words; the preset recommendation model is a recommendation word recommendation model that is at least related to time and location.

[0238] The keywords to be recommended based on the preset ranking are used as the target recommendation keywords.

[0239] Thirdly, embodiments of this disclosure provide an electronic device, including: at least one processor and a memory;

[0240] The memory stores computer-executed instructions;

[0241] The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the recommendation word determination method as described in the first aspect and various possible designs of the first aspect.

[0242] Fourthly, embodiments of this disclosure provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the recommendation word determination method described in the first aspect and various possible designs of the first aspect.

[0243] The above description is merely a preferred embodiment of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features disclosed in this disclosure that have similar functions.

[0244] Furthermore, while the operations are described in a specific order, this should not be construed as requiring these operations to be performed in the specific order shown or in a sequential order. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the above discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single embodiment. Conversely, various features described in the context of a single embodiment may also be implemented individually or in any suitable sub-combination in multiple embodiments.

[0245] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. A method for determining recommendation terms, characterized in that, include: Obtain the location information of the terminal, and based on the location information of the terminal, obtain the target historical search behavior data corresponding to the location information; The target's historical search behavior data is processed and analyzed to determine multiple search terms, and target keywords with local features are extracted from the multiple search terms; wherein, the local features are used to represent features associated with the location information; The location information and the target keywords with local characteristics are fused to generate multiple recommended words; a set of recommended words is obtained based on the multiple recommended words; the set of recommended words is merged into a recommendation recall set and participates in the ranking of the recommendation model; local words are pushed based on the location information; wherein, the recommendation recall set is determined in the following way: if the target user is a new user, the attribute corresponding to the target user is obtained, the attribute is generated by pushing information to the target user through cold start; a first keyword matching the attribute is obtained from the target keywords, and a first search result associated with the first keyword is obtained; if the target user is a historical user, user behavior and geographical location information are obtained based on search behavior trend statistics and after user authorization; a second keyword matching the user behavior and geographical location information is obtained from the target keywords, and a second search result associated with the second keyword is obtained; if the terminal used by the target user is detected to have a trigger operation for performing input events, the current behavior data of the target user is obtained after authorization by the target user, the current behavior data includes context information; target association information associated with the context information is obtained; the first keyword, the first search result, the second keyword, the second search result, and the target association information are combined to generate a recommendation recall set.

2. The method according to claim 1, characterized in that, The target historical search behavior data includes search behavior data corresponding to multiple search behavior sessions. The search behavior data corresponding to each search behavior session includes multiple search terms, interaction data corresponding to each search term, and search order corresponding to each search term. The interaction data and search order corresponding to the search terms are used to characterize the degree to which the search terms meet the search requirements.

3. The method according to claim 2, characterized in that, The process of processing and analyzing the target's historical search behavior data to identify multiple search terms, and extracting target keywords with local characteristics from these multiple search terms, includes: Based on the multiple search terms corresponding to the search behavior sessions, the search behavior sessions are aggregated to obtain multiple sets of search behavior sessions; For each set of search behavior sessions, the search terms in the set of search behavior sessions are deduplicated to obtain multiple target search terms; Based on the interaction data and search order corresponding to each of the search terms, determine the interaction data and search order corresponding to each of the target search terms; Based on the interaction data and search order corresponding to each target search term, candidate search terms are determined from multiple search behavior session sets; Based on the candidate search terms, the target keywords are determined, including local keywords.

4. The method according to claim 3, characterized in that, The process of determining the target keyword based on the candidate search terms includes: Search for search terms related to local content information from the candidate search terms; Search terms related to local content information are used as the target keywords.

5. The method according to claim 1 or 2, characterized in that, The method further includes: Obtain media resource information; Extract keywords including geographic location and / or local keywords from the media resource information; The target keywords are the keywords of the geographical location and / or local keywords, which have local characteristics.

6. The method according to claim 1, characterized in that, The target keywords are multiple, and the location information includes historical location information and current location information; the location information and the target keywords are fused to generate multiple recommended words, including: The location information and the target keywords are deduplicated; The deduplicated results are then concatenated with keywords to generate the multiple recommended terms.

7. A device for determining recommendation terms, characterized in that, include: The first acquisition module is used to acquire the location information of the terminal, and based on the location information of the terminal, acquire the target historical search behavior data corresponding to the location information; The second acquisition module is used to process and analyze the target historical search behavior data, determine multiple search terms, and extract target keywords with local features from the multiple search terms; wherein, the local features are used to represent features associated with the location information; A recommendation word generation module is used to fuse the location information and the target keywords with local features to generate multiple recommendation words; obtain a recommendation word set based on the multiple recommendation words; fuse the recommendation word set into a recommendation recall set to participate in the recommendation model ranking; and push local words based on the location information; wherein, the recommendation recall set is determined by the following methods: if the target user is a new user, then obtain the attribute corresponding to the target user, the attribute being generated by pushing information to the target user through cold start; obtain a first keyword matching the attribute from the target keywords, and obtain a first search result associated with the first keyword; if the target user is a historical user, then... Based on search behavior trend statistics, user behavior and geographic location information are obtained after user authorization; a second keyword matching the user behavior and geographic location information is obtained from the target keyword, and a second search result associated with the second keyword is obtained; if the terminal used by the target user is detected to have a trigger operation for performing input events, the target user's current behavior data, including context information, is obtained after user authorization; target association information associated with the context information is obtained; and a recommendation recall set is generated by combining the first keyword, the first search result, the second keyword, the second search result, and the target association information.

8. An electronic device, characterized in that, include: At least one processor and memory; The memory stores computer-executed instructions; The at least one processor executes computer execution instructions stored in the memory, causing the at least one processor to perform the recommendation word determination method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, implement the recommendation word determination method as described in any one of claims 1 to 6.

Images