Recommendation word processing method, electronic device and computer-readable storage medium

Abstract

The present application proposes a method for processing recommendation words, an electronic device, and a computer-readable storage medium. The method includes: obtaining relevant information of the current user who triggers the interactive interface, the relevant information at least including user behavior information; determining a reference user with similar behavior to the current user based on the relevant information; determining the operation object information corresponding to the reference user as a recommendation word with a behavior matching relationship with the relevant information; and providing the recommendation word to be displayed at an associated position in the interactive interface. By adopting the solution of the embodiment of the present application, the generated recommendation word is close to the user's interaction intention and personal habits, which can reduce the time cost of user interaction, reduce the complexity of operation, and effectively improve the conversion rate of user interaction.

Description

Technical Field

[0001] The present application relates to the field of cloud computing, and in particular to the field of cloud platform query technology. Background Art

[0002] Some websites provide search functions for users to search for the content they need. In order to guide users' browsing and purchasing intentions, one or more recommended words will be displayed to users when there is no input in the query input box to improve users' search efficiency.

[0003] The current recommendation scheme usually builds a vocabulary library based on popular lists and current website activities, selects recommended words from the vocabulary library in order and recommends them to users. The recommended words for different users are the same, which cannot meet the personalized query needs of some users. Summary of the invention

[0004] The present application provides a recommendation word processing method, an electronic device and a computer-readable storage medium to meet the personalized recommendation needs of different users.

[0005] In one aspect of the present application, a method for processing a recommendation word is provided, comprising:

[0006] Obtaining relevant information of the current user who triggered the interactive interface, wherein the relevant information at least includes user behavior information;

[0007] Determine, based on the relevant information, a reference user having similar behavior to the current user;

[0008] Determining the operation object information corresponding to the reference user as a recommendation word having a behavior matching relationship with the relevant information;

[0009] The recommendation word is provided for display at a relevant position in the interactive interface.

[0010] Another aspect of the present application provides a method for processing a recommendation word, comprising:

[0011] Obtaining relevant information of the current user who triggered the interactive interface, wherein the relevant information at least includes user behavior information;

[0012] Determining a recommendation word having a behavior matching relationship with the relevant information, the recommendation word including operation object information corresponding to a reference user having a similar behavior to the current user, the reference user being determined based on the relevant information;

[0013] The recommended word is provided at a related position in the interactive interface.

[0014] Another aspect of the present application provides a recommendation word processing device, including:

[0015] An information acquisition module, used to acquire relevant information of the current user who triggers the interactive interface, wherein the relevant information at least includes user behavior information;

[0016] A user determination module, configured to determine a reference user having similar behavior to the current user based on the relevant information;

[0017] A recommendation word determination module, used to determine the operation object information corresponding to the reference user as a recommendation word having a behavior matching relationship with the relevant information;

[0018] The recommendation word providing module is used to provide the recommendation word for display at a related position in the interactive interface.

[0019] Another aspect of the present application provides a recommendation word processing device, including:

[0020] An information acquisition module, used to acquire relevant information of the current user who triggers the interactive interface, wherein the relevant information at least includes user behavior information;

[0021] A recommendation word determination module, configured to determine a recommendation word having a behavior matching relationship with the relevant information, wherein the recommendation word includes operation object information corresponding to a reference user having a similar behavior to the current user, and the reference user is determined based on the relevant information;

[0022] The recommendation word providing module is used to provide the recommendation word at the associated position in the interactive interface.

[0023] In another aspect of the present application, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory, wherein the processor implements the method provided in any embodiment of the present application when executing the computer program.

[0024] In another aspect of the present application, a computer-readable storage medium is provided, in which a computer program is stored. When the computer program is executed by a processor, the method provided in any embodiment of the present application is implemented.

[0025] In another aspect of the present application, a computer program product is provided, including a computer program, which implements the method provided in any embodiment of the present application when executed by a processor.

[0026] Compared with the prior art, this application has the following advantages:

[0027] According to the embodiment of the present application, after the user who is currently performing the interaction triggers the interactive interface, relevant information of the current user, including at least user behavior information, is obtained, and a reference user with similar behavior to the current user is determined based on the relevant information, and the operation object information corresponding to the reference user is further determined as a recommendation word with a behavior matching relationship with the relevant information, and the above recommendation word is provided at a related position in the interactive interface. The above recommendation process is based on the behavior recommendation words of other users with similar behaviors to the current user. Behavioral similarity has a more explicit connection than the matching of text semantics, and is not easily subject to the current algorithm's understanding of the text. It has a better understanding effect on polysemous words and difficult-to-understand text sequences. Therefore, by adopting the solution of the embodiment of the present application, the generated recommendation words are close to the user's interactive intentions and personal habits, which can reduce the time cost of user interaction, reduce the cumbersomeness of operation, and effectively improve the conversion rate of user interaction. In addition, the above recommendation process refers to user behavior information, which can better reflect the user's personal interests than user personal information, and can further enhance the effect of recommendation.

[0028] In addition, the embodiments of the present application can also determine recommended words in combination with the semantic matching relationship of user behavior information, fully mine and utilize user behavior information, and can further improve the matching degree between the recommended words and the user's personal interests. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 A schematic diagram of an application scenario for implementing the method of an embodiment of the present application.

[0030] Figure 2 A schematic flowchart of a method for processing recommended words provided in an embodiment of the present application.

[0031] Figure 3 A schematic flowchart of a recommendation word processing method provided in another embodiment of the present application.

[0032] Figure 4 This is a structural block diagram of a recommendation word processing device provided in one embodiment of the present application.

[0033] Figure 5 This is a structural block diagram of a recommendation word processing device provided in another embodiment of the present application.

[0034] Figure 6 is a structural block diagram of an electronic device according to an embodiment of the present application. DETAILED DESCRIPTION

[0035] Many specific details are described in the following description to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the connotation of the present application, so the present application is not limited by the specific implementation disclosed below.

[0036] In order to more clearly demonstrate the recommendation word processing method provided in the embodiments of the present application, an application scenario that can be used to implement the method is first introduced.

[0037] The solution of the embodiment of the present application can be applied to various application scenarios based on interaction based on recommendation words. An entry for selecting or inputting recommendation words is provided on the interactive interface. The recommended recommendation words can be used as a reference for users to input information or to facilitate users to trigger interaction by directly clicking, thereby reducing user interaction operations and time consumption.

[0038] The recommended word processing method provided in the embodiment of the present application can be applied to the scene of cloud service search. Cloud service search refers to the cloud platform (that is, the cloud computing platform) providing a search portal through a page or a client. Software service providers or website operators, as business users of cloud services, can find the cloud service resources they need through the search portal. For example, a website operator with a need to create a website can search for various functional modules based on the search portal, such as various basic functional modules such as online consultation, video, e-commerce services, and member login, to achieve the construction of access systems such as web pages, applets, and clients corresponding to the computer or mobile phone. By building the above-mentioned access system through the cloud platform, the cost of purchasing physical hardware can be saved, the virtual resources of cloud computing can be used to complete the required computing requirements, and the efficiency of service construction can be improved, and various services can be accelerated to adopt advanced architectures, enable and elastically expand resources in sequence, and the security and confidentiality requirements of data can also be achieved through the cloud platform. Of course, it can be understood that the embodiment of the present application can also be applied to other search scenarios, such as searches initiated by the commodity search page of a trading website or the search homepage of a search engine.

[0039] Figure 1 The diagram is a schematic diagram of an exemplary application scenario for implementing the method of the embodiment of the present application. In the application scenario of cloud service search, the cloud platform provides search services, and the users are business users such as software service providers or website operators, and the operation objects corresponding to the user operation behaviors are cloud service resources. When the search interface is triggered, the client where the merchant user is located sends a recommendation request to the cloud platform, and the cloud platform extracts the relevant information of merchant user A carried in the recommendation request, which includes at least user behavior information, and may also include usage environment information, personal information, etc. The cloud platform further determines merchant user B who has similar behavior to merchant user A from the user behavior database based on the relevant information, and returns the operation object information corresponding to merchant user B as a recommendation word to the client where merchant customer A is located, and displays it as a background word in the background of the recommendation word input box in the search interface.

[0040] In the above recommendation process, the behavior recommendation words based on other users who have similar behaviors to a user have a more obvious connection than the matching of text semantics, and are not easily restricted by the current algorithm's understanding of text. They have a better understanding effect on polysemous words and difficult-to-understand text sequences. Therefore, by adopting the solution of the embodiment of the present application, the generated recommendation words are close to the user's interaction intention and personal habits, which can reduce the time cost of user interaction, reduce the cumbersomeness of operation, and effectively improve the conversion rate of user interaction. In addition, the above recommendation process refers to user behavior information, which can better reflect the user's personal interests than user personal information, and can further enhance the effect of recommendation.

[0041] The recommended word processing method provided by the present application can also be applied to voice interaction scenarios to provide users with a reference for voice input. The user can trigger the interactive interface through specific operations or voice wake-up, determine the recommended word based on relevant information including at least user behavior information, and further display the recommended word at a related position on the interactive interface, so that the user can perform voice input based on the recommended word and output the content corresponding to the recommended word.

[0042] The interactive interface involved in the embodiments of the present application may be a website page, a client page, or a device interface. In the search scenario, the interactive interface may specifically be the homepage or other pages of the search website or search client, or other websites including pages with search portals or pages of other clients; in the voice interaction scenario, the interactive interface may be the device interface of a user terminal or a voice interaction device. The interactive interface may be displayed on a user terminal or an electronic device, or may be displayed with the aid of virtual technologies such as virtual reality (VR) and augmented reality (AR).

[0043] In the embodiment of the present application, the user's related information may include user behavior information, which records the user's current or at least one historical time period, and may include behavior information within at least one time range before the current time point. For example, the behavior information within the time range from a certain time point to the current time point is taken as real-time behavior information, and the behavior information within other time ranges before the current time point is taken as historical behavior information, then the user behavior information may include one or more of the real-time behavior information and one or more historical behavior information.

[0044] User behavior information can record at least one type of information such as the operation type, operation time, operation location, and operation object corresponding to a specific operation of the user. Among them, the operation type is, for example, browsing behavior, clicking behavior, purchasing behavior, searching behavior, voice interaction behavior, etc. on the website. The user behavior information can correspond to one or more of the behavior information of the above-mentioned various operation types, that is, at least one of browsing behavior information, clicking behavior information, searching behavior information, interactive behavior information, and purchasing behavior information. The operation object corresponds to the page content browsed by the user on the website, the commodity object purchased, the commodity object searched, the interactive voice content or data object, etc. The operation object information is the relevant information corresponding to the operation object, which can include the name of the operation object, the classification of the operation object, the identification of the operation object (such as an abbreviation), etc.

[0045] User behavior information can be obtained from the record file corresponding to the client. In one implementation, the client captures the user behavior locally and records it in a cookie file (client record file), so the user behavior information can be read from the client's local cookie file; in another implementation, the user behavior information can be read from the server. The client page is embedded with JS (JavaScript) code. When the user accesses the page, the JS code will record the user behavior and send it to the server to generate an access log. Therefore, the user behavior information can be read from the client access log collected by the server.

[0046] In addition to user behavior information, relevant information may also include one or more of the current user's environmental information and personal information. Environmental information refers to information related to the user's network environment, time and space environment, etc. For example, information related to the network environment may be the context information of the current interactive interface, and information related to the time and space environment may be the current user's location information, time information, etc. Among them, the location information may be determined based on the location address of the positioning system in the terminal where the client is located or the network IP address. Personal information may include personal characteristics of the current user, such as the user's interests and hobbies, and personal attributes provided by the user.

[0047] Before using the above-obtained relevant information, in order to improve the efficiency of information processing and the effect of recommendation, the relevant information can be pre-processed. For example, data cleaning can be performed on the relevant information to remove redundant and irrelevant information, such as punctuation marks, meaningless words, garbled characters, etc.; for information that is too long, content can be deleted, one or more keywords representing key content can be retained, and only keywords can be used for subsequent determination of recommended words; data compliance verification can also be performed to delete data that does not meet the predetermined rules; or synonyms in the relevant information can be identified, and words with the same semantics can be merged to avoid subsequent meaningless repeated processing.

[0048] In an optional example, when the interactive interface is triggered, the client where the interactive interface is located can send a recommendation request to the server, and then extract the relevant information carried in the recommendation request. After determining the recommendation word, the recommendation word is added to the interactive interface for display. In another optional example, when the interactive interface is triggered, the recorded user-related information can be obtained from the server based on the user's personal information.

[0049] The user who currently triggers the interactive interface is the current user. Based on the relevant information, reference users with similar behaviors to the current user can be further determined. That is to say, the relevant information including the user behavior information characterizes the personal interest points of the user behavior. The reference user determined based on the relevant information has similar behaviors to the current user. The operation object information of the operation object corresponding to the reference user's behavior can be recommended to the current user as a recommendation word. It can be understood that the operation object information has a behavior matching relationship with the relevant information. For example, when the current user A searches for computing resources based on the client of the cloud platform, a reference user B with similar behaviors to the current user A is determined based on the relevant information of user A. It is identified from the behavior record of reference user B that "cloud server ECS" has been searched. Then, the "cloud server ECS" is further recommended to the current user A as the operation object information of reference user B.

[0050] When determining a reference user with similar behavior to the current user based on relevant information, the relevant information including user behavior information is used as a basis. In an optional example, a user who corresponds to the same operation object in the operation behavior as the current user can be used as a reference user. Specifically, the operation behavior of the current user and other users can be first obtained, and the operation objects that appear together in the same operation behavior can be found to form an operation object combination. Further, based on the corresponding operation behavior in the operation object combination, the users who appear together in the operation behavior corresponding to the operation object can be found as a user combination. The user combination includes the current user and one or more other users, and the one or more other users are also reference users. The current user and other users in the operation object combination have visited the operation objects in the operation object combination. Among them, the operation object combination includes two or more operation objects, which can be composed of two or more operation objects, or more than two operation objects can be composed of the operation object combination. Similarly, the user combination can include two or more users.

[0051] In another optional example, users who co-appear in the same operation behavior and correspond to the same operation object can also be found and added to the user combination, and one or more other users in the user combination are used as reference users of the current user, and the operation objects that co-appear in the same operation behavior, or the operation objects that co-appear in the operation behaviors corresponding to the users in the user combination, constitute the operation object combination. Of course, other arbitrary applicable methods can also be used to determine the operation object combination and the user combination, and this application does not limit this. In the screening of the operation object combination and the user combination, a corresponding threshold of the number of co-occurrences can also be set, and the co-occurrences exceeding the threshold are regarded as co-occurring operation objects or users.

[0052] Accordingly, when the operation object information corresponding to the reference user is determined as a recommendation word with a behavioral matching relationship with the relevant information, the graph structure data can be constructed based on the correspondence between the operation object combination and the user combination. The graph structure data is a data unit with points and edges as the structure. The operation objects and users are nodes in the graph structure data, and the probability of co-occurrence between nodes is the node connection, that is, the edge weight. Furthermore, based on the weight of the node connection, the similarity score of the operation object as a point in the graph structure data can be further calculated.

[0053] For example, according to the operation objects a and b corresponding to the current user A and the reference user B, respectively, a graph structure data consisting of four nodes is constructed. The weight of the node connection between the current user A and the reference user B can be the probability of the two appearing together in the same operation behavior, and the weight of the node connection between the operation object a and the operation object b can be the probability of the two appearing together in the same operation behavior. With respect to the similarity scores of the operation objects a and b obtained further, it can be understood that the operation object with a relatively high similarity score is closer to the personal habits of the non-corresponding users in the user combination, and can be used as a mutual recommendation object. Therefore, the information of the operation object with a high similarity score can be selected as the recommendation word.

[0054] The graph structure data composed of the operation object combination and the user combination can be extracted from the overall graph structure data composed of all operation object combinations and user combinations, or can be composed of a single pair of operation object combinations and user combinations.

[0055] Among them, the operation behaviors of multiple users can be captured and collected and collected into a corresponding user behavior database, and the users in the user behavior database can be used as a source of reference users to find reference users with similar behaviors to the current user from the user behavior database.

[0056] The above process of determining the recommended words provided to the current user based on the operation object information corresponding to the reference user can be implemented by a user-based collaborative filtering algorithm, which can be specifically constructed as a machine learning model. The combination of operation objects and the combination of users can constitute a graph structure data of a swing angle structure (swing shape), so that the calculation of similarity has higher noise resistance and accuracy. For example, a pair of operation objects and a pair of users can be formed into a quadrilateral structure, and the activity of the graph structure data is calculated as the similarity score of the operation object. Of course, other applicable models can also be selected according to actual needs, and the embodiments of the present application do not limit this.

[0057] When determining the recommended words based on the relevant information as mentioned above, the recommended words can also be determined from the perspective of the semantics represented by the relevant information. By combining it with the above-mentioned scheme of determining the recommended words for other users with similar behaviors based on the relevant information, the user behavior information can be further fully mined and utilized, thereby improving the matching degree between the recommended words and the user's personal interests.

[0058] Specifically, words with semantic matching relationships with relevant information can be searched as recommended words. The semantic matching relationship can be determined by calculating the matching relationship of semantic vectors. The semantic vector corresponding to the relevant information is determined by semantic encoding of the relevant information, and the words whose corresponding semantic vectors match the semantic vectors of the relevant information are further searched as recommended words. In other words, the semantic vectors of the recommended words match the semantic vectors of the relevant information. The matching between vectors can be represented by the similarity between the vectors. If the similarity exceeds the set threshold, it can be used as a matching vector.

[0059] Among them, semantic coding can be achieved through a semantic coding neural network. The semantic coding neural network can encode relevant information in text form into vectors, which can be obtained by training with a pre-trained DSSM twin-tower model (Deep Structured Semantic Model). The twin-tower model is easy to implement in engineering and has good system robustness. During the encoding process, the relevant information in text form is mapped to a low-dimensional vector space through the input layer and converted into a low-dimensional vector. The vector is further mapped to a high-dimensional space through a bag-of-words model in the representation layer. The vector mapped to the high-dimensional space is then converted into a high-dimensional semantic vector through multiple hidden layers, and the high-dimensional semantic vector is used as the vector corresponding to the relevant information. The matching between semantic vectors can be achieved by calculating the vector similarity through a semantic vector search system.

[0060] When searching for words whose corresponding semantic vectors match the semantic vectors of related information, if each word in the pre-configured corpus is matched, it will consume a large amount of computing resources and matching time, and the matching efficiency is low. In an optional example, the matching of semantic vectors can be achieved by using a vector recall method combined with a nearest neighbor fast search algorithm, and the words in the corpus are divided into different corpus sets according to the degree of similarity, so that similar corpora are placed in the same corpus set, and words that match the semantic vectors of related information are searched in each corpus set. After finding words that match the semantic vectors of related information in a corpus set, the next corpus set for searching for recommended words can be determined based on the matched words, so that a corpus set with a relatively high similarity can be selected for recommended word selection based on the similarity between the corpus set and the related information, for example, the next one or more corpus sets similar to the determined matching words are selected for recommended word selection, so that recommended words can be selected in a direction similar to the related information, and some corpus sets with a low similarity to the determined matching words are screened out.

[0061] In the process of selecting recommended words in a corpus collection, after determining a matching word, words with poor relevance to the matching word can be excluded, which can further save the computational complexity of the excluded words.

[0062] Recommended words that are determined based on the behavioral matching relationship and the semantic matching relationship with relevant information can be combined into a combination of recommended words. When combining, different weights can be configured for the behavioral matching recommended words and the semantic matching recommended words. Synonym merging and data compliance verification can also be performed to improve the efficiency of obtaining interactive results.

[0063] It is understandable that without obtaining the user behavior information of the current user, it is also possible to directly recommend words based on static information such as environmental information and personal information. For example, in the cloud service resource search scenario, a "website template" can be provided as a background word for new merchant users to facilitate merchants in website construction and service management.

[0064] When providing recommended words in the interactive interface, they can be displayed in the relevant position in the interactive interface, so that users can see the recommended words in time when searching, and further decide to use the recommended words, or enter related recommended words based on the recommended words, or enter other recommended words after determining that the recommended words do not meet the interactive intention. According to the current user's interactive operation based on the recommended words, interactive results based on the recommended words can be provided. For example, in the search scenario, if the user enters the recommended word, search results can be obtained based on the recommended word and provided to the current user's client for display; in the voice interaction scenario, if the user enters voice based on the recommended word, interactive content based on the voice can be provided.

[0065] The associated position in the interactive interface can be an associated position with the recommendation word input box, specifically, it can be the background of the query input box, and the recommendation word is displayed in the form of a background word of the recommendation word input box. The associated position can also be a position near the recommendation word input box, and the recommendation word is added to the page or displayed near the recommendation word input box in the form of a floating window or a new page element. The specific display form and position can be set according to actual needs, and this application does not limit this.

[0066] It should be noted that the above process of determining the reference user based on the relevant information of the current user and determining the recommended word based on the operation object information of the reference user can be implemented on the server or on the client. In order to obtain a better recommendation effect, the types of relevant information used can be more complex, more user behavior information can be collected in the user behavior database, and a more accurate algorithm can be used to determine similar users. Then, the above process can be implemented on the server or even the cloud server.

[0067] The embodiment of the present application also provides a recommendation word processing process from the client's perspective. After the current user triggers the interactive interface on the client, the client obtains the relevant information of the current user and further determines the recommendation words that have a behavioral matching relationship with the relevant information. This step can be implemented locally on the client, or by sending the relevant information to the server and obtaining the recommendation words determined by the server based on the relevant information, and further displaying the recommendation words at the associated position in the interactive interface. Among them, the relevant information includes at least user behavior information. In an optional example, the file that records the user behavior information on the client, such as a cookie file, can be accessed by accessing the associated file path of the client where the interactive interface is located, and the user behavior information can be extracted from it.

[0068] Figure 2 FIG. 1 is a schematic flow chart of a method for processing recommended words provided by an embodiment of the present application. The method may optionally be executed by a client or a server where the current user is located, but is not limited thereto. Figure 2 As shown, the method includes:

[0069] S210, obtaining relevant information of the current user who triggers the interactive interface, wherein the relevant information at least includes user behavior information;

[0070] S220, determining a reference user having similar behavior to the current user based on the relevant information;

[0071] S230, determining the operation object information corresponding to the reference user as a recommendation word having a behavior matching relationship with the relevant information;

[0072] S240: Provide the recommendation word to be displayed at a related position in the interactive interface.

[0073] In some optional embodiments, obtaining relevant information of the current user who triggers the interactive interface includes: extracting relevant information carried in the recommendation request sent when the interactive interface is triggered.

[0074] In some optional embodiments, the user behavior information includes at least one of browsing behavior information, clicking behavior information, searching behavior information, interactive behavior information, and purchasing behavior information, and the user behavior information includes behavior information within at least one time range before the current time point.

[0075] In some optional embodiments, determining a reference user having similar behavior to the current user based on the relevant information includes:

[0076] According to the operation behaviors of the current user and the reference user on the operation objects, searching for the operation objects that appear together in the same operation behavior to form an operation object combination;

[0077] Users that co-appear in the operation behaviors corresponding to the operation objects of the operation object combination are extracted to form a user combination, wherein the user combination includes the current user and a reference user.

[0078] In some optional embodiments, determining the operation object information corresponding to the reference user as a recommendation word having a behavior matching relationship with the relevant information includes:

[0079] Constructing graph structure data based on the correspondence between the operation object combination and the user combination, wherein the operation object and the user are nodes in the graph structure data, and the probability of common occurrence between the nodes is the weight of the node connection;

[0080] Determine the similarity score of the operation objects in the operation object combination based on the weights of the node connections;

[0081] The operation object information corresponding to the operation object with a higher similarity score ranking is selected as the recommendation word having a behavior matching relationship with the relevant information.

[0082] In some optional embodiments, the method further includes:

[0083] Determine a recommended word that has a semantic matching relationship with the relevant information.

[0084] In some optional embodiments, determining the recommended words having a semantic matching relationship with the relevant information includes:

[0085] Performing semantic encoding on the relevant information to obtain a semantic vector corresponding to the relevant information;

[0086] Search for a word whose corresponding semantic vector matches the semantic vector corresponding to the relevant information as a recommended word having a semantic matching relationship with the relevant information.

[0087] In some optional embodiments, the searching for a word whose corresponding semantic vector matches the semantic vector corresponding to the relevant information includes:

[0088] Divide the words in the corpus into different corpus sets according to their similarity;

[0089] Searching the corpus set for words that match the semantic vector corresponding to the relevant information, wherein the next corpus set for searching is determined according to the matching words in the previous corpus set.

[0090] In some optional embodiments, before determining the recommendation words having a semantic matching relationship with the relevant information, the method includes:

[0091] The relevant information is pruned to retain at least one keyword representing key content.

[0092] In some optional embodiments, the method further includes:

[0093] According to the interactive operation of the current user based on the recommended word, an interactive result based on the recommended word is provided.

[0094] In some optional embodiments, the relevant information also includes environmental information and / or personal information of the current user, the environmental information includes at least one of context information, time information, and location information of the interactive interface, and the personal information includes personal characteristics of the current user.

[0095] In some optional embodiments, the method further includes:

[0096] The relevant information is subjected to data preprocessing, wherein the data preprocessing includes at least one of data cleaning, data compliance verification, and similar data merging.

[0097] According to the embodiment of the present application, after the user who is currently performing the interaction triggers the interactive interface, relevant information of the current user, including at least user behavior information, is obtained, and a reference user with similar behavior to the current user is determined based on the relevant information, and the operation object information corresponding to the reference user is further determined as a recommendation word with a behavior matching relationship with the relevant information, and the above recommendation word is provided at a related position in the interactive interface. The above recommendation process is based on the behavior recommendation words of other users with similar behaviors to the current user. Behavioral similarity has a more explicit connection than the matching of text semantics, and is not easily subject to the current algorithm's understanding of the text. It has a better understanding effect on polysemous words and difficult-to-understand text sequences. Therefore, by adopting the solution of the embodiment of the present application, the generated recommendation words are close to the user's interactive intentions and personal habits, which can reduce the time cost of user interaction, reduce the cumbersomeness of operation, and effectively improve the conversion rate of user interaction. In addition, the above recommendation process refers to user behavior information, which can better reflect the user's personal interests than user personal information, and can further enhance the effect of recommendation.

[0098] Corresponding to the method provided in the above embodiment, the embodiment of the present application also provides a method for processing recommended words, which method can optionally be executed by the client where the current user is located, but is not limited to this. Figure 3 As shown, the method includes:

[0099] S310, obtaining relevant information of the current user who triggers the interactive interface, wherein the relevant information at least includes user behavior information;

[0100] S320, determining a recommendation word having a behavior matching relationship with the relevant information, the recommendation word including operation object information corresponding to a reference user having a similar behavior to the current user, the reference user being determined based on the relevant information;

[0101] S330: Provide the recommended word at a related position in the interactive interface.

[0102] In some optional embodiments, obtaining relevant information of the current user who triggers the interactive interface includes: reading user behavior information by accessing an associated file path of a client where the interactive interface is located.

[0103] According to the embodiment of the present application, after the user who is currently performing the interaction triggers the interactive interface, relevant information of the current user, including at least user behavior information, is obtained, and a reference user with similar behavior to the current user is determined based on the relevant information, and the operation object information corresponding to the reference user is further determined as a recommendation word with a behavior matching relationship with the relevant information, and the above recommendation word is provided at a related position in the interactive interface. The above recommendation process is based on the behavior recommendation words of other users with similar behaviors to the current user. Behavioral similarity has a more explicit connection than the matching of text semantics, and is not easily subject to the current algorithm's understanding of the text. It has a better understanding effect on polysemous words and difficult-to-understand text sequences. Therefore, by adopting the solution of the embodiment of the present application, the generated recommendation words are close to the user's interactive intentions and personal habits, which can reduce the time cost of user interaction, reduce the cumbersomeness of operation, and effectively improve the conversion rate of user interaction. In addition, the above recommendation process refers to user behavior information, which can better reflect the user's personal interests than user personal information, and can further enhance the effect of recommendation.

[0104] Corresponding to the application scenario and the method provided in the embodiment of the present application, the embodiment of the present application also provides a recommendation word processing device 400. Figure 4 , the apparatus 400 may include:

[0105] An information acquisition module 410 is used to acquire relevant information of the current user who triggers the interactive interface, wherein the relevant information at least includes user behavior information;

[0106] A user determination module 420, configured to determine a reference user having similar behavior to the current user based on the relevant information;

[0107] A recommendation word determination module 430 is used to determine the operation object information corresponding to the reference user as a recommendation word having a behavior matching relationship with the relevant information;

[0108] The recommendation word providing module 440 is used to provide the recommendation word for display at a related position in the interactive interface.

[0109] Optionally, the information acquisition module is specifically used to extract relevant information carried in the recommendation request sent when the interactive interface is triggered.

[0110] Optionally, the user behavior information includes at least one of browsing behavior information, clicking behavior information, searching behavior information, interactive behavior information, and purchasing behavior information, and the user behavior information includes behavior information within at least one time range before the current time point.

[0111] Optionally, the user determination module includes:

[0112] An object combination submodule, used to search for operation objects that appear together in the same operation behavior according to the operation behaviors of the current user and the reference user on the operation objects, and form an operation object combination;

[0113] The user combination submodule is used to extract users that co-appear in the operation behaviors corresponding to the operation objects of the operation object combination to form a user combination, wherein the user combination includes the current user and a reference user.

[0114] Optionally, the recommendation word determination module includes:

[0115] A graph construction submodule, for constructing graph structure data based on the correspondence between the operation object combination and the user combination, wherein the operation object and the user are nodes in the graph structure data, and the probability of common occurrence between the nodes is the weight of the node connection;

[0116] A similarity calculation submodule, used to determine the similarity scores of the operation objects in the operation object combination based on the weights of the node connections;

[0117] The object selection submodule is used to select the operation object information corresponding to the operation object with the highest similarity score as the recommendation word having a behavior matching relationship with the relevant information.

[0118] Optionally, the device further comprises:

[0119] The semantic recommendation module is used to determine the recommended words that have a semantic matching relationship with the relevant information.

[0120] Optionally, the semantic recommendation module includes:

[0121] A semantic vector acquisition submodule, used for semantically encoding the relevant information to obtain a semantic vector corresponding to the relevant information;

[0122] The word search submodule is used to search for words whose corresponding semantic vectors match the semantic vectors corresponding to the relevant information as recommended words having a semantic matching relationship with the relevant information.

[0123] Optionally, the word search submodule is specifically used to divide the words in the corpus into different corpus sets according to the degree of similarity; search for words that match the semantic vectors corresponding to the relevant information in the corpus set, wherein the next corpus set for search is determined based on the matching words in the previous corpus set.

[0124] Optionally, the device further comprises:

[0125] The content deletion module is used to delete the content of the relevant information before determining the recommended words that have a semantic matching relationship with the relevant information, so as to retain at least one keyword representing the key content.

[0126] Optionally, the device further comprises:

[0127] The interactive result providing module is used to provide an interactive result based on the recommended word according to the interactive operation of the current user based on the recommended word.

[0128] Optionally, the relevant information also includes environmental information and / or personal information of the current user, the environmental information includes at least one of context information, time information, and location information of the interactive interface, and the personal information includes personal characteristics of the current user.

[0129] Optionally, the device further comprises:

[0130] The preprocessing module is used to perform data preprocessing on the relevant information, wherein the data preprocessing includes at least one of data cleaning, data compliance verification, and similar data merging.

[0131] According to the embodiment of the present application, after the user who is currently performing the interaction triggers the interactive interface, relevant information of the current user, including at least user behavior information, is obtained, and a reference user with similar behavior to the current user is determined based on the relevant information, and the operation object information corresponding to the reference user is further determined as a recommendation word with a behavior matching relationship with the relevant information, and the above recommendation word is provided at a related position in the interactive interface. The above recommendation process is based on the behavior recommendation words of other users with similar behaviors to the current user. Behavioral similarity has a more explicit connection than the matching of text semantics, and is not easily subject to the current algorithm's understanding of the text. It has a better understanding effect on polysemous words and difficult-to-understand text sequences. Therefore, by adopting the solution of the embodiment of the present application, the generated recommendation words are close to the user's interactive intentions and personal habits, which can reduce the time cost of user interaction, reduce the cumbersomeness of operation, and effectively improve the conversion rate of user interaction. In addition, the above recommendation process refers to user behavior information, which can better reflect the user's personal interests than user personal information, and can further enhance the effect of recommendation.

[0132] This embodiment of the application also provides another recommended word processing device 500, referring to Figure 5 , the apparatus 500 may include:

[0133] An information acquisition module 510 is used to acquire relevant information of the current user who triggers the interactive interface, wherein the relevant information at least includes user behavior information;

[0134] A recommendation word determination module 520, configured to determine a recommendation word having a behavior matching relationship with the relevant information, wherein the recommendation word includes operation object information corresponding to a reference user having a similar behavior to the current user, and the reference user is determined based on the relevant information;

[0135] The recommendation word providing module 530 is used to provide the recommendation word at the associated position in the interactive interface.

[0136] Optionally, the information acquisition module is specifically used to read the user behavior information by accessing the associated file path of the client where the interactive interface is located.

[0137] The device provided in the embodiment of the present application can be used to implement the method provided in the embodiment of the present application and has corresponding beneficial effects.

[0138] An embodiment of the present application also provides an electronic device. Figure 6 The structural block diagram of an electronic device according to an embodiment of the present application is shown. The electronic device may be a recommendation word processing service device or a database engine device.

[0139] like Figure 6 As shown, the electronic device includes: a memory 610 and a processor 620. The memory 610 stores a computer program that can be run on the processor 620. When the processor 620 executes the computer program, the recommended word processing method in any embodiment of the present application is implemented. The number of the memory 610 and the processor 620 can be one or more.

[0140] like Figure 6 As shown, the electronic device also includes:

[0141] The communication interface 630 is used to communicate with external devices and perform data exchange transmission.

[0142] If the memory 610, the processor 620 and the communication interface 630 are implemented independently, the memory 610, the processor 620 and the communication interface 630 can be connected to each other through a bus and communicate with each other. The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 6 Only one thick line is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0143] Optionally, in a specific implementation, if the memory 610, the processor 620 and the communication interface 630 are integrated on a chip, the memory 610, the processor 620 and the communication interface 630 can communicate with each other through an internal interface.

[0144] An embodiment of the present application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the method provided in any embodiment of the present application.

[0145] An embodiment of the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the method provided in any embodiment of the present application.

[0146] An embodiment of the present application also provides a chip, which includes a processor for calling and executing instructions stored in the memory from the memory, so that a communication device equipped with the chip executes the method provided by the embodiment of the present application.

[0147] An embodiment of the present application also provides a chip, including: an input interface, an output interface, a processor and a memory, wherein the input interface, the output interface, the processor and the memory are connected via an internal connection path, and the processor is used to execute the code in the memory. When the code is executed, the processor is used to execute the method provided in the embodiment of the application.

[0148] It should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. It is worth noting that the processor may be a processor supporting the Advanced RISC Machines (ARM) architecture.

[0149] Further, optionally, the above-mentioned memory may include a read-only memory and a random access memory, and may also include a non-volatile random access memory. The memory may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may include a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may include a random access memory (RAM), which is used as an external cache. By way of example but not limitation, many forms of RAM are available. For example, static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), Sync Link DRAM (SLDRAM) and direct RAM bus random access memory (Direct Rambus RAM, DR RAM).

[0150] In the above embodiments, it can be implemented in whole or in part by software, hardware, firmware or any combination thereof. When implemented using software, it can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the process or function according to the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium.

[0151] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine different embodiments or examples described in this specification and the features of different embodiments or examples, unless they are contradictory.

[0152] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of the features. In the description of this application, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0153] Any process or method description in the flow chart or otherwise described herein can be understood to represent a module, fragment or portion of a code including one or more executable instructions for implementing the steps of a specific logical function or process. And the scope of the preferred embodiment of the present application includes other implementations, in which the functions may not be performed in the order shown or discussed, including in a substantially simultaneous manner or in a reverse order according to the functions involved.

[0154] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, which can be embodied in any computer-readable medium for use by an instruction execution system, apparatus or device (such as a computer-based system, a system including a processor or other system that can fetch instructions from an instruction execution system, apparatus or device and execute instructions), or used in combination with these instruction execution systems, apparatuses or devices.

[0155] It should be understood that the various parts of the present application can be implemented with hardware, software, firmware or a combination thereof. In the above embodiments, multiple steps or methods can be implemented with software or firmware stored in a memory and executed by a suitable instruction execution system. All or part of the steps of the above embodiment method can be completed by instructing the relevant hardware through a program, which can be stored in a computer-readable storage medium, and when the program is executed, it includes one of the steps of the method embodiment or a combination thereof.

[0156] In addition, each functional unit in each embodiment of the present application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into one module. The above-mentioned integrated module can be implemented in the form of hardware or in the form of a software functional module. If the above-mentioned integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. The storage medium can be a read-only memory, a disk or an optical disk, etc.

[0157] The above are only specific implementations of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of various changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A method for processing a recommendation word, comprising: Obtaining relevant information of the current user who triggered the interactive interface, wherein the relevant information at least includes user behavior information; Determine, based on the relevant information, a reference user having similar behavior to the current user; Determining the operation object information corresponding to the reference user as a recommendation word having a behavior matching relationship with the relevant information; Constructing graph structure data based on the correspondence between the operation object combination and the user combination, wherein the operation object and the user are nodes in the graph structure data, and the probability of common occurrence between the nodes is the weight of the node connection; Determine the similarity score of the operation objects in the operation object combination based on the weights of the node connections; Selecting the operation object information corresponding to the operation object with the highest similarity score as the recommendation word having a behavior matching relationship with the relevant information; providing the recommendation word for display at a relevant position in the interactive interface; The operation object combination includes operation objects that appear together in the same operation behavior of the current user and the reference user; the user combination includes users that appear together in the operation behaviors corresponding to the operation objects of the operation object combination.

2. The method according to claim 1, wherein: The acquisition of relevant information of the current user triggering the interactive interface includes: Extract relevant information carried in the recommendation request sent when the interactive interface is triggered.

3. The method according to claim 1, wherein: The user behavior information includes at least one of browsing behavior information, clicking behavior information, searching behavior information, interactive behavior information, and purchasing behavior information, and the user behavior information includes behavior information within at least one time range before the current time point.

4. The method according to claim 1, wherein: The determining, based on the relevant information, a reference user having similar behavior to the current user comprises: According to the operation behaviors of the current user and the reference user on the operation objects, searching for the operation objects that appear together in the same operation behavior to form an operation object combination; Users that co-appear in the operation behaviors corresponding to the operation objects of the operation object combination are extracted to form a user combination, wherein the user combination includes the current user and a reference user.

5. The method according to claim 1, wherein: The method further comprises: Determine a recommended word that has a semantic matching relationship with the relevant information.

6. The method according to claim 5, wherein: The determining of the recommended words having a semantic matching relationship with the relevant information includes: Performing semantic encoding on the relevant information to obtain a semantic vector corresponding to the relevant information; Search for a word whose corresponding semantic vector matches the semantic vector corresponding to the relevant information as a recommended word having a semantic matching relationship with the relevant information.

7. The method according to claim 6, wherein: The words whose corresponding semantic vectors are found to match the semantic vectors corresponding to the relevant information include: Divide the words in the corpus into different corpus sets according to their similarity; Searching the corpus set for words that match the semantic vector corresponding to the relevant information, wherein the next corpus set for searching is determined according to the matching words in the previous corpus set.

8. The method according to claim 1, wherein: Before determining the recommended words having a semantic matching relationship with the relevant information, the method includes: The relevant information is pruned to retain at least one keyword representing key content.

9. The method according to claim 1, wherein: The method further comprises: According to the interactive operation of the current user based on the recommended word, an interactive result based on the recommended word is provided.

10. The method according to claim 1, wherein: The relevant information also includes the environment information and / or personal information of the current user, the environment information includes at least one of the context information, time information, and location information of the interactive interface, and the personal information includes personal characteristics of the current user.

11. The method according to claim 1, wherein: The method further comprises: The relevant information is subjected to data preprocessing, wherein the data preprocessing includes at least one of data cleaning, data compliance verification, and similar data merging.

12. An electronic device comprising a memory, a processor and a computer program stored in the memory, wherein the processor implements the method according to any one of claims 1 to 11 when executing the computer program.

Images