Method and device for acquiring article information, method and device for acquiring knowledge graph and medium

Abstract

The invention discloses a method and device for obtaining article information, a method and device for obtaining a knowledge graph and a medium, and belongs to the field of computers and communication. The knowledge graph comprises a first query, first item information associated with the first query and second item information associated with the first item information, the method comprises the steps that a search request is received, the search request comprises a second query, and the first query and the second query are two similar queries; based on the second query and the knowledge graph, obtaining a first query similar to the second query; based on the first query and the knowledge graph, obtaining first item information having an association relationship with the first query; and based on the first article information and the knowledge graph, obtaining second article information having an association relationship with the first article information. According to the invention, generalization of article information acquisition can be improved.

Description

Technical Field

[0001] The present application relates to the field of computers and communications, and in particular to a method for obtaining object information, a method for obtaining a knowledge graph, a device and a medium. Background Art

[0002] The knowledge graph is applied to the item search scenario. The knowledge graph includes multiple queries and at least one item information corresponding to each query. The multiple queries include a first query. In the item search scenario, a second query to be searched by a user is received. The second query is similar to the first query. Based on the second query, the first query similar to the second query is obtained from the knowledge graph, and at least one item information corresponding to the first query is obtained, and the at least one item information is displayed to the user. For any item corresponding to the item information, the item information includes the link address of the conversion page of the item and / or the description information of the item, etc.

[0003] For example, a shopping website is an application for an item search scenario. Assume that the knowledge graph includes a telephone and item information of multiple mobile phones corresponding to the telephone. The multiple mobile phones are of different models and / or brands. The user inputs the second query to be searched as "telephone" to the shopping website. The shopping website receives the second query to be searched as "telephone", obtains the first query similar to the second query as "telephone" from the knowledge graph based on "telephone", obtains at least one item information corresponding to the first query including the item information of the multiple mobile phones, and displays the item information of the multiple mobile phones to the user. When the user clicks on the item information of a certain mobile phone, the shopping website obtains the item information of the mobile phone, and the item information of the mobile phone includes the link address of the transaction page of the mobile phone (a conversion page), displays the transaction page based on the link address, and the user can purchase the mobile phone on the transaction page.

[0004] After receiving the second query, the acquired item information is at least one item information corresponding to the first query similar to the second query, and the amount of acquired item information is small, resulting in low generalization of the acquired item information. Summary of the invention

[0005] The present application provides a method for obtaining item information, a method for obtaining a knowledge graph, a device and a medium to improve the generalization of obtaining item information. The technical solution is as follows:

[0006] In a first aspect, the present application provides a method for obtaining item information, the method is applied to a server, the server includes a knowledge graph, the knowledge graph includes a first query, first item information associated with the first query, and second item information associated with the first item information, the first item information is information of an item on which a first operation is performed among at least one item searched based on the first query, the first operation includes conversion and / or click, and the first item corresponding to the first item information and the second item corresponding to the second item information are items that can be converted at the same time. In the method, a search request is received, the search request includes a second query, and the first query and the second query are two similar queries. Based on the second query and the knowledge graph, a first query similar to the second query is obtained. Based on the first query and the knowledge graph, the first item information associated with the first query is obtained. Based on the first item information and the knowledge graph, the second item information associated with the first item information is obtained. Based on the first item information and the knowledge graph, the second item information associated with the first item information is obtained.

[0007] Since there is a first item information associated with the first query and a second item information associated with the first item information, not only can the first item information be obtained based on the second query and the knowledge graph included in the search request, but the second item information can also be obtained based on the first item information and the knowledge graph, thereby obtaining more item information and improving the generalization of the obtained item information.

[0008] In a possible implementation, the knowledge graph also includes group information of a user group, and there is an association relationship between the group information and the first query. The first query is a query searched by a user in the user group, and the search request also includes identification information of the first user, and the first user is a user in the user group. The user group to which the first user belongs is obtained based on the identification information of the first user. Based on the second query, the group information and the knowledge graph, a first query similar to the second query and having an association relationship with the group information is obtained. Since the first query searched by a user in the user group to which the first user belongs is obtained, the accuracy of obtaining the item information based on the first query can be improved, and the click-through rate of users clicking on the item information can be improved.

[0009] In another possible implementation, the attribute of the association relationship between the first item information and the second item information includes the group information, and the first item and the second item are items converted by users in the user group. Based on the first item information, the group information, and the knowledge graph, the second item information having an association relationship with the first item information and an attribute including the group information is obtained. Since the first item and the second item are items converted by users in the user group, this can improve the accuracy of obtaining the second item information and improve the click-through rate of users clicking on the second item information.

[0010] In another possible implementation, the attribute of the association relationship between the first item information and the second item information includes the identification information of the first user, the first item and the second item are items converted by the first user, and the search request also includes the identification information of the first user. Based on the first item information, the identification information, and the knowledge graph, the second item information having an association relationship with the first item information that includes the identification information is obtained. Since the first item and the second item are items converted by the first user, this can improve the accuracy of obtaining the second item information and improve the click-through rate of users clicking on the second item information.

[0011] In a second aspect, the present application provides a method for obtaining a knowledge graph, in which m query queries and n item information corresponding to the m queries are obtained, where m is an integer greater than 1, and n is an integer greater than m. The item information corresponding to the query is information about an item on which a first operation is performed among at least one item searched based on the query, and the first operation includes conversion and / or click. A knowledge graph is obtained based on the m queries and the n item information, and the knowledge graph includes the m queries and the n item information, and an association relationship exists between each query and the item information corresponding to each query. At least one item combination is obtained based on the knowledge graph, and each item combination includes two item information among the n item information, and the two item information are information about two items that can be converted at the same time. An association relationship is added between the two item information included in each item combination in the knowledge graph.

[0012] Since the two item information included in each item combination are information about two items that can be converted at the same time, an association relationship is added between the two item information included in each item combination in the knowledge graph. In this way, when obtaining item information through the knowledge graph, more item information can be obtained, thereby improving the generalization of the obtained item information.

[0013] In a possible implementation, the n item information includes first item information and second item information. Based on the knowledge graph, the first number of times the first item and the second item are converted simultaneously, the second number of times the first item is converted, and the third number of times the second item is converted, the first item is the item corresponding to the first item information, and the second item is the item corresponding to the second item information. Based on the first number, the second number, and the third number, the frequency information of the simultaneous conversion of the first item and the second item is obtained. When the frequency information meets the first condition, the first item information and the second item information are combined into a first item combination. Since the first item information and the second item information are combined into a first item combination when the frequency information of the simultaneous conversion of the first item and the second item meets the first condition, the association relationship between the two item information in the first item combination is increased, the accuracy of the knowledge graph can be improved, and the accuracy of obtaining item information can be improved based on the knowledge graph.

[0014] In another possible implementation, the frequency information includes one or more of the following: support between the first item and the second item, credibility of the first item to the second item, or promotion of the first item to the second item. The support is obtained by calculating the ratio between the first number and the total number of conversions of n items, where the n items are items corresponding to the n item information. The credibility is obtained by calculating the ratio between the first number and the second number. The promotion is obtained by calculating the ratio between the credibility and the third number.

[0015] In another possible implementation, when the support exceeds the support threshold, the credibility exceeds the credibility threshold, and the item category to which the first item belongs is the same as the item category to which the second item belongs, the first item information and the second item information are combined into a first item combination. Alternatively, when the support exceeds the support threshold, the credibility exceeds the credibility threshold, and the lift exceeds the lift threshold, the first item information and the second item information are combined into a first item combination. This can improve the accuracy of the first item combination, improve the accuracy of the knowledge graph, and improve the accuracy of obtaining item information based on the knowledge graph.

[0016] In another possible implementation, the m queries are queries searched by users in the same user group, and the attribute of the association relationship between the first item information and the second item information includes group information of the user group. In this way, when the user obtains item information, it is convenient to obtain item information based on the user group to which the user belongs, thereby improving the accuracy of obtaining item information.

[0017] In another possible implementation, based on the implicit association mining model, the third item information and the fourth item information with implicit association are identified from the knowledge graph, and the attribute of the third item corresponding to the third item information is related to the attribute of the fourth item corresponding to the fourth item information. The third item information and the fourth item information are combined into a second item combination. Since the attribute of the third item corresponding to the third item information is related to the attribute of the fourth item corresponding to the fourth item information, adding an association relationship between the two item information in the second item combination can improve the accuracy of the knowledge graph, and based on the knowledge graph, the accuracy and generalization of obtaining item information can be improved.

[0018] In another possible implementation, based on the knowledge graph, multiple query fragments of the same user are obtained, the query fragments include multiple queries searched by the user and the search time difference between any two adjacent queries is less than the time difference threshold. Multiple item information sets corresponding to the multiple query fragments are obtained, the item information sets corresponding to the query fragments include the item information corresponding to each query in the query fragment. Based on the multiple item information sets, the user's preference for simultaneously converting the fifth item corresponding to the fifth item information and the sixth item corresponding to the sixth item information is obtained, and the multiple item information sets include the fifth item information and the sixth item information. When the preference exceeds the preference threshold, the fifth item information and the sixth item information are combined into a third item combination.

[0019] Since the user's preference for converting the fifth item corresponding to the fifth item information and the sixth item corresponding to the sixth item information at the same time is obtained, when the preference exceeds the preference threshold, the fifth item information and the sixth item information are combined into a third item combination, and an association relationship is added between the two item information in the third item combination. This can improve the accuracy of the knowledge graph, and the accuracy of obtaining item information can be improved based on the knowledge graph.

[0020] In another possible implementation, the number of item information sets including the fifth item information and the sixth item information is counted; the ratio between the number of item information sets and the total number of the multiple item information sets is calculated to obtain the preference degree, thereby improving the accuracy of the obtained preference degree.

[0021] In another possible implementation, the attribute of the association relationship between the fifth item information and the sixth item information includes the identification information of the user. In this way, when the user obtains the item information, it is convenient to obtain the item information based on the identification information of the user, thereby improving the accuracy of obtaining the item information.

[0022] In another possible implementation, the m queries are queries searched by users in the same user group, and the knowledge graph also includes group information of the user group. There is an association relationship between the group information and the m queries.

[0023] In a third aspect, the present application provides a device for acquiring item information, which is used to execute the method in the first aspect or any possible implementation of the first aspect. Specifically, the device includes a unit for executing the method in the first aspect or any possible implementation of the first aspect.

[0024] In a fourth aspect, the present application provides a device for acquiring a knowledge graph, which is used to execute the method in the second aspect or any possible implementation of the second aspect. Specifically, the device includes a unit for executing the method in the second aspect or any possible implementation of the second aspect.

[0025] In a fifth aspect, the present application provides a computing device comprising at least one processor and a memory, wherein the at least one processor is used to couple with the memory, read and execute instructions in the memory, so as to implement the method in the first aspect or any possible implementation manner of the first aspect.

[0026] In a sixth aspect, the present application provides a computing device comprising at least one processor and a memory, wherein the at least one processor is used to couple with the memory, read and execute instructions in the memory, so as to implement the method in the second aspect or any possible implementation manner of the second aspect.

[0027] In the seventh aspect, the present application provides a computer program product, which includes a computer program stored in a computer-readable storage medium, and the computer program is loaded by a processor to implement the above-mentioned first aspect, second aspect, any possible implementation of the first aspect, or any possible implementation of the second aspect.

[0028] In an eighth aspect, the present application provides a computer-readable storage medium for storing a computer program, wherein the computer program is loaded by a processor to execute the method of the first aspect, the second aspect, any possible implementation of the first aspect, or any possible implementation of the second aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a structural diagram of a network architecture provided by an embodiment of the present application;

[0030] Figure 2 It is a schematic diagram of the structure of a knowledge graph provided in an embodiment of the present application;

[0031] Figure 3 It is a structural diagram of another network architecture provided by an embodiment of the present application;

[0032] Figure 4 It is a flow chart of a method for obtaining a knowledge graph provided in an embodiment of the present application;

[0033] Figure 5 It is a structural schematic diagram of a first knowledge graph provided in an embodiment of the present application;

[0034] Figure 6 is a schematic diagram of the structure of a second knowledge graph provided in an embodiment of the present application;

[0035] Figure 7 is a flow chart of a method for obtaining item information provided by an embodiment of the present application;

[0036] Figure 8 It is a schematic diagram of the structure of a device for obtaining item information provided in an embodiment of the present application;

[0037] Fig. 9 It is a schematic diagram of the structure of a device for obtaining a knowledge graph provided in an embodiment of the present application;

[0038] Fig.10 It is a schematic diagram of a device structure provided in an embodiment of the present application;

[0039] Fig.11 This is another schematic diagram of the device structure provided in an embodiment of the present application. DETAILED DESCRIPTION

[0040] The implementation methods of the present application will be further described in detail below with reference to the accompanying drawings.

[0041] See also Figure 1 The embodiment of the present application provides a network architecture 100, and the network architecture 100 includes: a client 101 and a server 102, and the client 101 can communicate with the server 102. Optionally, the number of clients 101 included in the network architecture 100 can be at least one.

[0042] The server 102 can obtain a knowledge graph, which includes a first query, first item information that is associated with the first query, and second item information that is associated with the first item information, wherein the first item information is information of an item on which a first operation is performed among at least one item searched based on the first query, the first operation includes a click and / or a conversion, and the first item corresponding to the first item information and the second item corresponding to the second item information are items that can be converted at the same time.

[0043] The client 101 may send a search request to the server 102 , where the search request includes a second query, and the first query and the second query are two similar queries.

[0044] The server 102 may receive the search request, and based on the second query and the knowledge graph, obtain a first query similar to the second query; based on the first query and the knowledge graph, obtain first item information associated with the first query; and based on the first item information and the knowledge graph, obtain second item information associated with the first item information.

[0045] Since the knowledge graph includes the first item information that is associated with the first query and the second item information that is associated with the first item information, when the server 102 receives the search request, in addition to being able to obtain the first item information that is associated with the first query information, it can also obtain the second item information that is associated with the first item information, thereby being able to obtain more item information and improving the generalization of the acquired item information.

[0046] In some embodiments, see Figure 2 , the first query, the first item information and the second item information in the knowledge graph are all different nodes of the knowledge graph.

[0047] For the first item information in the knowledge graph that is associated with the first query, the association between the first query and the first item information is represented by an edge connecting the first query and the first item information. Similarly, the association between other two nodes in the knowledge graph is also represented by an edge connecting the two nodes.

[0048] In some embodiments, for a node including the first query in the knowledge graph, the node may include information such as the search time of the first query in addition to the first query. The first item information includes a link address of a conversion page corresponding to the first item and / or description information of the first item, and the description information of the first item includes one or more of the following: the item category to which the first item belongs, the name, brand, color, price, supplier, model, size, or material of the first item, etc. Optionally, the item category to which the first item belongs includes a primary item category, a secondary item category, and / or a tertiary item category, etc.

[0049] In some embodiments, the nodes in the knowledge graph also include nodes such as group information of a user group and / or user information of a user. For a user belonging to the user group, there may be an association relationship between the group information of the user group and the user information of the user. For a query searched by the user, there may be an association relationship between the user information of the user and the query, and there may be an association relationship between the group information of the user group to which the user belongs and the query. Optionally, the user information of the user includes one or more of the following: the user's identification information, name, gender, or age, etc.

[0050] See also Figure 2 For the second item information that has an association relationship with the first item information, the second item information includes one or more of the following item information: explicitly associated item information that has an explicit association relationship with the first item information, implicitly associated item information that has an implicit association relationship with the first item information, or partially associated item information that has a partial association relationship with the first item information, etc.

[0051] The second item information is explicitly associated item information that has an explicit association relationship with the first item information, indicating that a user has converted the first item and the second item at the same time.

[0052] Optionally, the first item and the second item are items converted by users in the same user group, and there is an explicit association relationship between the first item information and the second item information, indicating that users in the user group frequently convert the first item and the second item at the same time.

[0053] In some embodiments, the knowledge graph may also include an attribute of an explicit association relationship between the first item information and the second item information, and the attribute of the explicit association relationship includes group information of the user group. Optionally, the group information of the user group may include one or more of the following: identification information of the user group, group name, or group category, etc.

[0054] In some embodiments, see Figure 2 The explicitly associated item information having an explicit association relationship with the first item information includes one or more of the following: similarly associated item information having a similar association relationship with the first item information, simply associated item information having a simple association relationship with the first item information, or causally associated item information having a causally associated relationship with the first item information, etc.

[0055] The second item information is similarly associated item information that has a similar association relationship with the first item information, indicating that the item category to which the first item corresponding to the first item information belongs is the same as the item category to which the second item corresponding to the second item information belongs. For example, the third-level item category to which the first item belongs is the same as the third-level item category to which the second item belongs, or the second-level item category to which the first item belongs is the same as the second-level item category to which the second item belongs, or the first-level item category to which the first item belongs is the same as the first-level item category to which the second item belongs.

[0056] The second item information is simply associated item information having a simple association relationship with the first item information, indicating that the first item corresponding to the first item information helps improve the conversion rate of the second item corresponding to the second item information, and vice versa.

[0057] The second item information is causally related item information that has a causal relationship with the first item information. Assuming that the first item information has a causal relationship with the second item information, the causal relationship is represented in the knowledge graph using a directed edge from the first item information to the second item information. The causal relationship means that the first item corresponding to the first item information helps improve the conversion rate of the second item corresponding to the second item information, but the second item corresponding to the second item information does not help improve the conversion rate of the first item corresponding to the first item information.

[0058] Among them, the second item information is implicitly associated item information that has an implicit association relationship with the first item information, indicating that there may not be a user who has converted the first item corresponding to the first item information and the second item corresponding to the second item information at the same time, but the attribute of the first item corresponding to the first item information is related to the attribute of the second item corresponding to the second item information. For example, the first item is a mobile phone, and the second item is a power bank. The first item belongs to the item category of electronic equipment, and the second item belongs to the item category of battery. Usually, electronic equipment includes batteries, so the attributes of the first item are related to the attributes of the second item, and there is an implicit association relationship between the item information of the two.

[0059] The second item information is locally associated item information that has a local association relationship with the first item information, indicating that the first item corresponding to the first item information and the second item corresponding to the second item information are two items that are simultaneously converted by the same user preference. Optionally, the knowledge graph may include attributes of the local association relationship, and the attributes of the local association relationship may include identification information of the user, etc.

[0060] In some embodiments, the server 102 includes registration information of a user registered in advance. Optionally, the registration information of the user includes one or more of the following information: identification information, age, gender, or region of the user.

[0061] In some embodiments, the search request sent by the user using the client 101 to the server 102 includes not only the second query but also the user's identification information. The server 102 can save the user's search log, which includes the user's identification information, the second query and the search time for the second query.

[0062] After searching for at least one item information based on the second query, the server 102 displays the at least one item information, and the user may click on a certain item information. For the item corresponding to the item information, the server 102 displays a conversion page for converting the item to the user. The search log also includes an item mapping relationship, and the server 102 adds a first record to the item mapping relationship included in the search log, and the first record includes the clicked item information. Optionally, the server 102 obtains the length of time the user stays on the conversion page of the item, and adds the length of time to the first record, that is, the first record may also include the length of time the user stays on the conversion page of the item.

[0063] When the user converts the item on the conversion page, the server 102 also adds a conversion mark in the first record, indicating that the item is converted by the user. Optionally, converting the item may be purchasing the item, or adding the item to a shopping cart. Alternatively, if the item is a virtual item (such as an application, an audio file, or a video file, etc.), the conversion item may be downloading the item.

[0064] Therefore, after the user uses the second query to search for at least one item information, the search log generated by the server 102 includes one or more of the following information: the user's identification information, the second query, and the item mapping relationship. The item mapping relationship includes at least one record, each record includes the item information clicked by the user and the user's stay time, and may also include a conversion mark, which is used to indicate that the user converted the item corresponding to the item information.

[0065] See also Figure 3, the network architecture 100 may include an application layer, a sorting layer, a recall layer, and a data layer, etc. The client 101 may be located in the application layer, and is used to provide a search interface. For example, the application layer may include a browser, a search bar, a panoramic search interface, or other search portals, etc. The sorting layer is used to sort the searched items and display the item links of the sorted items so that the user can click and open the conversion page of the item. The data layer includes the registration information registered by the user on the server 102, the search log generated by the server 102, and / or the item information on the server 102. The recall layer is used to obtain the knowledge graph based on the registration information, search log, and / or item information stored in the data layer.

[0066] In some embodiments, the client 101 may be a terminal device or an application client running on a terminal device. The terminal device may be a mobile phone, a personal computer, etc. The server 102 may be a server or a computing cluster, etc.

[0067] The network architecture 100 can be applied to an item search scenario, for example, the network architecture 100 can be applied to a shopping website or a search advertisement scenario, and the client 101 can be a shopping application client or a search client, etc.

[0068] See also Figure 4 , the present application embodiment provides a method 400 for obtaining a knowledge graph, the method 400 is applied to Figure 1 In the server 102 of the network architecture 100 shown, the method 400 includes the following process.

[0069] Step 401: obtain m queries and n item information corresponding to the m queries, where m is an integer greater than 1, and n is an integer greater than m. For each query, the item information corresponding to the query is information of an item on which a first operation is performed among at least one item searched based on the query, and the first operation includes conversion and / or click, etc.

[0070] For any user, for ease of explanation, the user is referred to as a first user, and the m queries include x queries searched by the first user, where x is an integer greater than or equal to 1 and less than or equal to m. The correspondence between the user information of the first user and the x queries may be stored in a first correspondence, and the correspondence between each query and the item information corresponding to each query may be stored in a second correspondence.

[0071] In some embodiments, in step 401, the user group to which the first user belongs may also be obtained, the correspondence between the user information of the first user and the group information of the user group may be saved in a third correspondence, and the correspondence between the group information of the user group and the x queries may be saved in a fourth correspondence.

[0072] In some embodiments, the first correspondence, the second correspondence, the third correspondence, and the fourth correspondence may be obtained through the following process 4011 to 4015 .

[0073] 4011: Get x search logs of the first user saved by the server.

[0074] The server saves a large number of search logs, and x search logs including identification information of the same user can be obtained from the search logs saved by the server. The user is called the first user, and the x search logs are the x search logs of the first user.

[0075] 4012: Obtain identification information of the first user, x queries, and z items of information corresponding to the x queries from the x search logs, where the n items of information include the z items of information, and z is an integer greater than or equal to x and less than or equal to n.

[0076] For each search log, a query included in the search log is obtained, and at least one item information included in the search log and clicked by the first user is obtained, and the at least one item information is used as the at least one item information corresponding to the query.

[0077] Optionally, the item information clicked by the first user and the dwell time corresponding to each item information included in the search log are obtained, and item information whose dwell time exceeds a time threshold is selected from the item information clicked by the first user, and at least one item information corresponding to the query includes the selected item information.

[0078] Optionally, for the item information whose retention time does not exceed the time threshold, the item information with a conversion mark is selected, and the at least one item information corresponding to the query also includes the selected item information with a conversion mark.

[0079] The x queries include queries obtained from each search log, and the z item information include item information corresponding to each query obtained based on each search log.

[0080] 4013: Obtain a user portrait of the first user, where the user portrait includes one or more of the following information: user information of the first user, sequence features of the first user, or numerical features of the first user.

[0081] In 4013, the registration information of the first user may be obtained from the server based on the identification information of the first user, and the user information of the first user may be obtained from the registration information of the first user.

[0082] In some embodiments, the sequence feature of the first user includes historical click behavior information of the first user and / or historical conversion row information of the first user. The historical click behavior information of the first user includes at least one item information clicked by the first user in the past N days and the dwell time corresponding to each item information, where N is an integer greater than 0. The historical conversion behavior information of the first user includes item information of at least one item converted by the first user in the past N days.

[0083] For the historical click behavior information of the first user, a search log with a search time in the past N days may be selected from the x search logs. Each item information clicked by the first user and the corresponding dwell time of each item information included in each selected search log are obtained to obtain the historical click behavior information of the first user.

[0084] For the historical conversion behavior information of the first user, a search log with a search time in the past N days may be selected from the x search logs. Information of each item marked with a conversion mark is obtained from each selected search log to obtain the historical conversion behavior information of the first user.

[0085] In some embodiments, the numerical feature of the first user includes the number of item information clicked by the first user in the past N days, and / or the number of item information converted by the first user in the past N days. Optionally, the number of item information included in the historical click behavior information of the first user may be counted to obtain the number of item information clicked by the first user in the past N days. The number of item information included in the historical conversion behavior information of the first user may be counted to obtain the number of item information converted by the first user in the past N days.

[0086] 4014: Based on the group clustering model and the user portrait of the first user, obtain the user group to which the first user belongs.

[0087] By repeatedly performing the above operations 4011 and 4013, user portraits of multiple users can be obtained, and then the group clustering model can be obtained by training the user portraits of the multiple users using an unsupervised machine learning method. Alternatively, for the multiple users, the technician can configure the user groups to which the multiple users belong, and form the group information and user portraits of the user groups of the multiple users into multiple training samples, each training sample including the group information and user portrait of the user group of one user, and train the group clustering model based on the multiple training samples.

[0088] In 4014, the user portrait of the first user is input into the group clustering model, the group clustering model obtains the group information of the user group of the first user based on the user portrait of the first user, and obtains the group information of the user group of the first user output by the group clustering model.

[0089] 4015: Save the correspondence between the user information of the first user and the x queries in a first correspondence, save the correspondence between each query in the x queries and the item information corresponding to each query in a second correspondence, save the correspondence between the user information of the first user and the group information of the user group of the first user in a third correspondence, and save the correspondence between the group information of the user group and the x queries in a fourth correspondence.

[0090] Repeat the above operations 4011-4015 to obtain the query searched by each other user, the item information, and the user group of each user, so that the first corresponding relationship saves the user information of multiple users and the queries searched by multiple users, the second corresponding relationship saves the multiple queries and the item information corresponding to the multiple queries, the third corresponding relationship saves the user information of multiple users and the group information of the user groups of multiple users, and the fourth corresponding relationship saves the group information of multiple user groups and the queries searched by users in the multiple user groups.

[0091] Step 402: A first knowledge graph is established based on the m queries and the n item information. The first knowledge graph includes the m queries and the n item information, and an association relationship exists between each query and the item information corresponding to each query.

[0092] In step 402, for each query, a first node including the query is established in the first knowledge graph, and at least one second node including at least one item information corresponding to the query is established in the first knowledge graph, each second node includes an item information, and an association relationship is added between the first node and each second node. Optionally, an edge is added between the first node and each second node to achieve adding an association relationship between the first node and each second node.

[0093] In some embodiments, the m queries include x queries searched by the first user, the first knowledge graph may also include user information of the first user and / or group information of the user group to which the first user belongs, there may be an association relationship between the user information of the first user and the x queries, there may be an association relationship between the user information of the first user and the group information of the user group, and there may be an association relationship between the group information of the user group and the x queries.

[0094] During implementation, a third node including user information of the first user and / or a fourth node including group information of the user group is established in the first knowledge graph, an association relationship is added between the third node and each first node, an association relationship is added between the fourth node and each first node, and an association relationship is added between the fourth node and the third node.

[0095] The first knowledge graph may also include user information of other users, group information of user groups to which other users belong, queries searched by other users, and at least one item information corresponding to each query searched by other users.

[0096] In some embodiments, for the multiple corresponding relationships obtained in step 401, a first knowledge graph can be established based on the multiple corresponding relationships, that is, the first knowledge graph is established based on the first corresponding relationship, the second corresponding relationship, the third corresponding relationship and the fourth corresponding relationship.

[0097] In implementation, a record including user information of a user and at least one query is obtained from the first corresponding relationship, the user information of the user and the at least one query are added to the first knowledge graph, and an association relationship is added between the user information of the user and the at least one query. For each query, at least one item information corresponding to the query is obtained from the second corresponding relationship, and the group information of the user group corresponding to the query is obtained from the fourth corresponding relationship, the group information of the user group and the at least one item information are added to the first knowledge graph, an association relationship is added between the query and the at least one item information, and an association relationship is added between the query and the group information of the user group. Based on the third corresponding relationship, it is determined whether the user group is the user group to which the user belongs. If the user group is the user group to which the user belongs, an association relationship is added between the user information of the user in the first knowledge graph and the group information of the user group.

[0098] See also Figure 5, the first knowledge graph may include group information 1 of a user group. User information of multiple users associated with the group information of the user group includes user information 1 of user 1 and user information 2 of user 2, multiple queries associated with the user information of user 1 include "baby products", "green vegetables" and "canned nuts", and multiple queries associated with the user information of user 2 include "diapers" and "wine to drink at home". At least one item information associated with "baby products" includes "body lotion" and "milk powder", at least one item information associated with "green vegetables" includes "green vegetables", at least one item information associated with "canned nuts" includes "walnuts", at least one item information associated with "diapers" includes "wet diapers", and at least one item information associated with "wine to drink at home" includes "beer" and "wine".

[0099] Since the first knowledge graph includes group information of user groups and user information of users, the content of the first knowledge graph is richer. Considering that user behaviors in the same user group have commonalities, grouping makes it convenient to mine global associations of items in the first knowledge graph according to user groups.

[0100] In some embodiments, the search time corresponding to each query is also saved in the first knowledge graph to facilitate mining of local associations of items based on the first knowledge graph.

[0101] Step 403: Acquire at least one item combination based on the first knowledge graph, each item combination includes two item information, and the two item information are information of two items in the first knowledge graph that can be converted simultaneously.

[0102] In step 403, the at least one item combination may be obtained in the following three ways, which are respectively the following way 1, way 2 and way 3.

[0103] Method 1: obtaining first item information and second item information having an explicit association relationship in a first knowledge graph, and combining the first item information and the second item information into a first item combination, wherein the at least one item combination includes the first item combination.

[0104] In some embodiments, method one can obtain the first item combination through the following process 4031-4033.

[0105] 4031: Based on the first knowledge graph, count the first number of times the first item and the second item are simultaneously converted, the second number of times the first item is converted, and the third number of times the second item is converted, where the first item is the item corresponding to the first item information, and the second item is the item corresponding to the second item information.

[0106] In 4031, for each user group's group information in the first knowledge graph, each query that is associated with the group information of the user group is obtained from the first knowledge graph, thereby obtaining multiple queries, and the multiple queries include each query that is associated with the group information of the user group. At least one item information associated with each query is obtained, thereby obtaining multiple item information, and the multiple item information includes at least one item information associated with each query.

[0107] The first item information and the second item information are any two items of the multiple items of information. At least one query that is associated with the first item information and the second item information is obtained from the multiple queries. The number of the at least one query is counted, and the number of the at least one query is used as the first number of simultaneous conversions of the first item and the second item.

[0108] At least one query associated with the first item information is obtained from the multiple queries, the number of the at least one query is counted, and the number of the at least one query is used as the second number of conversions of the first item.

[0109] At least one query associated with the second item information is obtained from the multiple queries, the number of the at least one query is counted, and the number of the at least one query is used as the third number of conversions of the second item.

[0110] 4032: Based on the first number, the second number and the third number, obtain frequency information of simultaneously transforming the first item and the second item.

[0111] The frequency information includes one or more of the following: support between the first item and the second item, credibility of the first item to the second item, promotion of the first item to the second item, or promotion of the second item to the first item;

[0112] In 4032, the support between the first item and the second item is obtained by calculating the ratio between the first number and the total number of conversions, where the total number of conversions is the sum of the number of times each of n items is converted, and the n items are items corresponding to the n item information; the credibility of the first item and the second item is obtained by calculating the ratio between the credibility of the first item and the second item and the third number; the improvement of the first item to the second item is obtained by calculating the ratio between the credibility of the first item and the second item and the second number; the improvement of the second item to the first item is obtained by calculating the ratio between the credibility of the first item and the second item and the second number.

[0113] 4033: When the frequency information satisfies the first condition, the first item information and the second item information are combined into a first item combination.

[0114] Whether the frequency information satisfies the first condition can be determined through the following process (11)-(14).

[0115] (11): Determine whether the support between the first item and the second item exceeds a support threshold, and whether the credibility between the first item and the second item exceeds a credibility threshold.

[0116] (12): If the support exceeds the support threshold and the credibility exceeds the credibility threshold, determine whether the item category of the first item and the item category of the second item are the same; if the item category of the first item and the item category of the second item are the same, execute (13); if the item category of the first item and the item category of the second item are different, execute (14).

[0117] The first item information includes an item category of the first item, and the second item information includes an item category of the second item. The item category here may be a tertiary item category, a secondary item category, or a primary item category.

[0118] (13): Combine the first item information and the second item information into a first item combination, and end.

[0119] When the item category of the first item and the item category of the second item are the same, it means that there is a similar association relationship between the first item information and the second item information.

[0120] (14): When the degree of improvement of the first item to the second item exceeds the improvement threshold, and / or when the degree of improvement of the second item to the first item exceeds the improvement threshold, the first item information and the second item information are combined into a first item combination.

[0121] When the degree of improvement of the first item to the second item exceeds the improvement threshold and the degree of improvement of the second item to the first item exceeds the improvement threshold, it indicates that there is a simple association relationship between the first item and the second item. When the degree of improvement of the first item to the second item exceeds the improvement threshold and the degree of improvement of the second item to the first item does not exceed the improvement threshold, it indicates that there is a causal association relationship between the first item and the second item. When the degree of improvement of the first item to the second item does not exceed the improvement threshold and the degree of improvement of the second item to the first item exceeds the improvement threshold, it indicates that there is a causal association relationship between the second item and the first item.

[0122] The first item combination corresponds to the user group, and the item information in the first item combination is the item information clicked by the users in the user group or the information of the items converted by the users.

[0123] By repeating the above process of 4031-4033, multiple first item combinations can be obtained.

[0124] Method 2: obtaining third item information and fourth item information having an implicit association relationship in the first knowledge graph, and combining the third item information and the fourth item information into a second item combination, wherein at least one item combination includes the second item combination.

[0125] In the second method, based on the implicit association mining model, the third item information and the fourth item information with implicit association are identified from the first knowledge graph, and the attribute of the third item corresponding to the third item information is related to the attribute of the fourth item corresponding to the fourth item information; the third item information and the fourth item information are combined into a second item combination.

[0126] Method three, obtaining fifth item information and sixth item information having a local association relationship in the first knowledge graph, and combining the fifth item information and the sixth item information into a third item combination, wherein the at least one item combination includes the third item combination.

[0127] In some embodiments, method three can obtain the first item combination through the following process 4131-4134.

[0128] 4131: Based on the first knowledge graph, multiple query fragments of the second user are obtained, for each query fragment, the query fragment includes multiple queries searched by the second user and the search time difference between any two adjacent queries is less than the time difference threshold, and the second user is the user corresponding to any user information in the first knowledge graph.

[0129] In 4131, user information of each user is obtained from the first knowledge graph, the user information of the user is referred to as the second user, t queries that are associated with the user information of the second user and the search time of each of the t queries are obtained from the first knowledge graph, and the t queries are divided into multiple query fragments, where t is an integer greater than 1.

[0130] 4132: Acquire multiple item information sets corresponding to the multiple query fragments. For each query fragment, the item information set corresponding to the query fragment includes item information corresponding to each query in the query fragment.

[0131] In 4132, for each query fragment, at least one item information associated with each query in the query fragment is obtained from the first knowledge graph, and the at least one item information associated with each query is formed into an item information set corresponding to the query.

[0132] 4133: Obtain the second user's preference for simultaneously converting a fifth item corresponding to the fifth item information and a sixth item corresponding to the sixth item information based on the multiple item information sets, wherein the multiple item information sets include the fifth item information and the sixth item information.

[0133] In 4133, the number of item information sets including the fifth item information and the sixth item information is counted, and the multiple item information sets include the fifth item information and the sixth item information; the ratio between the number of item information sets and the total number of the multiple item information sets is calculated to obtain the second user's preference for converting the fifth item and the sixth item at the same time.

[0134] 4134: When the preference exceeds the preference threshold, the fifth item information and the sixth item information are combined into a third item combination.

[0135] The third item combination corresponds to the second user and is two items that the second user prefers to convert at the same time. 4133 and 4134 may be repeatedly executed to obtain multiple third item combinations.

[0136] Step 404: Add an association relationship between two pieces of item information included in each item combination in the first knowledge graph to obtain a second knowledge graph.

[0137] For example, see Figure 6 For the first item combination, an association relationship is added between the first item information and the second item information included in the first item combination in the first knowledge graph. Optionally, the attribute of the association relationship includes group information of the user group corresponding to the first item combination.

[0138] If the first item information has a causal relationship with the second item information, the causal relationship between the first item information and the second item information is represented by an edge from the first item information to the second item information.

[0139] For the above-mentioned second item combination, an association relationship is added between the third item information and the fourth item information included in the second item combination in the first knowledge graph.

[0140] For the third item combination, an association relationship is added between the fifth item information and the sixth item information included in the third item combination in the first knowledge graph. Optionally, the attribute of the association relationship includes the identification information of the second user corresponding to the third item combination.

[0141] In an embodiment of the present application, the second knowledge graph includes at least one item information that has an association relationship with the query, and for each item information, the second knowledge graph also includes at least one other item information that has an association relationship with the item information. In this way, when searching for item information, based on the query, in addition to being able to search for item information that has an association relationship with the query, other item information that has an association relationship with the item information can also be searched. In this way, more item information can be searched, and the generalization of the search for item information is improved. In addition, the item information and the other item information may be item information of two different item categories, so that cross-category search for item information can be achieved.

[0142] See also Figure 7 The present application embodiment provides a method 700 for obtaining item information, wherein the method 700 is applied to Figure 1 In the network architecture 100 shown, the method 700 includes the following process.

[0143] Step 701: The client sends a search request to the server, where the search request includes a second query to be searched.

[0144] In some embodiments, the search request may also include identification information of the first user.

[0145] Step 702: The server receives the search request and obtains a first query similar to the second query based on the second query and the second knowledge graph.

[0146] The second knowledge graph can be Figure 4 The knowledge graph obtained by the method shown.

[0147] The first query is the query in the second knowledge graph.

[0148] In step 702, the server may use a text search method or a vector search method to obtain a first query similar to the second query from the second knowledge graph.

[0149] In some embodiments, the server can also obtain the user information of the first user from the saved registration information of each user based on the identification information of the first user included in the search request, and obtain the user group to which the first user belongs based on the user information of the first user. Based on the second query, the group information of the user group and the second knowledge graph, obtain a first query that is similar to the second query and has an association relationship with the group information.

[0150] In implementation, based on the group information of the user group, multiple queries that are associated with the group information of the user group are obtained from the second knowledge graph. Based on the second query, a text search method or a vector search method is used to obtain a first query similar to the second query from the multiple queries.

[0151] Optionally, based on the user information of the first user and the group clustering model, the user group to which the first user belongs is obtained. Alternatively, based on the user information of the first user, the user group to which the first user belongs is obtained from the correspondence between the user information and the user group stored in the third correspondence.

[0152] Step 703: The server obtains first item information that is associated with the first query based on the first query and the second knowledge graph.

[0153] In step 703, the server obtains first item information associated with the first query from the second knowledge graph based on the first query. The number of first item information associated with the first query in the second knowledge graph is at least one.

[0154] Step 704: The server obtains second item information that is associated with the first item information based on the first item information and the second knowledge graph.

[0155] The second item information associated with the first item information includes item information that has an explicit association with the first item information, item information that has an implicit association with the first item information, and / or item information that has a partial association with the first item information.

[0156] For item information that has an explicit association relationship with first item information, second item information that has a first association relationship with the first item information is obtained based on the first item information, group information of the user group and the second knowledge graph, and attributes of the first association relationship include the group information of the user group.

[0157] In some embodiments, the first association relationship may be a causal association relationship, and in the second knowledge graph, an edge from the first item information to the second item information is used to represent the causal association relationship. Based on the first item information, the group information of the user group, and the second knowledge graph, the second item information connected to the first item information using an edge is obtained, the edge is from the first item information to the second item information, and the attributes of the edge include the group information of the user group.

[0158] For item information that has a local association relationship with the first item information, second item information that has a second association relationship with the first item information is obtained based on the first item information, the identification information of the first user and the second knowledge graph, and the attributes of the second association relationship include the identification information of the first user.

[0159] In an embodiment of the present application, the second knowledge graph includes at least one first item information that has an association relationship with the query, and for each first item information, the second knowledge graph also includes at least one second item information that has an association relationship with the first item information. In this way, when a search request including the second query is received, in addition to being able to search for the first item information that has an association relationship with the first query based on the second query, the second item information that has an association relationship with the first item information can also be searched. In this way, more item information can be searched, and the generalization of the search for item information is improved. In addition, the first item information and the second item information may be item information of two different item categories, so that cross-category search for item information can be achieved.

[0160] See also Figure 8 The present application embodiment provides a device 800 for obtaining item information, wherein the device 800 is applied to Figure 1 In the server of the network architecture 100 shown, the device 800 includes a knowledge graph, the knowledge graph includes a first query, first item information associated with the first query, and second item information associated with the first item information, the first item information is information of an item on which a first operation is performed among at least one item searched based on the first query, the first operation includes conversion and / or click, the first item corresponding to the first item information and the second item corresponding to the second item information are items that can be converted at the same time, and the device 800 includes:

[0161] The receiving unit 801 is configured to receive a search request, where the search request includes a second query, and the first query and the second query are two similar queries;

[0162] The processing unit 802 is used to obtain a first query similar to the second query based on the second query and the knowledge graph;

[0163] The processing unit 802 is further configured to obtain first item information associated with the first query based on the first query and the knowledge graph;

[0164] The processing unit 802 is further configured to obtain second item information that is associated with the first item information based on the first item information and the knowledge graph.

[0165] Optionally, the detailed implementation process of receiving the search request by the receiving unit 801 is shown in Figure 7 The detailed implementation process of step 702 of the method 700 is not described in detail here.

[0166] Optionally, the detailed implementation process of the processing unit 802 obtaining the first query similar to the second query is shown in Figure 7 The detailed implementation process of step 702 of the method 700 is not described in detail here.

[0167] Optionally, the detailed implementation process of the processing unit 802 obtaining the first item information associated with the first query can be found in Figure 7 The detailed implementation process of step 703 of the method 700 is not described in detail here.

[0168] Optionally, the detailed implementation process of the processing unit 802 obtaining the second item information associated with the first item information can be found in Figure 7 The detailed implementation process of step 704 of the method 700 is not described in detail here.

[0169] Optionally, the knowledge graph further includes group information of a user group, the group information is associated with the first query, the first query is a query searched by a user in the user group, the search request further includes identification information of the first user, the first user is a user in the user group, and the processing unit 802 is configured to:

[0170] Acquire the user group to which the first user belongs based on the identification information of the first user;

[0171] Based on the second query, the group information and the knowledge graph, a first query that is similar to the second query and has an association relationship with the group information is obtained.

[0172] Optionally, the detailed implementation process of the processing unit 802 obtaining the first query which is similar to the second query and has an association relationship with the group information is as follows: Figure 7 The detailed implementation process of step 702 of the method 700 is not described in detail here.

[0173] Optionally, the attributes of the association relationship between the first item information and the second item information include the group information, the first item and the second item are items converted by users in the user group, and the processing unit 802 is used to obtain the second item information that has an association relationship with the first item information and has attributes including the group information based on the first item information, the group information and the knowledge graph.

[0174] Optionally, the detailed implementation process of the processing unit 802 obtaining the second item information having an attribute including the group information associated with the first item information is as follows: Figure 7 The detailed implementation process of step 704 of the method 700 is not described in detail here.

[0175] Optionally, the attributes of the association relationship between the first item information and the second item information include identification information of the first user, the first item and the second item are items converted by the first user, and the search request also includes the identification information of the first user. The processing unit 802 is used to obtain the second item information that has an association relationship with the first item information and has attributes including the identification information based on the first item information, the identification information and the knowledge graph.

[0176] Optionally, the detailed implementation process of the processing unit 802 obtaining the second item information having an attribute including the identification information associated with the first item information is as follows: Figure 7 The detailed implementation process of step 704 of the method 700 is not described in detail here.

[0177] In an embodiment of the present application, a receiving unit receives a search request, which includes a second query, and the first query and the second query are two similar queries. The processing unit obtains a first query similar to the second query based on the second query and the knowledge graph, obtains first item information that has an association relationship with the first query based on the first query and the knowledge graph, and obtains second item information that has an association relationship with the first item information based on the first item information and the knowledge graph. Since the first item information that has an association relationship with the first query and the second item information that has an association relationship with the first item information, the processing unit can not only obtain the first item information based on the second query and the knowledge graph included in the search request, but also obtain the second item information based on the first item information and the knowledge graph, so that more item information can be obtained and the generalization of the obtained item information is improved.

[0178] See also Fig. 9 The present application embodiment provides a device 900 for acquiring a knowledge graph, wherein the device 900 is applied to Figure 1 In the server of the network architecture 100 shown, the device 900 includes:

[0179] An acquisition unit 901 is configured to acquire m queries and n items of information corresponding to the m queries, where m is an integer greater than 1, and n is an integer greater than m. The item information corresponding to the query is information of an item on which a first operation is performed among at least one item searched based on the query, and the first operation includes conversion and / or click;

[0180] The acquisition unit 901 is further used to acquire a knowledge graph based on the m queries and the n item information, where the knowledge graph includes the m queries and the n item information, and there is an association relationship between each query and the item information corresponding to each query;

[0181] The acquisition unit 901 is further used to acquire at least one item combination based on the knowledge graph, each item combination includes two item information among the n item information, and the two item information are information of two items that can be converted at the same time;

[0182] The adding unit 902 is used to add an association relationship between two item information included in each item combination in the knowledge graph.

[0183] Optionally, the detailed implementation process of the acquisition unit 901 acquiring m queries and n item information corresponding to the m queries can be found in Figure 4 The detailed implementation process of step 401 of the method 400 is not described in detail here.

[0184] Optionally, the detailed implementation process of the acquisition unit 901 acquiring the knowledge graph based on the m queries and the n item information can be found in Figure 4 The detailed implementation process of step 402 of the method 400 is not described in detail here.

[0185] Optionally, the detailed implementation process of the acquisition unit 901 acquiring at least one item combination based on the knowledge graph is shown in Figure 4 The detailed implementation process of step 403 of the method 400 is not described in detail here.

[0186] Optionally, the detailed implementation process of adding unit 902 adding an association relationship between two items of information included in each item combination can be found in Figure 4 The detailed implementation process of step 404 of the method 400 is not described in detail here.

[0187] Optionally, the n items of information include first item information and second item information, and the acquiring unit 901 is configured to:

[0188] Based on the knowledge graph, count the first number of times the first item and the second item are simultaneously converted, the second number of times the first item is converted, and the third number of times the second item is converted, where the first item is the item corresponding to the first item information, and the second item is the item corresponding to the second item information;

[0189] Based on the first number, the second number, and the third number, obtaining frequency information of simultaneously transforming the first item and the second item;

[0190] When the frequency information satisfies the first condition, the first item information and the second item information are combined into a first item combination.

[0191] Optionally, the acquisition unit 901 counts the first number of times the first item and the second item are simultaneously converted, the second number of times the first item is converted, and the third number of times the second item is converted. Figure 4 The detailed implementation process of step 4031 of the method 400 is not described in detail here.

[0192] Optionally, the acquisition unit 901 acquires the frequency information of simultaneously converting the first item and the second item based on the first number, the second number and the third number. Figure 4 The detailed implementation process of step 4032 of the method 400 is not described in detail here.

[0193] Optionally, the detailed implementation process of the acquisition unit 901 combining the first item information and the second item information into a first item combination is shown in Figure 4 The detailed implementation process of step 4033 of the method 400 is not described in detail here.

[0194] Optionally, the frequency information includes one or more of the following: support between the first item and the second item, credibility of the first item to the second item, or promotion of the first item to the second item;

[0195] The acquisition unit 901 is used to:

[0196] The support is obtained by calculating the ratio between the first number and the total number of conversions of the n items, where the n items are the items corresponding to the n item information;

[0197] The confidence level is obtained by calculating the ratio between the first number and the second number;

[0198] The lift is calculated by calculating the ratio between the confidence and the third number.

[0199] Optionally, the acquisition unit 901 calculates the ratio between the first number and the total number of conversions of the n items to obtain the support degree. For a detailed implementation process, see Figure 4 The detailed implementation process of step 4032 of the method 400 is not described in detail here.

[0200] Optionally, the acquisition unit 901 calculates the ratio between the first number and the second number to obtain the credibility. Figure 4 The detailed implementation process of step 4032 of the method 400 is not described in detail here.

[0201] Optionally, the acquisition unit 901 calculates the ratio between the credibility and the third number to obtain the improvement degree. Figure 4 The detailed implementation process of step 4032 of the method 400 is not described in detail here.

[0202] Optionally, the acquiring unit 901 is configured to:

[0203] When the support exceeds the support threshold, the credibility exceeds the credibility threshold, and the item category to which the first item belongs and the item category to which the second item belongs are the same, the first item information and the second item information are combined into a first item combination; or

[0204] When the support exceeds a support threshold, the credibility exceeds a credibility threshold, and the lift exceeds a lift threshold, the first item information and the second item information are combined into a first item combination.

[0205] Optionally, the m queries are queries searched by users in the same user group, and the attribute of the association relationship between the first item information and the second item information includes group information of the user group.

[0206] Optionally, the acquiring unit 901 is configured to:

[0207] Based on the implicit association mining model, the third item information and the fourth item information with implicit association are identified from the knowledge graph, and the attribute of the third item corresponding to the third item information is related to the attribute of the fourth item corresponding to the fourth item information;

[0208] The third item information and the fourth item information are combined into a second item combination.

[0209] Optionally, the detailed implementation process of the acquisition unit 901 identifying the third item information and the fourth item information with implicit association from the knowledge graph is shown in Figure 4 The detailed implementation process of step 403 of the method 400 is not described in detail here.

[0210] Optionally, the detailed implementation process of the acquisition unit 901 combining the third item information and the fourth item information into a second item combination is shown in Figure 4 The detailed implementation process of step 403 of the method 400 is not described in detail here.

[0211] Optionally, the acquiring unit 901 is configured to:

[0212] Based on the knowledge graph, multiple query fragments of the same user are obtained, where the query fragments include multiple queries searched by the user and the search time difference between any two adjacent queries is less than the time difference threshold;

[0213] Obtain multiple item information sets corresponding to multiple query fragments, where the item information sets corresponding to the query fragments include item information corresponding to each query in the query fragments;

[0214] obtaining, based on a plurality of item information sets, a user's preference for simultaneously converting a fifth item corresponding to the fifth item information and a sixth item corresponding to the sixth item information; the plurality of item information sets including the fifth item information and the sixth item information;

[0215] When the preference exceeds the preference threshold, the fifth item information and the sixth item information are combined into a third item combination.

[0216] Optionally, the detailed implementation process of obtaining multiple query fragments of the same user by the obtaining unit 901 is shown in Figure 4 The detailed implementation process of step 4131 of the method 400 is not described in detail here.

[0217] Optionally, the detailed implementation process of the acquisition unit 901 acquiring multiple item information sets corresponding to multiple query fragments can be found in Figure 4 The detailed implementation process of step 4132 of the method 400 is not described in detail here.

[0218] Optionally, the acquisition unit 901 acquires the user's preference for simultaneously converting the fifth item corresponding to the fifth item information and the sixth item corresponding to the sixth item information based on the multiple item information sets. Figure 4 The detailed implementation process of step 4132 of the method 400 is not described in detail here.

[0219] Optionally, the detailed implementation process of the acquisition unit 901 combining the fifth item information and the sixth item information into a third item combination is shown in Figure 4 The detailed implementation process of step 4132 of the method 400 is not described in detail here.

[0220] Optionally, the acquiring unit 901 is configured to:

[0221] Counting the number of item information sets including the fifth item information and the sixth item information;

[0222] The preference degree is obtained by calculating the ratio between the number of the item information sets and the total number of the multiple item information sets.

[0223] Optionally, the detailed implementation process of the acquisition unit 901 counting the number of item information sets including the fifth item information and the sixth item information is as follows: Figure 4 The detailed implementation process of step 4133 of the method 400 is not described in detail here.

[0224] Optionally, the acquisition unit 901 calculates the ratio between the number of the item information set and the total number of the multiple item information sets to obtain the preference degree. Figure 4 The detailed implementation process of step 4133 of the method 400 is not described in detail here.

[0225] Optionally, the attribute of the association relationship between the fifth item information and the sixth item information includes user identification information.

[0226] Optionally, the m queries are queries searched by users in the same user group, and the knowledge graph also includes group information of the user group. There is an association relationship between the group information and the m queries.

[0227] In an embodiment of the present application, the acquisition unit acquires m queries and n item information corresponding to the m queries, acquires a first knowledge graph based on the m queries and the n item information, acquires at least one item combination based on the first knowledge graph, and adds an association relationship between two item information included in each item combination in the first knowledge graph to obtain a second knowledge graph. Since the two item information included in each item combination is information about two items that can be converted at the same time, an association relationship is added between the two item information included in each item combination in the first knowledge graph, so that when acquiring item information through the second knowledge graph, more item information can be acquired, thereby improving the generalization of acquired item information.

[0228] See also Fig.10 , the present application embodiment provides a schematic diagram of a device 1000. The device 1000 may be a server provided by any of the above embodiments, for example, the server may be Figure 1 The server in the network architecture 100 is shown. The device 1000 includes at least one processor 1001 , an internal connection 1002 , a memory 1003 and at least one transceiver 1004 .

[0229] The device 1000 is a hardware structure device that can be used to implement Figure 8 The functional modules in the device 800 are as follows. For example, those skilled in the art may think of Figure 8 The processing unit 802 in the device 800 shown can be implemented by the at least one processor 1001 calling the code in the memory 1003. Figure 8 The receiving unit 801 in the device 800 shown may be implemented by the at least one transceiver 1004 .

[0230] The device 1000 can also be used to implement the functions of the server in any of the above embodiments.

[0231] The processor 1001 may be a general-purpose central processing unit (CPU), a network processor (NP), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present application.

[0232] The internal connection 1002 may include a path to transmit information between the components. The internal connection 1002 may be a single board or a bus.

[0233] The at least one transceiver 1004 is used to communicate with other devices or communication networks.

[0234] The memory 1003 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may exist independently and be connected to the processor via a bus. The memory may also be integrated with the processor.

[0235] The memory 1003 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 1001. The processor 1001 is used to execute the application code stored in the memory 1003, and cooperate with at least one transceiver 1004, so that the device 1000 implements the functions of the method of the present patent.

[0236] In a specific implementation, as an embodiment, the processor 1001 may include one or more CPUs, such as Fig.10 CPU0 and CPU1 in.

[0237] In a specific implementation, as an embodiment, the device 1000 may include multiple processors, such as Fig.10 1001 and 1007 in the embodiment of the present invention. Each of these processors may be a single-CPU processor or a multi-CPU processor. The processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0238] See also Fig.11 , the present application embodiment provides a schematic diagram of a device 1100. The device 1100 may be a server provided by any of the above embodiments, for example, the server may be Figure 1 The server in the network architecture 100 is shown. The device 1100 includes at least one processor 1101 , an internal connection 1102 , a memory 1103 and at least one transceiver 1104 .

[0239] The device 1100 is a hardware structure device that can be used to implement Fig. 9 The functional modules in the device 900 are as follows. For example, those skilled in the art may think of Fig. 9 The obtaining unit 901 and the adding unit 902 in the device 900 shown can be implemented by the at least one processor 1101 calling the code in the memory 1103 .

[0240] The device 1100 can also be used to implement the functions of the server in any of the above embodiments.

[0241] The processor 1101 may be a general-purpose central processing unit (CPU), a network processor (NP), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits for controlling the execution of the program of the present application.

[0242] The internal connection 1102 may include a path to transmit information between the components. The internal connection 1102 may be a single board or a bus.

[0243] The at least one transceiver 1104 is used to communicate with other devices or communication networks.

[0244] The memory 1103 may be a read-only memory (ROM) or other types of static storage devices that can store static information and instructions, a random access memory (RAM) or other types of dynamic storage devices that can store information and instructions, or an electrically erasable programmable read-only memory (EEPROM), a compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical disc, laser disc, optical disc, digital versatile disc, Blu-ray disc, etc.), a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store the desired program code in the form of instructions or data structures and can be accessed by a computer, but is not limited thereto. The memory may exist independently and be connected to the processor via a bus. The memory may also be integrated with the processor.

[0245] The memory 1103 is used to store the application code for executing the solution of the present application, and the execution is controlled by the processor 1101. The processor 1101 is used to execute the application code stored in the memory 1103, and cooperate with at least one transceiver 1104, so that the device 1100 implements the functions of the method of the present patent.

[0246] In a specific implementation, as an embodiment, the processor 1101 may include one or more CPUs, such as Fig.11 CPU0 and CPU1 in.

[0247] In a specific implementation, as an embodiment, the device 1100 may include multiple processors, such as Fig.11 1 and 1107. Each of these processors may be a single-CPU processor or a multi-CPU processor. A processor herein may refer to one or more devices, circuits, and / or processing cores for processing data (e.g., computer program instructions).

[0248] A person skilled in the art will understand that all or part of the steps to implement the above embodiments may be accomplished by hardware or by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, and the above-mentioned storage medium may be a read-only memory, a disk or an optical disk, etc.

[0249] The above description is only an optional embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for obtaining item information, characterized in that: The method is applied to a server, the server includes a knowledge graph, the knowledge graph includes a first query, first item information associated with the first query, and second item information associated with the first item information, the first item information is information of an item on which a first operation is performed among at least one item searched based on the first query, the first operation includes conversion and / or click, the first item corresponding to the first item information and the second item corresponding to the second item information are items that can be converted at the same time, the method includes: receiving a search request, wherein the search request includes a second query, wherein the first query and the second query are two similar queries; Based on the second query and the knowledge graph, obtaining the first query similar to the second query; Based on the first query and the knowledge graph, obtaining the first item information associated with the first query; Based on the first item information and the knowledge graph, the second item information that is associated with the first item information is obtained.

2. The method according to claim 1, characterized in that The knowledge graph further includes group information of a user group, there is an association relationship between the group information and the first query, the first query is a query searched by a user in the user group, the search request further includes identification information of a first user, the first user is a user in the user group, and the method further includes: Acquire the user group to which the first user belongs based on the identification information of the first user; The acquiring, based on the second query and the knowledge graph, the first query that is similar to the second query includes: Based on the second query, the group information and the knowledge graph, the first query that is similar to the second query and has an association relationship with the group information is obtained.

3. The method according to claim 2, characterized in that The attribute of the association relationship between the first item information and the second item information includes the group information, the first item and the second item are items converted by users in the user group, and the acquiring, based on the first item information and the knowledge graph, the second item information that has an association relationship with the first item information includes: Based on the first item information, the group information and the knowledge graph, the second item information having an association relationship with the first item information including attributes of the group information is obtained.

4. The method according to claim 1 or 2, characterized in that: The attribute of the association relationship between the first item information and the second item information includes identification information of a first user, the first item and the second item are items converted by the first user, the search request also includes identification information of the first user, and acquiring the second item information that has an association relationship with the first item information based on the first item information and the knowledge graph includes: Based on the first item information, the identification information and the knowledge graph, the second item information having an association relationship with the first item information including attributes of the identification information is obtained.

5. A method for obtaining a knowledge graph, characterized in that: The method comprises: Obtaining m queries and n items of information corresponding to the m queries, where m is an integer greater than 1, and n is an integer greater than m, and the item information corresponding to the query is information of an item on which a first operation is performed among at least one item searched based on the query, and the first operation includes conversion and / or click; Acquire a knowledge graph based on the m queries and the n item information, wherein the knowledge graph includes the m queries and the n item information, and each query and the item information corresponding to each query are respectively associated with each other; Acquire at least one item combination based on the knowledge graph, each item combination including two item information among the n item information, and the two item information are information of two items that can be converted at the same time; An association relationship is added between two pieces of item information included in each item combination in the knowledge graph.

6. The method according to claim 5, characterized in that The n item information includes first item information and second item information, and acquiring at least one item combination based on the knowledge graph includes: Based on the knowledge graph, counting a first number of times a first item and a second item are simultaneously converted, a second number of times the first item is converted, and a third number of times the second item is converted, wherein the first item is an item corresponding to the first item information, and the second item is an item corresponding to the second item information; Based on the first number, the second number, and the third number, obtaining frequency information of simultaneously transforming the first item and the second item; When the frequency information satisfies a first condition, the first item information and the second item information are combined into a first item combination.

7. The method according to claim 6, characterized in that The frequency information includes one or more of the following: support between the first item and the second item, credibility of the first item to the second item, or promotion degree of the first item to the second item; The acquiring frequency information of simultaneously transforming the first item and the second item based on the first number, the second number, and the third number includes: The support degree is obtained by calculating a ratio between the first number of times and a total number of conversions of n items, where the n items are items corresponding to the n item information; Calculating a ratio between the first number of times and the second number of times to obtain the credibility; The improvement is obtained by calculating the ratio between the confidence level and the third tertiary coefficient.

8. The method according to claim 7, characterized in that When the frequency information satisfies a first condition, combining the first item information and the second item information into a first item combination includes: When the support exceeds a support threshold, the credibility exceeds a credibility threshold, and the item category to which the first item belongs is the same as the item category to which the second item belongs, the first item information and the second item information are combined into a first item combination; or When the support exceeds a support threshold, the credibility exceeds a credibility threshold, and the lift exceeds a lift threshold, the first item information and the second item information are combined into a first item combination.

9. The method according to any one of claims 6 to 8, characterized in that: The m queries are queries searched by users in the same user group, and the attribute of the association relationship between the first item information and the second item information includes group information of the user group.

10. The method according to any one of claims 5 to 9, characterized in that: The acquiring at least one item combination based on the knowledge graph includes: Based on the implicit association mining model, identifying third item information and fourth item information having an implicit association from the knowledge graph, wherein an attribute of the third item corresponding to the third item information is related to an attribute of the fourth item corresponding to the fourth item information; The third item information and the fourth item information are combined into a second item combination.

11. The method according to any one of claims 5 to 10, characterized in that: The acquiring at least one item combination based on the knowledge graph includes: Based on the knowledge graph, multiple query fragments of the same user are obtained, where the query fragments include multiple queries searched by the user and the search time difference between any two adjacent queries is less than a time difference threshold; Acquire multiple item information sets corresponding to the multiple query fragments, wherein the item information sets corresponding to the query fragments include item information corresponding to each query in the query fragments; obtaining, based on the multiple item information sets, a preference of the user for simultaneously converting a fifth item corresponding to the fifth item information and a sixth item corresponding to the sixth item information, wherein the multiple item information sets include the fifth item information and the sixth item information; When the preference exceeds a preference threshold, the fifth item information and the sixth item information are combined into a third item combination.

12. The method according to claim 11, characterized in that The obtaining, based on the multiple item information sets, the user's preference for simultaneously converting the fifth item corresponding to the fifth item information and the sixth item corresponding to the sixth item information includes: Counting the number of item information sets including the fifth item information and the sixth item information; The preference degree is obtained by calculating a ratio between the number of the item information sets and the total number of the plurality of item information sets.

13. The method according to claim 11 or 12, characterized in that The attribute of the association relationship between the fifth item information and the sixth item information includes identification information of the user.

14. The method according to any one of claims 5 to 13, characterized in that: The m queries are queries searched by users in the same user group. The knowledge graph also includes group information of the user group. There is an association relationship between the group information and the m queries.

15. A device for obtaining item information, characterized in that: The device includes a knowledge graph, the knowledge graph includes a first query, first item information associated with the first query, and second item information associated with the first item information, the first item information is information of an item on which a first operation is performed among at least one item searched based on the first query, the first operation includes conversion and / or click, a first item corresponding to the first item information and a second item corresponding to the second item information are items that can be converted at the same time, and the device includes: A receiving unit, configured to receive a search request, wherein the search request includes a second query, and the first query and the second query are two similar queries; A processing unit, configured to obtain the first query similar to the second query based on the second query and the knowledge graph; The processing unit is further configured to obtain the first item information associated with the first query based on the first query and the knowledge graph; The processing unit is further used to obtain the second item information that has an association relationship with the first item information based on the first item information and the knowledge graph.

16. The device according to claim 15, characterized in that The knowledge graph further includes group information of a user group, there is an association relationship between the group information and the first query, the first query is a query searched by users in the user group, the search request further includes identification information of a first user, the first user is a user in the user group, and the processing unit is used to: Acquire the user group to which the first user belongs based on the identification information of the first user; Based on the second query, the group information and the knowledge graph, the first query that is similar to the second query and has an association relationship with the group information is obtained.

17. The device according to claim 16, characterized in that The attributes of the association relationship between the first item information and the second item information include the group information, the first item and the second item are items converted by users in the user group, and the processing unit is used to obtain the second item information that has an association relationship with the first item information and has attributes including the group information based on the first item information, the group information and the knowledge graph.

18. The device according to claim 15 or 16, characterized in that The attributes of the association relationship between the first item information and the second item information include identification information of the first user, the first item and the second item are items converted by the first user, the search request also includes identification information of the first user, and the processing unit is used to obtain the second item information that has an association relationship with the first item information and has attributes including the identification information based on the first item information, the identification information and the knowledge graph.

19. A device for obtaining a knowledge graph, characterized in that: The device comprises: an acquisition unit, configured to acquire m queries and n items of information corresponding to the m queries, where m is an integer greater than 1, and n is an integer greater than m, and the item information corresponding to the query is information of an item on which a first operation is performed among at least one item searched based on the query, and the first operation includes conversion and / or click; The acquisition unit is further used to acquire a knowledge graph based on the m queries and the n item information, wherein the knowledge graph includes the m queries and the n item information, and there is an association relationship between each query and the item information corresponding to each query; The acquisition unit is further used to acquire at least one item combination based on the knowledge graph, each item combination includes two item information among the n item information, and the two item information are information of two items that can be converted at the same time; An adding unit is used to add an association relationship between two item information included in each item combination in the knowledge graph.

20. The device according to claim 19, characterized in that The n items of item information include first item information and second item information, and the acquiring unit is used to: Based on the knowledge graph, counting a first number of times a first item and a second item are simultaneously converted, a second number of times the first item is converted, and a third number of times the second item is converted, wherein the first item is an item corresponding to the first item information, and the second item is an item corresponding to the second item information; Based on the first number, the second number, and the third number, obtaining frequency information of simultaneously transforming the first item and the second item; When the frequency information satisfies a first condition, the first item information and the second item information are combined into a first item combination.

21. The device according to claim 20, characterized in that The frequency information includes one or more of the following: support between the first item and the second item, credibility of the first item to the second item, or promotion degree of the first item to the second item; The acquisition unit is used to: The support degree is obtained by calculating a ratio between the first number of times and a total number of conversions of n items, where the n items are items corresponding to the n item information; Calculating a ratio between the first number of times and the second number of times to obtain the credibility; The improvement is obtained by calculating the ratio between the confidence level and the third tertiary coefficient.

22. The device according to claim 21, characterized in that The acquisition unit is used to: When the support exceeds a support threshold, the credibility exceeds a credibility threshold, and the item category to which the first item belongs is the same as the item category to which the second item belongs, the first item information and the second item information are combined into a first item combination; or When the support exceeds a support threshold, the credibility exceeds a credibility threshold, and the lift exceeds a lift threshold, the first item information and the second item information are combined into a first item combination.

23. The device according to any one of claims 20 to 22, characterized in that The m queries are queries searched by users in the same user group, and the attribute of the association relationship between the first item information and the second item information includes group information of the user group.

24. The device according to any one of claims 19 to 23, characterized in that The acquisition unit is used to: Based on the implicit association mining model, identifying third item information and fourth item information having an implicit association from the knowledge graph, wherein an attribute of the third item corresponding to the third item information is related to an attribute of the fourth item corresponding to the fourth item information; The third item information and the fourth item information are combined into a second item combination.

25. The device according to any one of claims 19 to 24, characterized in that The acquisition unit is used to: Based on the knowledge graph, multiple query fragments of the same user are obtained, where the query fragments include multiple queries searched by the user and the search time difference between any two adjacent queries is less than a time difference threshold; Acquire multiple item information sets corresponding to the multiple query fragments, wherein the item information sets corresponding to the query fragments include item information corresponding to each query in the query fragments; obtaining, based on the multiple item information sets, a preference of the user for simultaneously converting a fifth item corresponding to the fifth item information and a sixth item corresponding to the sixth item information, wherein the multiple item information sets include the fifth item information and the sixth item information; When the preference exceeds a preference threshold, the fifth item information and the sixth item information are combined into a third item combination.

26. The device according to claim 25, characterized in that The acquisition unit is used to: Counting the number of item information sets including the fifth item information and the sixth item information; The preference degree is obtained by calculating a ratio between the number of the item information sets and the total number of the plurality of item information sets.

27. The device according to claim 25 or 26, characterized in that The attribute of the association relationship between the fifth item information and the sixth item information includes identification information of the user.

28. The device according to any one of claims 19 to 27, characterized in that The m queries are queries searched by users in the same user group. The knowledge graph also includes group information of the user group. There is an association relationship between the group information and the m queries.

29. A computing device, characterized in that Each device in the computing device includes at least one processor and at least one memory, wherein the at least one memory stores computer-readable instructions; the at least one processor executes the computer-readable instructions so that the computing device executes the method according to any one of claims 1 to 14.

30. A computer storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 14 is implemented.

31. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 14 is implemented.

Images