Information recommendation method, device, computer system and computer readable storage medium

Abstract

The present disclosure provides an information recommendation method, an information recommendation device, a computer system, a computer-readable storage medium, and a computer program product. The information recommendation method includes: generating a user preference matrix based on user behavior log information; generating a word frequency-inverse text frequency index matrix of an object based on object information stored in a database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in at least one object; integrating the user preference matrix and the word frequency-inverse text frequency index matrix and inserting them into a preset user behavior feature inverted index; performing aggregation calculation on the information in the user behavior feature inverted index to obtain user features for each user; and determining information recommended to each user based on each user feature.

Description

Technical Field

[0001] The present disclosure relates to the field of computer technology, and more specifically, to an information recommendation method, an information recommendation device, a computer system, a computer-readable storage medium, and a computer program product. Background Art

[0002] With the rapid development of Internet technology and information technology, more and more information interactions are realized through computers, networks, etc., and society is gradually entering an era of information overload. In this era, it becomes very difficult for users to obtain the information they are interested in from a large amount of information. In order to solve this problem, information recommendation technology came into being.

[0003] In the process of implementing the concept of the present disclosure, the inventors found that there are at least the following problems in the related art: the existing information recommendation method uses offline calculations through machine learning, has poor real-time performance, and is relatively complex to implement. Summary of the invention

[0004] In view of this, the present disclosure provides an information recommendation method, an information recommendation device, a computer system, a computer-readable storage medium, and a computer program product.

[0005] One aspect of the present disclosure provides an information recommendation method, comprising: generating a user preference matrix based on user behavior log information, wherein the user preference matrix includes the preference feature value of each user of at least one user to each object of at least one object; generating a word frequency-inverse text frequency index matrix of the objects based on the object information stored in a database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item of at least one feature item in the object information corresponding to each object of the at least one object; integrating the user preference matrix and the word frequency-inverse text frequency index matrix and inserting them into a preset user behavior feature inverted index; performing aggregation calculation on the information in the user behavior feature inverted index to obtain user features for each of the users; and determining the information recommended to each of the users based on each of the user features.

[0006] According to an embodiment of the present disclosure, generating a word frequency-inverse text frequency index matrix of an object based on object information stored in a database table includes: calling a term vector interface method to obtain at least one feature item of each of the objects from the object information stored in the database table; calculating each of the feature items based on a word frequency-inverse text frequency index algorithm to obtain a target word frequency-inverse text frequency index for each feature item in the object information corresponding to each of the objects; and forming the word frequency-inverse text frequency index matrix based on the target word frequency-inverse text frequency index.

[0007] According to an embodiment of the present disclosure, the user behavior log information includes a user identifier for determining the user, an object identifier for determining the object, and an operation type of the user on the object, and integrating the user preference matrix and the word frequency-inverse text frequency index matrix and inserting them into a preset user behavior feature inverted index includes: determining a target feature vector corresponding to a target object related to a target user according to an integration result of the user preference matrix and the word frequency-inverse text frequency index matrix, wherein the target feature vector includes at least one target feature item; determining a target operation type of the target user on the target object according to a preference feature value in the user preference matrix; determining a target weighting method for the target feature vector according to the target operation type; performing weighted calculation on each target feature item according to the target weighting method to obtain a weight of each target feature item in the object information corresponding to the target object; and inserting the weight and the corresponding target feature item, target user, and target object into the preset user behavior feature inverted index.

[0008] According to an embodiment of the present disclosure, the target feature vector includes multiple ones, and inserting the weights and the corresponding target feature items, target users and target objects into the preset user behavior feature inverted index includes: merging the multiple target feature vectors according to a preset merging strategy; calculating the target weight of each target feature item in the merged target feature vector; sorting the target feature items in the merged target feature vector according to the target weights to obtain a sorting result; and inserting the first preset value of target feature items in the sorting result and their corresponding target weights, target users and target objects into the preset user behavior feature inverted index.

[0009] According to an embodiment of the present disclosure, the information recommendation method further includes: obtaining a custom feature item filtering interface; and filtering the target feature items in the target feature vector using the custom feature item filtering interface to obtain new target feature items.

[0010] According to an embodiment of the present disclosure, information in the inverted index of user behavior features is aggregated and calculated to obtain user features for each of the users, including: filtering and obtaining target information according to custom filtering conditions; performing a first aggregation on the target information according to a preset nested structure to obtain at least one group of feature items and weights in the target information; and performing a second aggregation on the at least one group of feature items and weights according to the feature item names to sort the feature items according to the weight values ​​corresponding to the feature items, wherein the weight value is obtained by adding the weights of the same feature items in the at least one group of feature items and weights.

[0011] Another aspect of the present disclosure provides an information recommendation device, including: a first generation module, used to generate a user preference matrix based on user behavior log information, wherein the user preference matrix includes the preference feature value of each user of at least one user to each object of at least one object; a second generation module, used to generate a word frequency-inverse text frequency index matrix of the object based on the object information stored in a database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object of the at least one object; an integration module, used to integrate the user preference matrix and the word frequency-inverse text frequency index matrix and insert them into a preset user behavior feature inverted index; an aggregation module, used to perform aggregation calculation on the information in the user behavior feature inverted index to obtain user features for each of the users; and a determination module, used to determine the information recommended to each of the users based on each of the user features.

[0012] Another aspect of the present disclosure provides a computer system, including: one or more processors; and a memory for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the information recommendation method as described above.

[0013] Another aspect of the present disclosure provides a computer-readable storage medium having computer-executable instructions stored thereon, which are used to implement the information recommendation method as described above when executed.

[0014] Another aspect of the present disclosure provides a computer program product, which includes computer executable instructions, and when the instructions are executed, are used to implement the information recommendation method as described above.

[0015] According to an embodiment of the present disclosure, a user preference matrix is ​​generated according to user behavior log information, wherein the user preference matrix includes the preference feature value of each user in at least one user to each object in at least one object; a word frequency-inverse text frequency index matrix of the object is generated according to the object information stored in a database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in at least one object; the user preference matrix and the word frequency-inverse text frequency index matrix are integrated and inserted into a preset user behavior feature inverted index; information in the user behavior feature inverted index is aggregated and calculated to obtain user features for each user; and information technology means for determining recommendations to each user according to each user feature are used. Since the recommendation results are determined according to the log information and the aggregate calculations, the technical problem of poor real-time performance when recommending information based on machine learning is at least partially overcome, thereby achieving the technical effect of realizing a simple and easy information recommendation method. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The above and other objects, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0017] Figure 1 An exemplary system architecture to which the information recommendation method according to an embodiment of the present disclosure can be applied is schematically shown;

[0018] Figure 2 The flowchart of the information recommendation method according to the embodiment of the present disclosure is schematically shown;

[0019] Figure 3 The following schematically shows an architecture diagram of an information recommendation method according to an embodiment of the present disclosure;

[0020] Figure 4 The following schematically shows a calculation flow chart of an information recommendation method according to a specific embodiment of the present disclosure;

[0021] Figure 5 A flowchart schematically shows how to integrate a user preference matrix and a word frequency-inverse text frequency index matrix and insert them into a preset user behavior feature inverted index according to an embodiment of the present disclosure;

[0022] Figure 6 A schematic diagram schematically shows a method of merging different feature vectors according to an embodiment of the present disclosure;

[0023] Figure 7 A block diagram schematically shows an information recommendation device according to an embodiment of the present disclosure; and

[0024] Figure 8A block diagram of a computer system suitable for implementing the method described above according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0025] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0026] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0027] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art, unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification, and should not be interpreted in an idealized or overly rigid manner.

[0028] In the case of using expressions such as "at least one of A, B, and C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.). In the case of using expressions such as "at least one of A, B, or C, etc.", it should generally be interpreted in accordance with the meaning of the expression generally understood by those skilled in the art (for example, "a system having at least one of A, B, or C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0029] The embodiments of the present disclosure provide an information recommendation method, an information recommendation device, a computer system, a computer-readable storage medium, and a computer program product. The method includes generating a user preference matrix based on user behavior log information, wherein the user preference matrix includes the preference feature value of each user in at least one user for each object in at least one object; generating a word frequency-inverse text frequency index matrix of the object based on the object information stored in the database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in at least one object; integrating the user preference matrix and the word frequency-inverse text frequency index matrix and inserting them into a preset user behavior feature inverted index; performing aggregation calculation on the information in the user behavior feature inverted index to obtain user features for each user; and determining the information recommended to each user based on each user feature.

[0030] Figure 1 The exemplary system architecture 100 to which the information recommendation method according to the embodiment of the present disclosure can be applied is schematically shown. It should be noted that: Figure 1 What is shown is merely an example of a system architecture to which the embodiments of the present disclosure can be applied, in order to help those skilled in the art understand the technical content of the present disclosure, but it does not mean that the embodiments of the present disclosure cannot be used in other devices, systems, environments or scenarios.

[0031] like Figure 1 As shown, the system architecture 100 according to this embodiment may include terminal devices 101, 102, 103, a network 104 and a server 105. The network 104 is used to provide a medium for communication links between the terminal devices 101, 102, 103 and the server 105. The network 104 may include various connection types, such as wired and / or wireless communication links, etc.

[0032] Users can use terminal devices 101, 102, 103 to interact with server 105 through network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients and / or social platform software, etc.

[0033] The terminal devices 101 , 102 , and 103 may be various electronic devices having a display screen and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, and desktop computers.

[0034] The server 105 may be a server that provides various services, such as a background management server that provides support for websites browsed by users using the terminal devices 101, 102, and 103. The background management server may analyze and process received data such as user requests, and feed back processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0035] It should be noted that the information recommendation method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the information recommendation device provided in the embodiment of the present disclosure can generally be set in the server 105. The information recommendation method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Accordingly, the information recommendation device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the terminal devices 101, 102, 103 and / or the server 105. Alternatively, the information recommendation method provided in the embodiment of the present disclosure can also be executed by the terminal devices 101, 102, or 103, or can also be executed by other terminal devices different from the terminal devices 101, 102, or 103. Accordingly, the information recommendation device provided in the embodiment of the present disclosure can also be set in the terminal devices 101, 102, or 103, or can be set in other terminal devices different from the terminal devices 101, 102, or 103.

[0036] For example, the user behavior log information or the database table for storing object information may be originally stored in any one of the terminal devices 101, 102, or 103 (for example, the terminal device 101, but not limited thereto), or stored on an external storage device and may be imported into the terminal device 101. Then, the terminal device 101 may locally execute the information recommendation method provided by the embodiment of the present disclosure, or send the user behavior log information or the database table for storing object information to other terminal devices, servers, or server clusters, and the other terminal devices, servers, or server clusters that receive the user behavior log information or the database table for storing object information may execute the information recommendation method provided by the embodiment of the present disclosure.

[0037] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0038] Figure 2 The flowchart of the information recommendation method according to the embodiment of the present disclosure is schematically shown.

[0039] Figure 3The following schematically shows an architecture diagram of an information recommendation method according to an embodiment of the present disclosure.

[0040] Figure 4 The calculation flow chart of the information recommendation method according to a specific embodiment of the present disclosure is schematically shown.

[0041] like Figure 2 As shown, the method includes operations S201 to S205.

[0042] In operation S201 , a user preference matrix is ​​generated according to user behavior log information, wherein the user preference matrix includes a preference feature value of each user of at least one user to each object of at least one object.

[0043] According to the embodiments of the present disclosure, see Figure 3 As shown, various operations of users when browsing websites on the front end may include browsing behavior, clicking behavior, etc. In some embodiments, the operation may be specifically manifested as which articles or products the user browses, collects, likes, and comments on. By collecting logs in the front end, for example, the above-mentioned user behavior log information can be obtained. The collected log information can be reported to the data warehouse for storage on the one hand, and can also be uploaded to the streaming computing platform JRC (Spark) by docking with the real-time data pipeline JDQ on the other hand, for example, the real-time data pipeline JDQ is a high-throughput, high-expansion real-time data bus built by the big data platform based on Kafka. It is widely used in streaming data processing business and is a high-availability message pipeline that connects real-time data collection and the streaming computing platform JRC. It can be used for real-time consumption of the above-mentioned user behavior log information. The above-mentioned objects can, for example, refer to articles, products, comments, and other items with which users have behaviors and have text descriptions that can be fully indexed.

[0044] According to a specific embodiment of the present disclosure, see Figure 4 As shown, the above users may include user1, user2, user3, user4, user5, etc., the above objects may include document 1, document 2, document 3, document 4, etc., and the above preference feature values ​​may be expressed as Figure 4 The "1" or "N / A" form in the user preference matrix is ​​used to represent the relationship between the user and the object, where 1 can represent that the user has browsed an object (called a document for the inverted index and an item for the recommendation system), and 2 can represent that the user has clicked (or purchased) an object ( Figure 4 N / A may represent that the user has neither clicked nor browsed a certain object. The user preference matrix may be based on Figure 4The preference feature values ​​such as "1" or "N / A" in the user preference matrix are obtained.

[0045] In operation S202, a word frequency-inverse text frequency index matrix of the object is generated according to the object information stored in the database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in at least one object.

[0046] According to the embodiments of the present disclosure, see Figure 3 As shown, the data information corresponding to the website browsed by the user (including the above-mentioned object information) is pre-stored in Figure 3 In the database table (TABLE), the information in the database table can be fully synchronized to the object index (ITEM INDEX), and the full-text index fields in the object index (ITEM INDEX) need to set with_positions_offsets_payloads (including configuration information such as position, offset, and payload) so as to determine the field information corresponding to a certain index field (such as a specific object) in the database table according to the configuration information of with_positions_offsets_payloads (for example, it can represent at least one of the above-mentioned feature items). The above-mentioned term frequency-inverse text frequency index matrix can be obtained by, for example, calculating TF-IDF (term frequency-inverse text frequency index) for each feature item in the field information.

[0047] According to a specific embodiment of the present disclosure, see Figure 4 As shown, the feature vector (i.e., the at least one feature item) may include feature 1, feature 2, feature 3, feature 4, feature 5, etc. Taking feature 1 and document 1 as an example, the word frequency-inverse text frequency index may be expressed as the word frequency-inverse text frequency index of feature 1 in document 1, such as Figure 4 The value is, for example, 0.037. The above-mentioned word frequency-inverse text frequency index matrix can be based on Figure 4 The term frequency-inverse text frequency index in the TF-IDF matrix of the object (item) is obtained.

[0048] It should be noted that, in order to optimize the operation process, the object index (ITEM INDEX) can be, for example, an inverted index based on an object identifier. The feature vector can also be finally determined by, for example, sorting the TF-IDF (i.e., the term frequency-inverse text frequency index) value of each feature item and then selecting the feature items in the top n positions in the sorting result.

[0049] In operation S203, the user preference matrix and the word frequency-inverse text frequency index matrix are integrated and inserted into a preset user behavior feature inverted index.

[0050] According to the embodiments of the present disclosure, see Figure 3 As shown, the above preset user behavior feature inverted index can be expressed as Figure 3 The structure shown in the inverted index of user behavior features (USER ACTION INDEX) in the example may include: user_id, a unique identifier for indicating a user identity, such as a user PIN; item_id, a unique identifier for indicating an object, such as a document ID; type, a type of action performed by a user on an object, such as browsing action, liking action, etc.; time, a time when an action occurs; and keywords, an array of feature items after a user performs an action on an object, which may be expressed in the form of TERM1:WEIGHT1, TERM2:WEIGHT2, etc., where TERM1, TERM2, etc. may be used to indicate the name of a feature item, and WEIGHT1, WEIGHT2, etc. may be used to indicate the weight of a feature item. It should be noted that the name and weight of the same feature item need to use a nested object structure when designing the code.

[0051] According to a specific embodiment of the present disclosure, see Figure 4 As shown, after integrating the aforementioned user preference matrix and the aforementioned word frequency-inverse text frequency index matrix, for example, we can get Figure 4 The result corresponding to the user behavior inverted index part in the structure can be finally inserted into the above-mentioned preset user behavior feature inverted index (i.e. Figure 3 in USER ACTIONINDEX).

[0052] In operation S204, aggregation calculation is performed on the information in the inverted index of user behavior features to obtain user features for each user.

[0053] According to a specific embodiment of the present disclosure, see Figure 4 As shown, based on operation S204, for example, Figure 4 In the results shown in the user features section, the user features for a certain user can be determined based on the aggregated calculated values ​​of each row. For example, the user features of user1 can be determined to be feature 1 (0.160), feature 2 (0.120), feature 3 (0.180), and feature 5 (0.189).

[0054] In operation S205 , information recommended to each user is determined according to each user characteristic.

[0055] According to an embodiment of the present disclosure, after obtaining user features for user1 (including the features and the user's affinity for content represented by each feature) based on the above operations S201 to S204, for example, appropriate recommendation information can be provided to user1 based on the user features for user1.

[0056] Through the above-mentioned embodiments of the present disclosure, the calculation of user features can be converted into a search problem, the user behavior log information can be stored through an inverted index, and a feature vector for characterizing user features can be obtained based on aggregate calculation. The information construction method implemented in this way is simpler and can achieve real-time calculation.

[0057] The following is combined with specific embodiments and reference Figure 3 to Figure 6 ,right Figure 2 The method shown is further explained.

[0058] According to an embodiment of the present disclosure, the above-mentioned operation S202 may, for example, include: calling the term vector interface method to obtain at least one feature item of each object from the object information stored in the database table; calculating each feature item based on the term frequency-inverse text frequency index algorithm to obtain the target term frequency-inverse text frequency index of each feature item in the object information corresponding to each object; and forming a term frequency-inverse text frequency index matrix according to the target term frequency-inverse text frequency index.

[0059] According to the embodiments of the present disclosure, see Figure 3 As shown, the above process of operation S202 can be specifically expressed as follows: first, through the _termvectors API (i.e., the above term vector interface) of Elasticsearch (a distributed streaming data engine), for example, the word-based feature items related to the above object (i.e., the above field information) can be obtained, based on which, for example, the feature vector information of the related objects (i.e., at least one feature item of each of the above objects) can be obtained; then, each feature item in the feature vector information is calculated based on the TF-IDF algorithm, for example, the target term frequency-inverse text frequency index of each feature item in the object information corresponding to each object can be obtained, so that the above term frequency-inverse text frequency index matrix can be obtained based on this.

[0060] According to an embodiment of the present disclosure, the above-mentioned user behavior log information includes a user identifier for determining the user, an object identifier for determining the object, and the user's operation type on the object. Integrating the user preference matrix and the word frequency-inverse text frequency index matrix and inserting them into the preset user behavior feature inverted index includes: determining a target feature vector corresponding to a target object related to the target user according to the integration result of the user preference matrix and the word frequency-inverse text frequency index matrix, wherein the target feature vector includes at least one target feature item; determining the target operation type of the target user on the target object according to the preference feature value in the user preference matrix; determining a target weighting method for the target feature vector according to the target operation type; performing weighted calculation on each target feature item according to the target weighting method to obtain the weight of each target feature item in the object information corresponding to the target object; and inserting the weight and the corresponding target feature item, target user, and target object into the preset user behavior feature inverted index.

[0061] According to an embodiment of the present disclosure, the above-mentioned target feature vector includes multiple, and inserting the weights and corresponding target feature items, target users and target objects into the preset user behavior feature inverted index includes: merging the multiple target feature vectors according to a preset merging strategy; calculating the target weight of each target feature item in the merged target feature vector; sorting the target feature items in the merged target feature vector according to the target weight to obtain a sorting result; and inserting the first preset value of target feature items in the sorting result and their corresponding target weights, target users and target objects into the preset user behavior feature inverted index.

[0062] According to an embodiment of the present disclosure, based on the above operation S203, for example, it may also include: obtaining a custom feature item filtering interface; and filtering the target feature items in the target feature vector using the custom feature item filtering interface to obtain new target feature items.

[0063] Figure 5 The flowchart of integrating the user preference matrix and the word frequency-inverse text frequency index matrix and inserting them into the preset user behavior feature inverted index according to an embodiment of the present disclosure is schematically shown.

[0064] like Figure 5 As shown, this process is used to show Figure 3 The data processing process in the streaming computing platform in the embodiment may include operations S501 to S520.

[0065] In operation S501, a user behavior data stream is obtained.

[0066] According to an embodiment of the present disclosure, the operation Figure 3In the architecture diagram shown, for example, it can be shown that the stream computing platform JRC obtains the user behavior log information (ie, user behavior data stream) collected in real time based on the real-time data pipeline JDQ. Based on this operation, for example, the above-mentioned user preference matrix can be obtained.

[0067] In operation S502, feature vector information of the related object is obtained according to item_id.

[0068] According to an embodiment of the present disclosure, the operation Figure 3 In the architecture diagram shown, for example, it can be shown that based on the ES_termvectors API, the function of extracting word vector (i.e., feature vector) constraints can be used, that is, the object information corresponding to each object can be segmented and represented as word-based feature items (i.e., the above-mentioned field information) through the ES_termvectors API, and based on the with_positions_offsets_payloads information set in the object index (ITEM INDEX), the streaming computing platform JRC can obtain the object's field information (i.e., feature vector information) from the database table (TABLE) associated with the object index (ITEM INDEX) through the ES_termvectors API, wherein each field information can include at least one feature item. Based on this operation, for example, the above-mentioned word frequency-inverse text frequency index matrix can be obtained.

[0069] According to an embodiment of the present disclosure, through the user behavior data stream in operation S501, for example, the correspondence between the user (user_id) and the object (item_id) can be first determined. On this basis, by combining the relationship between the field information and the object (item_id) in operation S502, for example, the field information corresponding to the object that has a relationship with the user in the user behavior data stream (that is, the above-mentioned related object) and the corresponding word frequency-inverse text frequency index (that is, the above-mentioned feature vector information) can be finally determined.

[0070] According to an embodiment of the present disclosure, according to the ES_termvectors API, three types of information, namely, Field Statistics, Term Statistics, and Term Information, can be obtained. To calculate the TF*IDF weight of each term, for example, the following data is required:

[0071] a) document count: a field contains multiple documents;

[0072] b) document frequency: contains the temperature number of the current feature item;

[0073] c)Term frequency in the field: the frequency with which a feature term appears in the field.

[0074] It should be noted that noun-like words such as place names and proper nouns often have higher semantic weights and can be processed separately at the application layer.

[0075] In operation S503, Fields in the response result are iterated.

[0076] According to an embodiment of the present disclosure, the objects related to the user in the above user behavior data stream may include multiple objects, and the acquired field information (i.e., Fields) may include multiple objects. For each field information, it is necessary to perform corresponding processing in an iterative manner, such as determining and processing each feature item in the field information, processing the determined feature vector, etc. Each iterative processing, for example, may include operations S504 to S517.

[0077] In operation S504, whether there is a next Field.

[0078] In operation S505, a total word count variable is initialized by accumulating TF calculations.

[0079] In operation S506, the Terms in the Field are enumerated.

[0080] According to an embodiment of the present disclosure, the above operations S505 to S506 are all performed when the judgment result of operation S504 is yes.

[0081] According to an embodiment of the present disclosure, field information (Fields) may be a feature vector (Terms) composed of multiple feature terms (Terms), and for each feature term, corresponding processing needs to be performed by enumeration, such as filtering, etc. Each enumeration process may include operations S507 to S514, for example.

[0082] In operation S507, it is determined whether there is a next Term.

[0083] In operation S508, the Term frequency of the current Term is obtained.

[0084] In operation S509, TF (Term frequency) is accumulated.

[0085] In operation S510, a Term name is acquired.

[0086] In operation S511, a Term filtering interface is customized.

[0087] In operation S512, whether to filter.

[0088] In operation S513, the current Term is added to the feature vector set of the Field.

[0089] In operation S514, this iteration is completed.

[0090] According to an embodiment of the present disclosure, the above operations S508 to S514 are all performed when the judgment result of operation S507 is yes, and the above operations S513 to S514 are performed when the judgment result of operation S512 is yes, and when the judgment result of operation S512 is no, operation S514 is directly performed.

[0091] According to an embodiment of the present disclosure, the filtering condition of the filtering interface in the above operation S511 is set, for example, according to the Term name. It should be noted that the filtering condition can also be set, for example, according to the TF value, which is not limited here. In this embodiment, when processing each feature item in the field information, feature items that are not desired to be built into the user behavior features can be filtered out by setting a custom Term filtering interface.

[0092] It should be noted that whether a feature item needs to be built into the user behavior feature can also be determined by, for example, sorting all feature items by TF value. For example, feature items in the top n positions can be selected as feature items that need to be built into the user behavior feature.

[0093] According to an embodiment of the present disclosure, after all feature items in the field information are enumerated, that is, when the judgment result of the above operation S507 is no, for example, the feature vector finally determined in the above field information can be further processed, and the processing process can, for example, include operations S515 to S517.

[0094] In operation S515, the statistical information document count of the Field is obtained.

[0095] In operation S516, the feature vector set is iterated.

[0096] In operation S517, a weighted TD-IDF is calculated for each feature item.

[0097] According to the embodiments of the present disclosure, for example, different weighting methods can be set for feature items corresponding to field information of objects corresponding to different operation types according to different operation types of users on the objects. For example, the weighting methods of browsing behavior and liking behavior can be set to be different, so that the TF-IDF weight of each feature item can be determined in a targeted manner, thereby obtaining more accurate user behavior characteristics that are more in line with user affinity.

[0098] According to an embodiment of the present disclosure, after iterations for all field information are completed, that is, when the judgment result of the above operation S504 is no, for example, multiple field information can be merged to obtain user behavior characteristics, and the process can include operations S518 to S521.

[0099] In operation S518, feature vector sets in multiple Fields are merged.

[0100] According to the embodiments of the present disclosure, since the object information of different objects is different, after the above operations S504 to S517, the feature vectors finally determined in the Field corresponding to different objects may be different. Since there may be multiple objects that have a relationship with the user, it is necessary to accurately determine the user behavior characteristics based on the weight of each feature item in the corresponding object information. This operation is used, for example, to merge the feature items in the feature vectors finally determined corresponding to different Fields to obtain the weight of each feature item finally determined in the object information corresponding to each object.

[0101] In operation S519, a merging strategy of the same Term is executed.

[0102] According to an embodiment of the present disclosure, when multiple Fields are passed into the termvectors API, the aforementioned operation process will independently calculate a feature vector set for each Field. Therefore, it is necessary to merge the independent feature vector sets. When the term names are the same, the weight merging strategy may, for example, include: taking the average, taking the maximum value, adding or other custom implementation methods.

[0103] In operation S520, the top M are obtained according to the combined weight ranking.

[0104] Figure 6 The figure schematically shows a schematic diagram of merging different feature vectors according to an embodiment of the present disclosure.

[0105] like Figure 6 As shown, the feature items in FieldA: the weights include t1: 0.8, t2: 0.6:, t3: 0.5, the feature items in FieldB: the weights include t2: 0.6, t3: 0.5, t4: 0.7, based on the merging strategy of taking the average value, after operations S518~S519, for example, t1: 0.8, t2: 0.4, t3: 0.3, t4: 0.7 can be obtained, and after operation S520, for example, t1: 0.8, t4: 0.7, t2: 0.4, t3: 0.3 can be obtained.

[0106] According to an embodiment of the present disclosure, by sorting the feature items by weight and selecting the first n feature items, the obtained t1: 0.8, t4: 0.7, t2: 0.4, t3: 0.3, for example, can form user behavior characteristics.

[0107] In operation S521, the user behavior characteristics are saved.

[0108] According to an embodiment of the present disclosure, this operation is used, for example, to further write the user behavior characteristics formed above into Figure 3 The JSON example of writing data into the inverted index of user behavior features (USER ACTION INDEX) is as follows:

[0109]

[0110] It can provide a basis for subsequent aggregation queries.

[0111] According to an embodiment of the present disclosure, the above-mentioned operation S204 may, for example, include: filtering to obtain target information according to custom filtering conditions; performing a first aggregation on the target information according to a preset nested structure to obtain at least one group of feature items and weights in the target information; and performing a second aggregation on at least one group of feature items and weights according to the feature item names to sort the feature items according to the weight values ​​corresponding to the feature items, wherein the weight value is obtained by adding the weights of the same feature items in at least one group of feature items and weights.

[0112] According to an embodiment of the present disclosure, based on the above-mentioned user action feature inverted index (USER ACTION INDEX), for example, real-time aggregation calculation can be performed directly through the ES query DSL algorithm to obtain user action feature items. The query logic may include, for example:

[0113] a) The query part (query) completes the filtering, filtering the data of the user (user_id) under the specified behavior type (type) within a certain time window (range query created). The code logic related to this part can customize the filtering conditions according to the application scenario, and even add additional business attributes to the index for conditional filtering.

[0114] b) Aggregation part (aggs), aggregation is performed based on the results of the query part, and the calculation of user behavior features is completed through aggregation. The aggregation logic can be expressed as, for example, completing the aggregation of the nested object structure in the first layer of aggregation (aggs.agg_keywords, i.e. the first aggregation mentioned above), completing the aggregation by the feature item name in the second layer of aggregation (aggs.agg_word, i.e. the second aggregation mentioned above), and sorting in descending order according to the results (weights) of the third layer of aggregation, where, in the third layer of aggregation (aggs.agg_score), for example, the weights of the same feature words are added (SUM). The aggregation logic can be expressed as, for example:

[0115]

[0116]

[0117] Based on the above aggregation logic, for example, a set of feature items most relevant to the current query can be obtained.

[0118] According to the embodiments of the present disclosure, see Figure 3 As shown, real-time calculation of the inverted index of user action features (USER ACTION INDEX) is provided, for example, user features can be queried in real time through aggregate query. The query result can be expressed as a set of feature items and their corresponding weight values, which can provide support for subsequent business.

[0119] Through the above-mentioned embodiments of the present disclosure, a universal inverted index storage structure is provided, based on which user behavior characteristics can be calculated in real time using a fixed aggregation query method, ultimately providing support for recommendation systems and search systems.

[0120] Figure 7 The block diagram of the information recommendation device according to an embodiment of the present disclosure is schematically shown.

[0121] like Figure 7 As shown, the information recommendation device 700 includes a first generating module 710 , a second generating module 720 , an integrating module 730 , an aggregating module 740 and a determining module 750 .

[0122] The first generating module 710 is used to generate a user preference matrix according to the user behavior log information, wherein the user preference matrix includes a preference feature value of each user of at least one user to each object of at least one object.

[0123] The second generation module 720 is used to generate a word frequency-inverse text frequency index matrix of the object based on the object information stored in the database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in the at least one object.

[0124] The integration module 730 is used to integrate the user preference matrix and the word frequency-inverse text frequency index matrix and insert them into a preset user behavior feature inverted index.

[0125] The aggregation module 740 is used to perform aggregation calculation on the information in the user behavior feature inverted index to obtain user features for each of the users.

[0126] The determination module 750 is used to determine the information recommended to each of the users according to the characteristics of each of the users.

[0127] According to an embodiment of the present disclosure, a user preference matrix is ​​generated according to user behavior log information, wherein the user preference matrix includes the preference feature value of each user in at least one user to each object in at least one object; a word frequency-inverse text frequency index matrix of the object is generated according to the object information stored in a database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in at least one object; the user preference matrix and the word frequency-inverse text frequency index matrix are integrated and inserted into a preset user behavior feature inverted index; information in the user behavior feature inverted index is aggregated and calculated to obtain user features for each user; and information technology means for determining recommendations to each user according to each user feature are used. Since the recommendation results are determined according to the log information and the aggregate calculations, the technical problem of poor real-time performance when recommending information based on machine learning is at least partially overcome, thereby achieving the technical effect of realizing a simple and easy information recommendation method.

[0128] According to an embodiment of the present disclosure, the second generating module includes a first acquiring unit, a first calculating unit and a forming unit.

[0129] The first acquisition unit is used to call the term vector interface method to acquire at least one feature item of each object from the object information stored in the database table.

[0130] The first calculation unit is used to calculate each feature item based on a word frequency-inverse text frequency index algorithm to obtain a target word frequency-inverse text frequency index of each feature item in the object information corresponding to each object.

[0131] A forming unit is used to form a word frequency-inverse text frequency index matrix according to the target word frequency-inverse text frequency index.

[0132] According to an embodiment of the present disclosure, the above-mentioned user behavior log information includes a user identifier for determining the user, an object identifier for determining the object, and the user's operation type on the object, and the above-mentioned integration module includes a first determination unit, a second determination unit, a third determination unit, a second calculation unit and an insertion unit.

[0133] The first determination unit is used to determine a target feature vector corresponding to a target object related to a target user according to an integration result of a user preference matrix and a word frequency-inverse text frequency index matrix, wherein the target feature vector includes at least one target feature item.

[0134] The second determining unit is used to determine the target operation type of the target user on the target object according to the preference feature value in the user preference matrix.

[0135] The third determining unit is used to determine a target weighting method for the target feature vector according to the target operation type.

[0136] The second calculation unit is used to perform weighted calculation on each target feature item according to the target weighting method to obtain the weight of each target feature item in the object information corresponding to the target object.

[0137] The insertion unit is used to insert the weight and the corresponding target feature item, target user and target object into the preset user behavior feature inverted index.

[0138] According to an embodiment of the present disclosure, the target feature vector includes multiple ones, and the insertion unit includes a merging subunit, a calculating subunit, a sorting subunit and an inserting subunit.

[0139] The merging subunit is used to merge multiple target feature vectors according to a preset merging strategy.

[0140] The calculation subunit is used to calculate the target weight of each target feature item in the merged target feature vector.

[0141] The sorting subunit is used to sort the target feature items in the merged target feature vector according to the target weight to obtain a sorting result.

[0142] The inserting subunit is used to insert the first preset value target feature items in the sorting result and their corresponding target weights, target users and target objects into the preset user behavior feature inverted index.

[0143] According to an embodiment of the present disclosure, the above-mentioned information recommendation device further includes a second acquisition unit and a filtering unit.

[0144] The second acquisition unit is used to acquire a custom feature item filtering interface.

[0145] The filtering unit is used to filter the target feature items in the target feature vector using the custom feature item filtering interface to obtain new target feature items.

[0146] According to an embodiment of the present disclosure, the aggregation module includes a screening unit, a first aggregation unit and a second aggregation unit.

[0147] The filtering unit is used to filter and obtain target information according to the custom filtering conditions.

[0148] The first aggregation unit is used to perform a first aggregation on the target information according to a preset nested structure to obtain at least one set of feature items and weights in the target information.

[0149] The second aggregation unit is used to perform a second aggregation on at least one group of feature items and weights according to the feature item names, so as to sort the feature items according to the weight values ​​corresponding to the feature items, wherein the weight value is obtained by adding the weights of the same feature items in at least one group of feature items and weights.

[0150] According to the embodiments of the present invention, any one or more of the modules, units, and sub-units, or at least part of the functions of any one of them can be implemented in one module. According to the embodiments of the present invention, any one or more of the modules, units, and sub-units can be split into multiple modules for implementation. According to the embodiments of the present invention, any one or more of the modules, units, and sub-units can be at least partially implemented as hardware circuits, such as field programmable gate arrays (FPGAs), programmable logic arrays (PLAs), systems on chips, systems on substrates, systems on packages, application specific integrated circuits (ASICs), or can be implemented by hardware or firmware in any other reasonable way of integrating or packaging the circuit, or in any one of the three implementation methods of software, hardware, and firmware, or in any appropriate combination of any of them. Alternatively, according to the embodiments of the present invention, one or more of the modules, units, and sub-units can be at least partially implemented as computer program modules, and when the computer program modules are run, the corresponding functions can be performed.

[0151] For example, any of the first generation module 710, the second generation module 720, the integration module 730, the aggregation module 740, and the determination module 750 can be combined in one module / unit / sub-unit for implementation, or any of the modules / units / sub-units can be split into multiple modules / units / sub-units. Alternatively, at least part of the functions of one or more of these modules / units / sub-units can be combined with at least part of the functions of other modules / units / sub-units and implemented in one module / unit / sub-unit. According to an embodiment of the present disclosure, at least one of the first generation module 710, the second generation module 720, the integration module 730, the aggregation module 740, and the determination module 750 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or by any one of the three implementation methods of software, hardware, and firmware, or by a suitable combination of any of them. Alternatively, at least one of the first generating module 710 , the second generating module 720 , the integrating module 730 , the aggregating module 740 and the determining module 750 may be at least partially implemented as a computer program module, which may perform corresponding functions when executed.

[0152] It should be noted that the information recommendation device part in the embodiment of the present disclosure corresponds to the information recommendation method part in the embodiment of the present disclosure. The description of the information recommendation device part specifically refers to the information recommendation method part, which will not be repeated here.

[0153] Figure 8 A block diagram of a computer system suitable for implementing the method described above according to an embodiment of the present disclosure is schematically shown. Figure 8 The computer system shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0154] like Figure 8As shown, the computer system 800 according to the embodiment of the present disclosure includes a processor 801, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 802 or the program loaded from the storage part 808 to the random access memory (RAM) 803. The processor 801 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 801 may also include an onboard memory for caching purposes. The processor 801 may include a single processing unit or multiple processing units for performing different actions of the method flow according to the embodiment of the present disclosure.

[0155] In RAM 803, various programs and data required for the operation of system 800 are stored. Processor 801, ROM 802 and RAM 803 are connected to each other through bus 804. Processor 801 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 802 and / or RAM 803. It should be noted that the program can also be stored in one or more memories other than ROM 802 and RAM 803. Processor 801 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0156] According to an embodiment of the present disclosure, the system 800 may further include an input / output (I / O) interface 805, which is also connected to the bus 804. The system 800 may further include one or more of the following components connected to the I / O interface 805: an input section 806 including a keyboard, a mouse, etc.; an output section 807 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 808 including a hard disk, etc.; and a communication section 809 including a network interface card such as a LAN card, a modem, etc. The communication section 809 performs communication processing via a network such as the Internet. A drive 810 is also connected to the I / O interface 805 as needed. A removable medium 811, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 810 as needed, so that a computer program read therefrom is installed into the storage section 808 as needed.

[0157] According to an embodiment of the present disclosure, the method flow according to an embodiment of the present disclosure can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a computer-readable storage medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 809, and / or installed from the removable medium 811. When the computer program is executed by the processor 801, the above-mentioned functions defined in the system of the embodiment of the present disclosure are executed. According to an embodiment of the present disclosure, the system, equipment, device, module, unit, etc. described above can be implemented by a computer program module.

[0158] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0159] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium. For example, I may include, but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus, or a device.

[0160] For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 802 and / or the RAM 803 described above and / or one or more memories other than the ROM 802 and the RAM 803 .

[0161] An embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains a program code for executing the method provided by the embodiment of the present disclosure. When the computer program product runs on an electronic device, the program code is used to enable the electronic device to implement the information recommendation method provided by the embodiment of the present disclosure.

[0162] When the computer program is executed by the processor 801, the above functions defined in the system / device of the embodiment of the present disclosure are executed. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0163] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 809, and / or installed from a removable medium 811. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0164] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0165] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flowchart or block diagram may represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of the code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box may also occur in an order different from that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flowchart, and the combination of boxes in the block diagram or flowchart, can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions. It can be understood by those skilled in the art that the features recorded in the various embodiments and / or claims of the present disclosure can be combined and / or combined in a variety of ways, even if such a combination or combination is not explicitly recorded in the present disclosure. In particular, without departing from the spirit and teaching of the present disclosure, the features described in the various embodiments and / or claims of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0166] The embodiments of the present disclosure are described above. However, these embodiments are only for the purpose of illustration and are not intended to limit the scope of the present disclosure. Although the embodiments are described above separately, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. The scope of the present disclosure is defined by the attached claims and their equivalents. Without departing from the scope of the present disclosure, those skilled in the art may make various substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. An information recommendation method, comprising: Generate a user preference matrix according to the user behavior log information, wherein the user preference matrix includes a preference feature value of each user of at least one user to each object of at least one object; Generate a word frequency-inverse text frequency index matrix of the object according to the object information stored in the database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in the at least one object; Integrate the user preference matrix and the word frequency-inverse text frequency index matrix and insert them into a preset user behavior feature inverted index; Aggregate the information in the inverted index of the user behavior features to obtain user features for each of the users, including: According to the custom filtering condition, the target information is filtered from the information of the inverted index of the user behavior characteristics; Performing a first aggregation on the target information according to a preset nested structure to obtain at least one set of feature items and weights in the target information; and Performing a second aggregation on the at least one set of feature items and weights according to the feature item names, so as to sort the feature items according to the weight values ​​corresponding to the feature items, wherein the weight value is obtained by adding the weights of the same feature items in the at least one set of feature items and weights; and Determine information recommended to each of the users according to each of the user characteristics.

2. The method according to claim 1, wherein: Generating the word frequency-inverse text frequency index matrix of the object according to the object information stored in the database table includes: Calling the term vector interface method to obtain at least one feature item of each of the objects from the object information stored in the database table; Calculating each of the feature items based on a word frequency-inverse text frequency index algorithm to obtain a target word frequency-inverse text frequency index of each feature item in the object information corresponding to each of the objects; and The word frequency-inverse text frequency index matrix is ​​formed according to the target word frequency-inverse text frequency index.

3. The method according to claim 1, wherein: The user behavior log information includes a user identifier for determining the user, an object identifier for determining the object, and the type of operation performed by the user on the object. The user preference matrix and the word frequency-inverse text frequency index matrix are integrated and inserted into a preset user behavior feature inverted index, including: Determining a target feature vector corresponding to a target object related to a target user according to an integration result of the user preference matrix and the word frequency-inverse text frequency index matrix, wherein the target feature vector includes at least one target feature item; Determining a target operation type of the target user for the target object according to the preference feature values ​​in the user preference matrix; Determining a target weighting method for the target feature vector according to the target operation type; Performing weighted calculation on each of the target feature items according to the target weighting method to obtain the weight of each of the target feature items in the object information corresponding to the target object; and The weights and corresponding target feature items, target users and target objects are inserted into the preset user behavior feature inverted index.

4. The method according to claim 3, wherein: The target feature vector includes a plurality of target feature items, and inserting the weights and corresponding target feature items, target users, and target objects into the preset user behavior feature inverted index includes: Merging the plurality of target feature vectors according to a preset merging strategy; Calculating the target weight of each target feature item in the combined target feature vector; Sorting the target feature items in the merged target feature vector according to the target weight to obtain a sorting result; and The first preset value target feature items in the sorting result and their corresponding target weights, target users and target objects are inserted into the preset user behavior feature inverted index.

5. The method according to claim 3, further comprising: Get the custom feature item filtering interface; as well as The target feature items in the target feature vector are filtered using the custom feature item filtering interface to obtain new target feature items.

6. An information recommendation device, comprising: A first generating module, configured to generate a user preference matrix according to user behavior log information, wherein the user preference matrix includes a preference feature value of each user of at least one user to each object of at least one object; A second generating module is used to generate a word frequency-inverse text frequency index matrix of the object according to the object information stored in the database table, wherein the word frequency-inverse text frequency index matrix includes the word frequency-inverse text frequency index of each feature item in at least one feature item in the object information corresponding to each object in the at least one object; An integration module, used to integrate the user preference matrix and the word frequency-inverse text frequency index matrix and insert them into a preset user behavior feature inverted index; an aggregation module, configured to perform aggregation calculation on the information in the inverted index of the user behavior features to obtain user features for each of the users; and A determination module, used for determining information recommended to each of the users according to the characteristics of each of the users; Wherein, the aggregation module includes: A screening unit, used to screen the target information from the information of the user behavior feature inverted index according to a custom screening condition; A first aggregation unit, configured to perform a first aggregation on the target information according to a preset nested structure to obtain at least one set of feature items and weights in the target information; and A second aggregation unit is used to perform a second aggregation on the at least one group of feature items and weights according to the feature item names, so as to sort the feature items according to the weight values ​​corresponding to the feature items, wherein the weight value is obtained by adding the weights of the same feature items in the at least one group of feature items and weights.

7. A computer system comprising: one or more processors; a memory for storing one or more programs, When the one or more programs are executed by the one or more processors, the one or more processors implement the method according to any one of claims 1 to 5.

8. A computer-readable storage medium having executable instructions stored thereon, which, when executed by a processor, enables the processor to implement the method according to any one of claims 1 to 5.

9. A computer program product comprising computer executable instructions, wherein the instructions are used to implement the method according to any one of claims 1 to 5 when executed.

Images