Information recommendation method, device, equipment and computer-readable storage medium

Abstract

The present invention provides an information recommendation method, device, equipment and computer-readable storage medium, wherein the information recommendation method includes: semantically fusing the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information; semantically comparing each preference information with each feature information of the candidate occasion to obtain the similarity value corresponding to each feature information; determining the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; and recommending the occasion information according to the matching degree value. This scheme can comprehensively consider the information and preferences of multiple users, so as to accurately and comprehensively give occasion recommendations; specifically, it can effectively provide satisfactory restaurant recommendations based on the dining characteristics of multiple users, reduce the complexity of users selecting from a large number of restaurants, improve the efficiency of dining selection, and improve the satisfaction of diners with the selected restaurant or food.

Description

Technical Field

[0001] The present invention relates to the field of Internet technology, and in particular to an information recommendation method, device, equipment and computer-readable storage medium. Background Art

[0002] In recent years, with the development of information technology and the Internet, people have gradually entered an era of information overload from an era of information scarcity. Therefore, recommendation systems have emerged. The application of recommendation systems can help consumers quickly locate the products they need in the complex product information, thereby improving user experience, increasing customer stickiness, and preventing customer churn. Among them, restaurant recommendations are an important application area. Now many food delivery apps are committed to providing users with better restaurant recommendation services.

[0003] There are two main types of restaurant recommendation systems for multiple users: one is a multi-user restaurant recommendation system based on the majority rule, and the other is a multi-user restaurant recommendation system based on matrix decomposition;

[0004] The multi-user restaurant recommendation system for the minority obeys the majority usually makes recommendations based on the principle of minority obeys the majority through voting, asking or relying on the user's understanding of other users to make empirical judgments, so as to try to meet the needs of the dining party; although this method can consider the needs of more people to a certain extent, this approach is too simple and unscientific; whether it is voting or asking, it lacks user information and domain knowledge, and cannot comprehensively analyze the preferences of multiple users, and the recommendation accuracy is low. At the same time, some users may be unwilling to participate in the voting process due to face or other reasons, which makes this solution have some limitations.

[0005] A multi-user restaurant recommendation system based on matrix decomposition usually constructs a joint matrix between users and restaurants, and then obtains a user feature matrix and a restaurant feature matrix through matrix decomposition, so as to make recommendations based on these two matrices. This method can meet the dining needs of multiple users to a certain extent, but it lacks scalability and reasoning capabilities, and cannot mine the implicit relationship between users themselves and between users and restaurants, resulting in low recommendation accuracy and recall rates. For example, a user's personal information only contains information that he likes Sichuan cuisine, but no information related to his liking for spicy food. Matrix decomposition cannot infer that the user's favorite flavor is spicy.

[0006] It can be seen from the above that the restaurant recommendation system for multiple users in the prior art has the problem of low recommendation accuracy. Summary of the invention

[0007] The object of the present invention is to provide an information recommendation method, device, equipment and computer-readable storage medium to solve the problem of low recommendation accuracy of restaurant recommendation systems for multiple users in the prior art.

[0008] In order to solve the above technical problems, an embodiment of the present invention provides an information recommendation method, including:

[0009] Semantically fuse the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information;

[0010] Semantically comparing each of the preference information with each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information;

[0011] Determining a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0012] According to the matching degree value, occasion information is recommended.

[0013] Optionally, before semantically fusing the personal preference information of each user in the target user group, the following is further included:

[0014] Semantic completion is performed on the personal preference information and the candidate occasion information.

[0015] Optionally, the semantically completing the personal preference information and the candidate occasion information includes:

[0016] Obtaining common information of the occasion according to the personal preference information of the preset user and the characteristic information of the preset occasion;

[0017] According to the common occasion information, semantic completion is performed on the personal preference information and the candidate occasion information.

[0018] Optionally, the semantically completing the personal preference information and the candidate occasion information includes:

[0019] According to the upper and lower relationship information, semantically complete the personal preference information and the candidate occasion information; or

[0020] According to preset inference rules, semantic completion is performed on the personal preference information and candidate occasion information.

[0021] Optionally, the semantic fusion of the personal preference information of each user in the target user group to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information includes:

[0022] Obtain the semantic similarity value between each user's personal preference information;

[0023] According to the semantic similarity value, the personal preference information of each user is integrated to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

[0024] Optionally, determining the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value includes:

[0025] Obtaining a weight value corresponding to the preference information corresponding to each first similarity value;

[0026] Determining the matching degree value between the candidate occasion and the target user group according to the obtained weight value;

[0027] The first similarity value is the similarity value greater than a first threshold.

[0028] Optionally, the number of the candidate occasions is at least two, and the recommending of occasion information according to the matching degree value includes:

[0029] Determining a recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value;

[0030] Recommending occasion information according to the recommendation order;

[0031] The distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

[0032] Optionally, determining the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value includes:

[0033] Formula 1 is used to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value;

[0034] Wherein, the formula 1 is:

[0035]

[0036] P represents the recommended order;

[0037] α represents the weight value corresponding to the matching degree value;

[0038] represents the matching degree value of the jth candidate occasion, m ijrepresents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold;

[0039] n represents the total number of the first similarity values;

[0040] β represents the weight value corresponding to the popularity value corresponding to the candidate occasion;

[0041] x j represents the popularity value corresponding to the j-th candidate occasion;

[0042] λ represents the weight value corresponding to the distance value corresponding to the candidate occasion;

[0043] y j represents the distance value corresponding to the j-th candidate occasion;

[0044] l represents the total number of candidate occasions.

[0045] The embodiment of the present invention further provides an information recommendation device, comprising:

[0046] The first processing module is used to semantically fuse the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information;

[0047] A first comparison module is used to perform semantic comparison between each of the preference information and each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information;

[0048] A first determination module, used to determine a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0049] The first recommendation module is used to recommend occasion information according to the matching degree value.

[0050] Optionally, also include:

[0051] The second processing module is used to semantically complete the personal preference information and the candidate occasion information before semantically fusing the personal preference information of each user in the target user group.

[0052] Optionally, the second processing module includes:

[0053] The first processing submodule is used to obtain the common information of the occasion according to the personal preference information of the preset user and the characteristic information of the preset occasion;

[0054] The second processing submodule is used to semantically complete the personal preference information and the candidate occasion information according to the occasion common information.

[0055] Optionally, the semantically completing the personal preference information and the candidate occasion information includes:

[0056] According to the upper and lower relationship information, semantically complete the personal preference information and the candidate occasion information; or

[0057] According to preset inference rules, semantic completion is performed on the personal preference information and candidate occasion information.

[0058] Optionally, the first processing module includes:

[0059] The first acquisition submodule is used to obtain the semantic similarity value between the personal preference information of each user;

[0060] The third processing submodule is used to integrate the personal preference information of each user according to the semantic similarity value to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

[0061] Optionally, the first determining module includes:

[0062] A second acquisition submodule is used to obtain a weight value corresponding to the preference information corresponding to each first similarity value;

[0063] A first determination submodule, used to determine a matching degree value between the candidate occasion and the target user group according to the acquired weight value;

[0064] The first similarity value is the similarity value greater than a first threshold.

[0065] Optionally, the number of the candidate occasions is at least two, and the first recommendation module includes:

[0066] A second determination submodule, configured to determine a recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value;

[0067] A first recommendation submodule, used for recommending occasion information according to the recommendation order;

[0068] The distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

[0069] Optionally, the second determining submodule is specifically used to:

[0070] Formula 1 is used to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value;

[0071] Wherein, the formula 1 is:

[0072]

[0073] P represents the recommended order;

[0074] α represents the weight value corresponding to the matching degree value;

[0075] represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold;

[0076] n represents the total number of the first similarity values;

[0077] β represents the weight value corresponding to the popularity value corresponding to the candidate occasion;

[0078] x j represents the popularity value corresponding to the j-th candidate occasion;

[0079] λ represents the weight value corresponding to the distance value corresponding to the candidate occasion;

[0080] y j represents the distance value corresponding to the j-th candidate occasion;

[0081] l represents the total number of candidate occasions.

[0082] An embodiment of the present invention further provides an information recommendation device, comprising: a processor;

[0083] The processor is used to semantically fuse the personal preference information of each user in the target user group to obtain each piece of preference information corresponding to the target user group and a weight value corresponding to each piece of preference information;

[0084] Semantically comparing each of the preference information with each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information;

[0085] Determining a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0086] According to the matching degree value, occasion information is recommended.

[0087] Optionally, the processor is further configured to:

[0088] Before semantically fusing the personal preference information of each user in the target user group, semantic completion is performed on the personal preference information and the candidate occasion information.

[0089] Optionally, the processor is specifically configured to:

[0090] Obtaining common information of the occasion according to the personal preference information of the preset user and the characteristic information of the preset occasion;

[0091] According to the common occasion information, semantic completion is performed on the personal preference information and the candidate occasion information.

[0092] Optionally, the processor is specifically configured to:

[0093] According to the upper and lower relationship information, semantically complete the personal preference information and the candidate occasion information; or

[0094] According to preset inference rules, semantic completion is performed on the personal preference information and candidate occasion information.

[0095] Optionally, the processor is specifically configured to:

[0096] Obtain the semantic similarity value between each user's personal preference information;

[0097] According to the semantic similarity value, the personal preference information of each user is integrated to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

[0098] Optionally, the processor is specifically configured to:

[0099] Obtaining a weight value corresponding to the preference information corresponding to each first similarity value;

[0100] Determining the matching degree value between the candidate occasion and the target user group according to the obtained weight value;

[0101] The first similarity value is the similarity value greater than a first threshold.

[0102] Optionally, the processor is specifically configured to:

[0103] Determining a recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value;

[0104] Recommending occasion information according to the recommendation order;

[0105] The distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

[0106] Optionally, the processor is specifically configured to:

[0107] Formula 1 is used to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value;

[0108] Wherein, the formula 1 is:

[0109]

[0110] P represents the recommended order;

[0111] α represents the weight value corresponding to the matching degree value;

[0112] represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold;

[0113] n represents the total number of the first similarity values;

[0114] β represents the weight value corresponding to the popularity value corresponding to the candidate occasion;

[0115] x j represents the popularity value corresponding to the j-th candidate occasion;

[0116] λ represents the weight value corresponding to the distance value corresponding to the candidate occasion;

[0117] y j represents the distance value corresponding to the j-th candidate occasion;

[0118] l represents the total number of candidate occasions.

[0119] An embodiment of the present invention further provides an information recommendation device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the program, the above-mentioned information recommendation method is implemented.

[0120] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the above-mentioned information recommendation method are implemented.

[0121] The beneficial effects of the above technical solution of the present invention are as follows:

[0122] In the above scheme, the information recommendation method semantically fuses the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information; semantically compares each preference information with each feature information of the candidate occasion to obtain the similarity value corresponding to each feature information; determines the matching degree value between the candidate occasion and the target user group based on the similarity value and the weight value; recommends occasion information based on the matching degree value; can comprehensively consider the information and preferences of multiple users, so as to accurately and comprehensively give occasion recommendations; specifically, can effectively provide satisfactory restaurant recommendations based on the dining characteristics of multiple users, reduce the complexity of users' selection from a large number of restaurants, improve the efficiency of dining selection, and also can improve the satisfaction of diners with the selected restaurant or food; it well solves the problem of low recommendation accuracy of restaurant recommendation systems for multiple users in the prior art. BRIEF DESCRIPTION OF THE DRAWINGS

[0123] Figure 1 A schematic diagram of a flow chart of an information recommendation method according to an embodiment of the present invention;

[0124] Figure 2 A schematic diagram of a specific implementation flow of the information recommendation method according to an embodiment of the present invention;

[0125] Figure 3 This is a schematic diagram of a user-restaurant shared knowledge base according to an embodiment of the present invention;

[0126] Figure 4 This is a schematic diagram of a portion of the personal information ontology of user u1 according to an embodiment of the present invention;

[0127] Figure 5 This is a schematic diagram of a partial restaurant information body of restaurant R1 according to an embodiment of the present invention;

[0128] Figure 6 This is a schematic diagram of semantic completion of the personal information of user u1 according to an embodiment of the present invention;

[0129] Figure 7 This is a schematic diagram of a user personal information ontology after semantic completion according to an embodiment of the present invention;

[0130] Figure 8 This is a schematic diagram of the structure of an information recommendation device according to an embodiment of the present invention;

[0131] Fig. 9 It is a schematic diagram of the structure of an information recommendation device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0132] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0133] The present invention aims at the problem of low recommendation accuracy of restaurant recommendation systems for multiple users in the prior art, and provides an information recommendation method, such as Figure 1 As shown, including:

[0134] Step 11: semantically fuse the personal preference information of each user in the target user group to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information;

[0135] Step 12: semantically compare each of the preference information with each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information;

[0136] Step 13: Determine the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0137] Step 14: Recommending occasion information based on the matching degree value.

[0138] The information recommendation method provided by the embodiment of the present invention obtains each preference information corresponding to the target user group and the weight value corresponding to each preference information by semantically fusing the personal preference information of each user in the target user group; semantically compares each preference information with each feature information of the candidate occasion to obtain the similarity value corresponding to each feature information; determines the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; recommends occasion information according to the matching degree value; can comprehensively consider the information and preferences of multiple users, so as to accurately and comprehensively give occasion recommendations; specifically, can effectively provide satisfactory restaurant recommendations according to the dining characteristics of multiple users, reduce the complexity of users' selection from a large number of restaurants, improve the efficiency of dining selection, and also can improve the satisfaction of diners with the selected restaurant or food; and well solves the problem of low recommendation accuracy of restaurant recommendation systems for multiple users in the prior art.

[0139] Furthermore, considering that the user or occasion information is missing or incomplete, in an embodiment of the present invention, before semantically fusing the personal preference information of each user in the target user group, it also includes: semantically completing the personal preference information and candidate occasion information.

[0140] The semantically completing the personal preference information and candidate occasion information includes: obtaining common occasion information based on the personal preference information of a preset user and characteristic information of a preset occasion; and semantically completing the personal preference information and candidate occasion information based on the common occasion information.

[0141] Specifically, the semantically completing the personal preference information and the candidate occasion information includes: semantically completing the personal preference information and the candidate occasion information according to the hierarchical relationship information; or semantically completing the personal preference information and the candidate occasion information according to a preset inference rule.

[0142] The inference rules may be obtained through machine learning or may be predefined, which is not limited here.

[0143] In an embodiment of the present invention, the personal preference information of each user in the target user group is semantically fused to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information, including: obtaining a semantic similarity value between the personal preference information of each user; and integrating the personal preference information of each user according to the semantic similarity value to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

[0144] Specifically, determining the matching degree value between the candidate occasions and the target user group based on the similarity value and the weight value includes: obtaining the weight value corresponding to the preference information corresponding to each first similarity value; determining the matching degree value between the candidate occasions and the target user group based on the obtained weight value; wherein the first similarity value is the similarity value greater than a first threshold value.

[0145] The first threshold may be 0.5.

[0146] In an embodiment of the present invention, the number of candidate occasions is at least two, and recommending occasion information based on the matching degree value includes: determining a recommendation order of the candidate occasions based on the matching degree value, popularity values ​​corresponding to the candidate occasions, and distance values; recommending occasion information based on the recommendation order; wherein the distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

[0147] Specifically, determining the recommended order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value includes: using formula 1 to determine the recommended order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; wherein the formula 1 is:

[0148]

[0149] P represents the recommendation order; α represents the weight value corresponding to the matching degree value;

[0150] represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold;

[0151] n represents the total number of the first similarity values; β represents the weight value corresponding to the popularity value corresponding to the candidate occasion; x j represents the popularity value corresponding to the jth candidate occasion; λ represents the weight value corresponding to the distance value corresponding to the candidate occasion; y j represents the distance value corresponding to the jth candidate occasion; l represents the total number of candidate occasions.

[0152] Correspondingly, in the embodiment of the present invention, determining the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value may include:

[0153] Formula 2 is used to determine the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0154] Wherein, the formula 2 is:

[0155]

[0156] matchRate represents the matching degree value; m i represents the weight value of the i-th first similarity value of the candidate occasion; wherein, the first similarity value is the similarity value greater than a first threshold; and n represents the total number of the first similarity values.

[0157] The information recommendation method provided by the embodiment of the present invention is further described below, taking a restaurant as an example.

[0158] First, two concepts involved in the following contents of the embodiments of the present invention are introduced:

[0159] The first one, ontology, is a conceptual abstraction of things in a domain and the relationships between things. It is organized in a hierarchical structure and is used to describe, represent and obtain knowledge in the relevant domain. Ontology can provide semantic knowledge and common understanding in the domain, determine the commonly recognized concepts (classes) and the relationships (attributes) between concepts in the domain, and share them between people, people and machines, and machines and machines. As a form of knowledge expression, ontology can well describe user information and restaurant information, which is of great significance for realizing multi-user restaurant recommendations.

[0160] The second one, semantic similarity, measures the similarity between semantic information from a quantitative perspective. As an important research direction in the field of natural language processing, word semantic similarity has been widely used in word sense disambiguation, retrieval, information extraction, and ontology matching and clipping. The typical method of calculating semantic similarity is to train a word vector model on a corpus based on the word2vec algorithm, and then calculate the cosine distance between words based on the word vector model as the semantic similarity of the words. For example, the word vectors of slightly spicy and spicy are:

[0161] Slightly spicy = (-0.878781 -2.112356 1.204208 1.041078 -1.249182 0.272419 ... -1.952982 2.179033 -0.258987 1.129659 0.363234);

[0162] Spicy = (0.033173 -0.252127 -0.794737 -1.141235 0.678041 2.522937 -0.833398 ... 1.822892 0.152297 -1.032652 -0.725105 0.790077);

[0163] The semantic similarity is: SimVec = cos (mildly spicy, spicy) = 0.78.

[0164] In view of the above technical problems, an embodiment of the present invention provides an information recommendation method. Taking restaurant recommendation as an example, a multi-user restaurant recommendation system based on semantic technology can be used to implement the method. The specific operation can be as follows: Figure 2 As shown, it includes: firstly building a user dinner group ( Figure 2 u1 represents user 1, u2 represents user 2, and un represents user n) and knowledge base (which may include user-restaurant common knowledge base, user personal information knowledge base, and restaurant information knowledge base); then semantic completion is performed on the user personal information knowledge base and the restaurant information knowledge base; secondly, based on the dining group, semantic fusion is performed on the semantically completed user personal information knowledge base; then semantic (ontology) matching is performed on the fused user information knowledge base and the semantically completed restaurant knowledge base, and finally restaurant recommendations are made based on the matching results. The specific operation is described as follows:

[0165] 1. About the construction of dining groups and knowledge base

[0166] (1) Establish a dining group

[0167] Many restaurant recommendation apps now have social functions, such as setting up the "My Follow" module in the app, which allows users to follow other users. After following, users can see where their friends have eaten and what kind of comments they have made about which restaurants. Based on this social function, a module for building a social group can be added, that is, when multiple users u1, u2, ... un have a meal together, they can use the social relationships in the recommendation system to establish a meal group. The role of this group is to conduct a comprehensive analysis of the personal information and dining preferences of each member in the group, so as to recommend restaurants that satisfy everyone.

[0168] (2) The knowledge base part mainly includes three parts: user-restaurant shared knowledge base, user personal information knowledge base, and restaurant information knowledge base.

[0169] The user-restaurant shared knowledge base is a shared semantic ontology constructed by experts (based on experience) based on all users' personal information and restaurant information (or semi-automatically, using existing technology to construct based on recipe keywords, etc.). This knowledge base contains all the knowledge related to recommendations shared by user preferences and restaurant characteristics (all classification features), and is a semantic abstraction of user preferences and restaurant characteristics. For example, the use of cuisine, taste, etc. is a generalization of the user's favorite cuisine and restaurant cuisine characteristics. The user-restaurant shared knowledge base is the foundation of the restaurant recommendation system. For more information about the user-restaurant shared knowledge base, see Figure 3 , Figure 3 The word "contain" means inclusion, "is-a" means belonging to, and "has-taste" means the taste possessed.

[0170] The user personal information knowledge base is a collection of each user's personal information ontology. Each user has his or her own information ontology, which represents the user's dining preferences and characteristics from a semantic level. For example, the information ontology of user u1 is Figure 4 : This ontology information shows that user u1 likes Japanese cuisine, Sichuan cuisine, and likes spicy and sweet flavors. This information can be used for subsequent restaurant recommendations.

[0171] The restaurant information knowledge base is a collection of each restaurant's information ontology. Each restaurant has its own information ontology, which represents the restaurant's information and characteristics from a semantic level. For example, the information ontology of restaurant R1 Figure 5 : This ontology information shows that restaurant R1 is a Sichuan restaurant, with special dishes such as spicy and fish-flavored. This information can be used for subsequent restaurant recommendations.

[0172] 2. About semantic completion of user personal information knowledge base and restaurant information knowledge base

[0173] The restaurant recommendation system can be combined with semantic technology to provide a semantic completion solution for the missing or incomplete characteristics of user or restaurant information. The solution can use the hierarchical relationship and reasoning ability of semantic technology to explore the implicit relationship between users themselves, users and communities, and restaurants and communities. The semantic completion solution can make recommendations more accurate and comprehensive. The specific semantic completion solution is as follows, taking the semantic completion of user personal information as an example, and the semantic completion of restaurant information is the same.

[0174] (1) First, semantic matching is performed between the user u1's user personal information ontology and the user restaurant's shared ontology. The basic principle of semantic matching is based on the calculation of semantic similarity. When the concept similarity in the two ontologies is greater than 0.5 and is a pair of concepts with the maximum value, matching is performed (matching is performed when the maximum similarity is obtained, and the maximum similarity is greater than 0.5);

[0175] (2) Then semantic completion is performed based on certain rules. The rules can be as follows:

[0176] Rule 1: semantic supplementation based on the hierarchical relationship. When the leaf node A in the user's personal information ontology matches the non-leaf node B in the user-restaurant shared ontology, node A is semantically supplemented, that is, the leaf node of the user-restaurant shared ontology node B is added to the user's personal information ontology node A as a subordinate child node of node A. This is because the leaf node in the user's information ontology matches the non-leaf node in the user-restaurant shared ontology, indicating that the leaf node in the user's information ontology is a higher-level abstraction without fine-grained division. Therefore, it can be assumed that the user will like the characteristics of the recipes after fine-grained division.

[0177] Rule 2: semantic supplement based on reasoning. When there are certain reasoning rules in the user restaurant common ontology that can infer some implicit preference information, the inferred part is added to the user personal information ontology for semantic supplement. For example, Hunan cuisine has the taste of fish flavor. If a user likes Hunan cuisine, then the user likes fish flavor. Vice versa.

[0178] Rule 3: When leaf node A in the user’s personal information ontology matches leaf node B in the user-restaurant shared ontology, no semantic supplement is required.

[0179] For example: Based on rules 1, 2, and 3, the personal information ontology of user u1 can be semantically supplemented (semantic matching) as follows: Figure 6 , the user personal information ontology after semantic completion is as follows Figure 7 shown.

[0180] The semantic completion of the restaurant information ontology is similar to the principle of semantic completion of the user personal information ontology, which will not be repeated here.

[0181] 3. Based on dining groups, semantic fusion of semantically completed user personal information knowledge base

[0182] This process is based on semantic similarity technology to semantically fuse the personal information ontology of each user in the dining group.

[0183] If there are n users (u1, u2...un) in the current user group, the corresponding semantically completed ontology is (o1, o2...on). Based on the calculation of semantic similarity, the n ontologies (o1, o2...on) are merged into a large ontology with weights.

[0184] The weight can be calculated as follows: in the process of ontology fusion (merging), when the similarity of concepts (features) in two ontologies is greater than 0.5 and is the maximum value of a pair of concepts, matching is performed. When a concept (feature) is a leaf node and is matched by m user ontologies, the weight coefficient of the leaf concept is m. The weight of a leaf node that has not been matched is 1. Finally, an ontology that integrates the preference information of multiple users is constructed.

[0185] 4. Semantic matching between the fused user information knowledge base and the semantically completed restaurant knowledge base

[0186] The weights of the leaf nodes of the restaurant information ontology are all set to 1. Then, based on the semantic similarity algorithm, the semantic ontology with weights after semantic fusion is semantically matched with the restaurant knowledge base after semantic completion. The weighted sum of the matched nodes (similarity greater than the first threshold) is performed to calculate the matching degree between the user and the restaurant. The formula is as follows:

[0187] Where n is the number of matching concepts (i.e. the first threshold above), m i is the weight of the i-th matching concept in the fused user information ontology.

[0188] 5. About forming a restaurant recommendation list (restaurant recommendations based on matching results)

[0189] Specifically, after calculating the matching degree between the user and the restaurant, the popularity value and distance value of each restaurant are weighted, and finally a restaurant recommendation list is obtained. The formula for generating the restaurant recommendation list can be:

[0190]

[0191] Where P represents the restaurant recommendation list (i.e., the above recommendation order); j is the jth restaurant, m ij represents the matching degree value of the matching pair of the i-th similarity of the j-th restaurant that is greater than the first threshold, and n represents the total number of matching pairs. represents the restaurant matching degree of the jth restaurant (i.e., the matching degree value mentioned above);

[0192] x j represents the restaurant popularity value corresponding to the jth restaurant, y j represents the distance value corresponding to the j-th restaurant (the distance value refers to the distance between the location of the restaurant and the location of the dining user group), α, β, and λ represent the restaurant matching degree, restaurant popularity value, and weight value of the distance value (which can be obtained based on experience), which can be 0.6, 0.3, and 0.2 respectively.

[0193] Finally, based on the sorting results, recommendations are made to the user group that is gathering for dinner, providing a list of restaurants that best suit everyone’s preferences for everyone to choose from.

[0194] There are two situations in actual application regarding the distance between the location of the dining user group and the restaurant location:

[0195] The first type is that all users are in one location. In this case, the distance between the user's location and the restaurant's location can be used as the distance between the location of the dining user group and the restaurant's location.

[0196] The second type is that the users in the dinner user group are located in at least two locations. In this case, the number of users corresponding to each location can be used as the weight value of the location, and the distance value between each location and the restaurant location can be weighted summed and averaged (averaged according to the total number of different locations of the users). The result is used as the distance value between the location of the dinner user group and the restaurant location, but it is not limited to this.

[0197] For example:

[0198] User 1, User 2, User 3, User 4, and User 5 form a dining user group. Among them, User 1, User 2, and User 3 are located at location A, and User 4 and User 5 are located at location B. The weighted value corresponding to location A is 0.6, and the weighted value corresponding to location B is 0.4. The distance value S between the location of the dining user group and the restaurant location C can be calculated as follows:

[0199] Get the distance value 1 between location A and restaurant location C and the distance value 2 between location B and restaurant location C;

[0200] The distance value 1 is multiplied by the weighted value 0.6 corresponding to position A to obtain the first value;

[0201] The distance value 2 is multiplied by the weighted value 0.4 corresponding to position B to obtain the second value;

[0202] Add the first value and the second value, divide by 2, and use the result as S.

[0203] As can be seen from the above, the solution provided by the embodiment of the present invention is aimed at the common situation of multiple users dining together in real life, such as company dinners, class dinners, and friends dinners, and provides a restaurant recommendation system for multiple users based on semantic technology:

[0204] (1) The system comprehensively considers the information and preferences of multiple users through steps such as knowledge base construction, semantic completion, semantic fusion, and ontology matching, and can accurately and comprehensively recommend restaurants;

[0205] (2) The restaurant recommendation system is combined with semantic technology. In view of the fact that user or restaurant information is missing or incomplete, a semantic completion solution is provided. The solution uses the upper and lower relationships and reasoning capabilities of semantic technology to explore the implicit relationships between users themselves, users and communities, and restaurants and communities. The semantic completion solution can make recommendations more accurate and comprehensive. For example, through semantic technology, it can be inferred that the user likes spicy dishes based on the fact that the user likes Sichuan cuisine, so spicy dishes can be recommended to the user, thereby improving the recall rate and accuracy.

[0206] (3) This scheme also provides a semantic fusion scheme based on the semantic similarity calculation method, which can fuse the personalized information of multiple users to form a weighted multi-user information ontology, providing a basis for subsequent restaurant recommendations.

[0207] Through the above three points, this solution can effectively provide a comprehensive and highly accurate restaurant recommendation solution based on the dining characteristics of multiple users, provide satisfactory restaurant recommendations, reduce the complexity of users' selection from a large number of restaurants, improve the efficiency of dining selection, and also improve the satisfaction of diners with the selected restaurant or food.

[0208] The embodiment of the present invention also provides an information recommendation device, such as Figure 8 As shown, including:

[0209] The first processing module 81 is used to semantically fuse the personal preference information of each user in the target user group to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information;

[0210] A first comparison module 82, configured to perform semantic comparison between each of the preference information and each of the feature information of the candidate occasions, and obtain a similarity value corresponding to each of the feature information;

[0211] A first determination module 83, used to determine a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0212] The first recommendation module 84 is used to recommend occasion information according to the matching degree value.

[0213] The information recommendation device provided by the embodiment of the present invention obtains each preference information corresponding to the target user group and the weight value corresponding to each preference information by semantically fusing the personal preference information of each user in the target user group; semantically compares each preference information with each feature information of the candidate occasion to obtain the similarity value corresponding to each feature information; determines the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; recommends occasion information according to the matching degree value; can comprehensively consider the information and preferences of multiple users, so as to accurately and comprehensively give occasion recommendations; specifically, can effectively provide satisfactory restaurant recommendations according to the dining characteristics of multiple users, reduce the complexity of users' selection from a large number of restaurants, improve the efficiency of dining selection, and also can improve the satisfaction of diners with the selected restaurant or food; and well solves the problem of low recommendation accuracy of restaurant recommendation systems for multiple users in the prior art.

[0214] Furthermore, the information recommendation device further includes: a second processing module, which is used to semantically complete the personal preference information and the candidate occasion information before semantically fusing the personal preference information of each user in the target user group.

[0215] Among them, the second processing module includes: a first processing submodule, used to obtain common information of occasions based on personal preference information of a preset user and characteristic information of a preset occasion; a second processing submodule, used to semantically complete the personal preference information and candidate occasion information based on the common information of the occasion.

[0216] Specifically, the semantically completing the personal preference information and the candidate occasion information includes: semantically completing the personal preference information and the candidate occasion information according to the hierarchical relationship information; or, semantically completing the personal preference information and the candidate occasion information according to a preset inference rule.

[0217] In an embodiment of the present invention, the first processing module includes: a first acquisition submodule, used to obtain the semantic similarity value between the personal preference information of each user; a third processing submodule, used to integrate the personal preference information of each user according to the semantic similarity value, to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information.

[0218] Specifically, the first determination module includes: a second acquisition submodule, used to obtain the weight value corresponding to the preference information corresponding to each first similarity value; a first determination submodule, used to determine the matching degree value between the candidate occasion and the target user group based on the acquired weight value; wherein, the first similarity value is the similarity value greater than the first threshold.

[0219] In an embodiment of the present invention, the number of candidate occasions is at least two, and the first recommendation module includes: a second determination submodule, used to determine the recommendation order of the candidate occasions based on the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; a first recommendation submodule, used to recommend occasion information based on the recommendation order; wherein the distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

[0220] Specifically, the second determination submodule is specifically used to: use Formula 1 to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; wherein the Formula 1 is:

[0221]

[0222] P represents the recommendation order; α represents the weight value corresponding to the matching degree value;

[0223] represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold;

[0224] n represents the total number of the first similarity values; β represents the weight value corresponding to the popularity value corresponding to the candidate occasion; x j represents the popularity value corresponding to the jth candidate occasion; λ represents the weight value corresponding to the distance value corresponding to the candidate occasion; y j represents the distance value corresponding to the jth candidate occasion; l represents the total number of candidate occasions.

[0225] Among them, the implementation embodiments of the above-mentioned information recommendation method are all applicable to the embodiments of the information recommendation device and can achieve the same technical effects.

[0226] The embodiment of the present invention also provides an information recommendation device, such as Fig. 9 As shown, it includes: a processor 91;

[0227] The processor 91 is used to semantically fuse the personal preference information of each user in the target user group to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information;

[0228] Semantically comparing each of the preference information with each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information;

[0229] Determining a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value;

[0230] According to the matching degree value, occasion information is recommended.

[0231] The information recommendation device may also include a transceiver, which is not limited here.

[0232] The information recommendation device provided by the embodiment of the present invention obtains each preference information corresponding to the target user group and the weight value corresponding to each preference information by semantically fusing the personal preference information of each user in the target user group; semantically compares each preference information with each feature information of the candidate occasion to obtain the similarity value corresponding to each feature information; determines the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; recommends occasion information according to the matching degree value; can comprehensively consider the information and preferences of multiple users, so as to accurately and comprehensively give occasion recommendations; specifically, can effectively provide satisfactory restaurant recommendations according to the dining characteristics of multiple users, reduce the complexity of users' selection from a large number of restaurants, improve the efficiency of dining selection, and also can improve the satisfaction of diners with the selected restaurant or food; and well solves the problem of low recommendation accuracy of restaurant recommendation systems for multiple users in the prior art.

[0233] Furthermore, the processor is also used to: before semantically fusing the personal preference information of each user in the target user group, semantically complete the personal preference information and the candidate occasion information.

[0234] The processor is specifically used to: obtain common information of occasions according to personal preference information of preset users and characteristic information of preset occasions; and semantically complete the personal preference information and candidate occasion information according to the common information of occasions.

[0235] Specifically, the processor is used to: semantically complete the personal preference information and candidate occasion information according to the hierarchical relationship information; or semantically complete the personal preference information and candidate occasion information according to a preset inference rule.

[0236] In an embodiment of the present invention, the processor is specifically used to: obtain a semantic similarity value between personal preference information of each user; and integrate the personal preference information of each user according to the semantic similarity value to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

[0237] Specifically, the processor is used to: obtain the weight value corresponding to the preference information corresponding to each first similarity value; determine the matching degree value between the candidate occasion and the target user group based on the obtained weight value; wherein the first similarity value is the similarity value greater than a first threshold.

[0238] In an embodiment of the present invention, the processor is specifically used to: determine the recommendation order of the candidate occasions based on the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; recommend occasion information based on the recommendation order; wherein the distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

[0239] Specifically, the processor is specifically used to: use Formula 1 to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; wherein the Formula 1 is:

[0240]

[0241] P represents the recommendation order; α represents the weight value corresponding to the matching degree value;

[0242] represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold;

[0243] n represents the total number of the first similarity values; β represents the weight value corresponding to the popularity value corresponding to the candidate occasion; x j represents the popularity value corresponding to the jth candidate occasion; λ represents the weight value corresponding to the distance value corresponding to the candidate occasion; y j represents the distance value corresponding to the jth candidate occasion; l represents the total number of candidate occasions.

[0244] Among them, the implementation embodiments of the above-mentioned information recommendation method are all applicable to the embodiments of the information recommendation device and can also achieve the same technical effects.

[0245] An embodiment of the present invention further provides an information recommendation device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; when the processor executes the program, the above-mentioned information recommendation method is implemented.

[0246] Among them, the implementation embodiments of the above-mentioned information recommendation method are all applicable to the embodiments of the information recommendation device and can also achieve the same technical effects.

[0247] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the steps in the above-mentioned information recommendation method are implemented.

[0248] Among them, the implementation embodiments of the above-mentioned information recommendation method are all applicable to the embodiments of the computer-readable storage medium and can achieve the same technical effects.

[0249] It should be noted that many functional components described in this specification are referred to as modules / sub-modules in order to more particularly emphasize the independence of their implementation methods.

[0250] In the embodiment of the present invention, module / submodule can be implemented with software so that it can be performed by various types of processors. For example, an executable code module of an identification can include one or more physical or logical blocks of computer instructions, for example, it can be constructed as an object, process or function. Nevertheless, the executable code of the identified module does not need to be physically located together, but can include different instructions stored in different positions, and when these instructions are logically combined together, it constitutes a module and realizes the specified purpose of this module.

[0251] In fact, executable code module can be a single instruction or many instructions, and can even be distributed on a plurality of different code segments, distributed among different programs, and distributed across a plurality of memory devices. Similarly, operating data can be identified in the module, and can be implemented and organized in the data structure of any appropriate type according to any appropriate form. The operating data can be collected as a single data set, or can be distributed in different locations (including on different storage devices), and can only be present on a system or network as an electronic signal at least in part.

[0252] When a module can be implemented by software, considering the level of existing hardware technology, a person skilled in the art can build a corresponding hardware circuit to implement the corresponding function of the module that can be implemented by software without considering the cost. The hardware circuit includes a conventional very large scale integration (VLSI) circuit or gate array and existing semiconductors such as logic chips, transistors, or other discrete components. The module can also be implemented by a programmable hardware device, such as a field programmable gate array, a programmable array logic, a programmable logic device, etc.

[0253] The above is a preferred embodiment of the present invention. It should be pointed out that for ordinary personnel in this technical field, several improvements and modifications can be made without departing from the principles of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. An information recommendation method, characterized in that: include: Semantically fuse the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information; Semantically comparing each of the preference information with each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information; Determining a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; Recommending occasion information according to the matching degree value; Before semantically fusing the personal preference information of each user in the target user group, the following steps are also included: Performing semantic completion on the personal preference information and candidate occasion information; The semantic fusion of the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information includes: Obtain the semantic similarity value between each user's personal preference information; According to the semantic similarity value, the personal preference information of each user is integrated to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

2. The information recommendation method according to claim 1, characterized in that: The semantically completing the personal preference information and the candidate occasion information includes: Obtaining common information of the occasion according to the personal preference information of the preset user and the characteristic information of the preset occasion; According to the common occasion information, semantic completion is performed on the personal preference information and the candidate occasion information.

3. The information recommendation method according to claim 1 or 2, characterized in that: The semantically completing the personal preference information and the candidate occasion information includes: According to the upper and lower relationship information, semantically complete the personal preference information and the candidate occasion information; or According to preset inference rules, semantic completion is performed on the personal preference information and candidate occasion information.

4. The information recommendation method according to claim 1, characterized in that: Determining the matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value includes: Obtaining a weight value corresponding to the preference information corresponding to each first similarity value; Determining the matching degree value between the candidate occasion and the target user group according to the obtained weight value; The first similarity value is the similarity value greater than a first threshold.

5. The information recommendation method according to claim 1, characterized in that: The number of the candidate occasions is at least two, and the recommending of occasion information according to the matching degree value includes: Determining a recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; Recommending occasion information according to the recommendation order; The distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

6. The information recommendation method according to claim 5, characterized in that: The determining the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value includes: Formula 1 is used to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; Wherein, the formula 1 is: P represents the recommended order; α represents the weight value corresponding to the matching degree value; represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold; n represents the total number of the first similarity values; β represents the weight value corresponding to the popularity value corresponding to the candidate occasion; x j represents the popularity value corresponding to the j-th candidate occasion; λ represents the weight value corresponding to the distance value corresponding to the candidate occasion; y j represents the distance value corresponding to the j-th candidate occasion; l represents the total number of candidate occasions.

7. An information recommendation device, characterized in that: include: The first processing module is used to semantically fuse the personal preference information of each user in the target user group to obtain the preference information corresponding to the target user group and the weight value corresponding to each preference information; A first comparison module is used to perform semantic comparison between each of the preference information and each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information; A first determination module, used to determine a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; A first recommendation module, used for recommending occasion information according to the matching degree value; Wherein, the information recommendation device further includes: The second processing module is used to semantically complete the personal preference information and the candidate occasion information before semantically fusing the personal preference information of each user in the target user group; Wherein, the first processing module includes: The first acquisition submodule is used to obtain the semantic similarity value between the personal preference information of each user; The third processing submodule is used to integrate the personal preference information of each user according to the semantic similarity value to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

8. The information recommendation device according to claim 7, characterized in that: The second processing module comprises: The first processing submodule is used to obtain the common information of the occasion according to the personal preference information of the preset user and the characteristic information of the preset occasion; The second processing submodule is used to semantically complete the personal preference information and the candidate occasion information according to the occasion common information.

9. The information recommendation device according to claim 7 or 8, characterized in that: The semantically completing the personal preference information and the candidate occasion information includes: According to the upper and lower relationship information, semantically complete the personal preference information and the candidate occasion information; or According to preset inference rules, semantic completion is performed on the personal preference information and candidate occasion information.

10. The information recommendation device according to claim 7, characterized in that: The first determining module includes: A second acquisition submodule is used to obtain a weight value corresponding to the preference information corresponding to each first similarity value; A first determination submodule, used to determine a matching degree value between the candidate occasion and the target user group according to the acquired weight value; The first similarity value is the similarity value greater than a first threshold.

11. The information recommendation device according to claim 7, characterized in that: The number of the candidate occasions is at least two, and the first recommendation module includes: A second determination submodule, configured to determine a recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; A first recommendation submodule, used for recommending occasion information according to the recommendation order; The distance value is the distance value between the geographical location of the candidate occasion and the geographical location of the target user group.

12. The information recommendation device according to claim 11, characterized in that: The second determining submodule is specifically used for: Formula 1 is used to determine the recommendation order of the candidate occasions according to the matching degree value, the popularity value corresponding to the candidate occasions, and the distance value; Wherein, the formula 1 is: P represents the recommended order; α represents the weight value corresponding to the matching degree value; represents the matching degree value of the jth candidate occasion, m ij represents the i-th first similarity value of the j-th candidate occasion; the first similarity value is the similarity value greater than the first threshold; n represents the total number of the first similarity values; β represents the weight value corresponding to the popularity value corresponding to the candidate occasion; x j represents the popularity value corresponding to the j-th candidate occasion; λ represents the weight value corresponding to the distance value corresponding to the candidate occasion; y j represents the distance value corresponding to the j-th candidate occasion; l represents the total number of candidate occasions.

13. An information recommendation device, characterized in that: include: processor; The processor is used to semantically fuse the personal preference information of each user in the target user group to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information; Semantically comparing each of the preference information with each of the feature information of the candidate occasions to obtain a similarity value corresponding to each of the feature information; Determining a matching degree value between the candidate occasion and the target user group according to the similarity value and the weight value; Recommending occasion information according to the matching degree value; Wherein, the processor is further used for: Before semantically fusing the personal preference information of each user in the target user group, semantically completing the personal preference information and the candidate occasion information; The processor is specifically used for: Obtain the semantic similarity value between each user's personal preference information; According to the semantic similarity value, the personal preference information of each user is integrated to obtain each preference information corresponding to the target user group and a weight value corresponding to each preference information.

14. An information recommendation device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor; characterized in that: When the processor executes the program, the information recommendation method according to any one of claims 1 to 6 is implemented.

15. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps in the information recommendation method according to any one of claims 1 to 6 are implemented.

Images