Information recommendation method, device, computer equipment and storage medium

Abstract

The present invention relates to artificial intelligence technology, and discloses an information recommendation method, device, computer equipment and storage medium, which belongs to the field of Internet technology. The information recommendation method can obtain a first feature tag according to the user's identification data, and construct a user-tag weight matrix associated with the user's interests and basic information according to the first feature tag and the second feature tag input by the user; considering that one recommended information may correspond to multiple recommended tags, in order to improve the accuracy of the recommendation, the recommended tags and corresponding tag weights of each recommended information are obtained from the recommendation library to generate a recommended-tag weight matrix; a user-recommendation association matrix that associates user tags with recommended tags is constructed according to the user-tag weight matrix and the recommended-tag weight matrix, and a recommendation list that meets the user's interests and basic information is generated by calculating the similarity between the user tag and the recommended tag weight in the user-recommendation association matrix, thereby achieving the purpose of accurate recommendation.

Description

Technical Field

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

[0002] Intelligent algorithm recommendation refers to pushing the right content to the right people. It has obvious advantages, such as higher input-output ratio, wider coverage, and higher personalization. The usage rate of intelligent recommendation algorithms on websites is increasing and has become a norm. Currently, there are two main widely used recommendation methods: collaborative recommendation for recommending popular information and content recommendation based on users' different interests and hobbies.

[0003] The collaborative recommendation method emphasizes the current hot spots when presenting information and public information. The recommended content is mainly based on popularity and timeliness, while ignoring the personalized needs of users / employees. The content recommendation method extracts corresponding features from each pushed content to represent the content features; generates user preferences based on the feature data of user interests in content (such as: browsed content, browsing time, number of likes, etc.); and pushes the content that users are most likely to be interested in based on content features and user preferences. However, for lengthy and cumbersome content, the extracted features can only represent one aspect of the content, not all of it, which affects the accuracy of the pushed content; content is only recommended based on the user's past interest in certain content, and content cannot be recommended based on the user's potential interests; and suitable content cannot be recommended for new users.

[0004] In summary, existing recommendation methods cannot accurately recommend content based on user situations (such as users' potential interests, new users whose hobbies are not known). Summary of the invention

[0005] In view of the problem that existing recommendation methods cannot make accurate recommendations based on user conditions, a method for information recommendation is provided to enable accurate recommendations based on user feature data.

[0006] To achieve the above object, the present invention provides an information recommendation method, comprising:

[0007] Collecting identification data of the user, and obtaining a first feature tag of the user according to the identification data;

[0008] Acquire a second feature tag input by the user;

[0009] Constructing a user-label weight matrix according to the first feature label and the second feature label;

[0010] Obtain the recommended labels and corresponding label weights of each recommended information from the recommendation library, and generate a recommendation-label weight matrix;

[0011] Constructing a user-recommendation association matrix according to the user-tag weight matrix and the recommendation-tag weight matrix;

[0012] The similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix is ​​calculated, and a recommendation list is generated according to the similarity.

[0013] Optionally, collecting identification data of the user and acquiring a first feature tag of the user according to the identification data includes:

[0014] Collecting identification data of the user, and obtaining basic information of the user from the human resources system according to the identification data;

[0015] Extracting feature data from the basic information, and converting the feature data into a first feature tag;

[0016] The first feature tag includes at least one of a department tag, a position tag, an interest tag, a birthday tag, a zodiac tag, an address tag, a gender tag, and an age tag.

[0017] Optionally, constructing a user-label weight matrix according to the first feature label and the second feature label includes:

[0018] Collecting current information, wherein the current information includes at least one of current time data, current address data, current weather data, and current temperature data;

[0019] Matching the current information with the first feature tag to obtain matching data;

[0020] Determine a tag weight according to the matching data and the second feature tag;

[0021] The user-tag weight matrix is ​​generated based on the tag weights.

[0022] Optionally, the recommendation library includes recommendation information carrying preset tags and recommendation information not carrying preset tags;

[0023] The step of obtaining the recommendation tags and corresponding tag weights of each recommendation information from the recommendation library and generating a recommendation-tag weight matrix includes:

[0024] Segmenting the titles of the recommended information in the recommendation library respectively, and generating recommendation tags corresponding to the recommended information according to the segmented words;

[0025] Determining the weight of the recommended tag and the weight of the preset tag based on the proportion of the recommended tag in the recommended information and the proportion of the preset tag in the recommended information;

[0026] When the recommendation information carries a preset tag, a recommendation-tag weight matrix is ​​generated based on the weight of the recommendation tag of the recommendation information and the weight of the preset tag; when the recommendation information does not carry a preset tag, a recommendation-tag weight matrix is ​​generated based on the weight of the recommendation tag.

[0027] Optionally, constructing a user-recommendation association matrix according to the user-tag weight matrix and the recommendation-tag weight matrix includes:

[0028] The user tags in the user-tag weight matrix are associated with the recommendation information in the recommendation-tag weight matrix according to the user tags and the recommendation tags, and the user-recommendation association matrix is ​​generated by combining them.

[0029] Optionally, calculating the similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix, and generating a recommendation list according to the similarity includes:

[0030] Calculate the value evaluation of the recommendation tag corresponding to each recommendation information in the user-recommendation association matrix;

[0031] Using the value assessment amount as the similarity of the recommended tag weight;

[0032] The recommendation information corresponding to the recommendation tag weights is sorted according to the order of the similarities from large to small to generate the recommendation list.

[0033] Optionally, the recommendation list adopts an integrated recommendation mode, a multi-row parallel recommendation mode or a pipeline recommendation mode.

[0034] To achieve the above object, the present invention further provides an information recommendation device, comprising:

[0035] an acquisition unit, configured to collect identification data of a user, and acquire a first feature tag of the user according to the identification data;

[0036] A receiving unit, configured to receive a second feature tag input by the user;

[0037] A first construction unit, configured to construct a user-label weight matrix according to the first feature label and the second feature label;

[0038] The second construction unit is used to obtain the recommendation tags and corresponding tag weights of each recommendation information from the recommendation library, and generate a recommendation-tag weight matrix;

[0039] A third construction unit, configured to construct a user-recommendation association matrix according to the user-label weight matrix and the recommendation-label weight matrix;

[0040] The generating unit is used to calculate the similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix, and generate a recommendation list according to the similarity.

[0041] To achieve the above objectives, the present invention also provides a computer device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program.

[0042] To achieve the above object, the present invention further provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the steps of the above method when executed by a processor.

[0043] The information recommendation method, device, computer equipment and storage medium provided by the present invention can obtain a first feature tag according to the user's identification data, and construct a user-tag weight matrix associated with the user's interests and basic information according to the first feature tag and the second feature tag input by the user; considering that one recommended information may correspond to multiple recommended tags, in order to improve the accuracy of the recommendation, the recommended tags and corresponding tag weights of each recommended information are obtained from the recommendation library to generate a recommendation-tag weight matrix; a user-recommendation association matrix that associates user tags with recommended tags is constructed according to the user-tag weight matrix and the recommendation-tag weight matrix, and a recommendation list that meets the user's interests and basic information is generated by calculating the similarity between the user tag and the recommended tag weight in the user-recommendation association matrix, thereby achieving the purpose of accurate recommendation. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A method flow chart of an embodiment of the information recommendation method of the present invention;

[0045] Figure 2 A flowchart of an embodiment of the present invention for obtaining a first feature tag according to identification data;

[0046] Figure 3 A flowchart of an embodiment of constructing a user-label weight matrix for the present invention;

[0047] Figure 4 A flowchart of an embodiment of generating a recommendation-label weight matrix for the present invention;

[0048] Figure 5 A flowchart of an embodiment of the present invention for generating a recommendation list according to similarity;

[0049] Figure 6 A module diagram of an embodiment of the information recommendation device of the present invention;

[0050] Figure 7A module diagram of an embodiment of the first building unit of the present invention;

[0051] Figure 8 A module diagram of an embodiment of the second building unit of the present invention;

[0052] Fig. 9 The figure is a hardware architecture diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0053] In order to make the purpose, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not intended to limit the present application. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0054] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0055] The information recommendation method, device, computer equipment and storage medium provided by the present invention involve artificial intelligence technology and can be applied to insurance, finance, medical care, leasing and other business fields. The present invention can obtain a first feature tag according to the user's identification data, and construct a user-tag weight matrix associated with the user's interests and basic information according to the first feature tag and the second feature tag input by the user; considering that one recommended information may correspond to multiple recommended tags, in order to improve the accuracy of the recommendation, the recommended tags and corresponding tag weights of each recommended information are obtained from the recommendation library to generate a recommendation-tag weight matrix; a user-recommendation association matrix that associates user tags with recommended tags is constructed according to the user-tag weight matrix and the recommendation-tag weight matrix, and a recommendation list that meets the user's interests and basic information is generated by calculating the similarity between the user tag and the recommended tag weight in the user-recommendation association matrix, thereby achieving the purpose of accurate recommendation.

[0056] Embodiment 1

[0057] See also Figure 1 , an information recommendation method of this embodiment includes the following steps:

[0058] S1. Collect identification data of the user, and obtain a first feature tag of the user according to the identification data.

[0059] It should be noted that the first feature tag may include at least one of a department tag, a position tag, an interest tag, a birthday tag, a zodiac tag, an address tag, a gender tag, and an age tag.

[0060] In this embodiment, the information recommendation method is mainly used to recommend information to employees within the enterprise, and the users are mainly employees of the enterprise. The identification data refers to data used to identify user information, such as employee number, ID number, etc.

[0061] For further information, see Figure 2 The step S1 shown may include the following steps:

[0062] S11. Collect the user's identification data, and obtain the user's basic information from the human resources system based on the identification data.

[0063] In this embodiment, the basic information may include the user's resume information, department information, position information, etc. The human resources system stores the basic information of the user. The human resources system of the enterprise is queried according to the identification data to obtain the basic information of the user associated with the identification data.

[0064] S12. Extract feature data from the basic information, and convert the feature data into a first feature tag.

[0065] In this embodiment, corresponding feature data can be extracted according to the labels in the basic information. For example, corresponding feature data can be extracted according to the labels of each item in the resume information to obtain the user's age, birthday, zodiac sign, residential address, gender, major, and interests; the user's department information, position information, and other information can also be obtained directly from the basic information, and the extracted feature data can be matched with the preset label to determine the first feature tag that matches it, thereby obtaining the first feature tag in a unified format.

[0066] S2. Obtain a second feature tag input by the user.

[0067] It should be noted that there can be multiple second feature tags, which are interest and hobby tags entered by the user (such as sports tags, entertainment tags, game tags, astronomy hobby tags, literature and art tags or musical instrument tags, etc.), or they can be empty. If the user has no interests or hobbies, he or she can choose not to enter data and set the second feature tag to empty.

[0068] S3. Construct a user-label weight matrix according to the first feature label and the second feature label.

[0069] Further, see Figure 3 As shown, step S3 may include the following steps:

[0070] S31. Collect current information.

[0071] In this embodiment, the current information can be obtained through the user's client.

[0072] As an example but not limitation, the current information may include at least one of current time data, current address data, current weather data, and current temperature data.

[0073] S32. Match the current information with the first feature tag to obtain matching data.

[0074] In this embodiment, the current information is matched with the first feature tag to determine whether there is data belonging to the current information in the first feature tag. If there is data that can be highlighted in a targeted manner, the information matching the current information is generally selected in the recommended information to improve the real-time nature of the recommendation. For example, if the date in the current information is the user's birthday, that is, the current time data matches the user's birthday tag (the day is the user's birthday), the birthday tag is extracted, and the birthday tag is used as a high-priority tag.

[0075] S33. Determine a tag weight according to the matching data and the second feature tag.

[0076] In this embodiment, if there is matching data, the weight of each label in the first feature label is adjusted according to the preset rules, the weight of the matched first feature label is increased, and the weight of the unmatched first feature label is reduced. The preset rules may include: when the feature labels in the first feature label and the second feature label involve group labels, it is judged whether the weight value of the group label is less than the third weight threshold, if so, the third weight threshold is used as the weight value of the group label, if not, the weight value of the group label is kept unchanged; when the feature labels in the first feature label and the second feature label involve interest and hobby labels, it is judged whether the weight value of the interest and hobby label is less than the third weight threshold, if so, the third weight threshold is used as the weight value of the interest and hobby label, if not, the weight value of the interest and hobby label is kept unchanged.

[0077] As an example but not a limitation, considering that some users have common or similar behavioral hobbies (preferences), interest tags with the same hobbies can be classified as group tags, and it is determined whether the weight value of the group tag is less than the third weight threshold. If so, the weight value of the group tag is increased, and the third weight threshold is used as the weight value of the group tag. If the user's preference types are wide and have low repetition in the group user data, the type of recommended information that the user liked in the past, that is, the recommended tag, is extracted as the user tag, and the weight value of the user tag is increased; if the user provides personal information / personalized settings (interests, hobbies, specialties, etc.), these tags can be used as the most important evaluation data, and the data provided by the user is given priority, and the weight value of this type of tag is increased.

[0078] Furthermore, in step S33, when the second feature tag is not empty, the second feature tag is matched with the first feature tag. If they do not match, the weight of the second feature tag is set to the first weight threshold according to the total number of user feature tags; if they match, the weight of the second feature tag is increased and the weight is increased to the second weight threshold (the second weight threshold is greater than the first weight threshold).

[0079] S34. Generate the user-tag weight matrix based on the tag weights.

[0080] In this embodiment, a user-label weight matrix is ​​constructed according to the weight values ​​of each first feature label, the weight values ​​of the second feature label, and the preset label.

[0081] The preset tags are all tags provided in advance, including the first feature tag and the second feature tag.

[0082] Take the preset 10 tags (b0, b1, ..., b9) as an example: User A involves 3 tags: b2, b5 and b8, among which the first feature tag includes b2 and b5, the second feature tag is b8, the weight value of b2 is 0.5, the weight value of b5 is 0.3, and the weight value of b8 is 0.2. The tag weight matrix of user A is:

[0083]

[0084] S4. Obtain the recommended tags and corresponding tag weights of each recommended information from the recommendation library, and generate a recommendation-tag weight matrix.

[0085] The recommendation library includes recommendation information with preset tags and recommendation information without preset tags.

[0086] In this embodiment, the recommendation information is stored in the recommendation library. Each piece of recommendation information is associated with at least one recommendation tag. The recommendation information may be pre-tagged or untagged.

[0087] Further, see Figure 4 As shown, step S4 may include the following steps:

[0088] S41. Segment the titles of the recommended information in the recommendation library respectively, and generate recommendation tags corresponding to the recommended information according to the segmentation.

[0089] In this embodiment, the title of the recommended information is segmented, keywords are extracted, the keywords are matched with preset tags, and the matched tags are used as recommended tags.

[0090] S42. Determine the weight of the recommended tag and the weight of the preset tag based on the recommended tag ratio of the recommended information and the preset tag ratio of the recommended information.

[0091] Furthermore, when the recommendation information carries a preset tag, the number of recommended tags in the content of the recommendation information and the number of the preset tags in the content are obtained, and the preset tags are matched with the recommended tags. If they match, the matched recommended tags are extracted, the proportion of each recommended tag in the recommendation information is counted, and the tag weight is set according to the proportion (e.g., the weight of the tag is the proportion of the tag); if they do not match, the proportion of each recommended tag and the preset tag in the recommendation information is counted, and the tag weight is set according to the proportion (e.g., a table associating proportion values ​​with tag weights is provided, each tag weight value corresponds to a proportion interval, and the corresponding weight value is selected according to the proportion of the tag). When the recommendation information does not carry a preset tag (i.e., the proportion of the preset tag is zero), the number of recommended tags in the content of the recommendation information is obtained, the proportion of each recommended tag in the recommendation information is counted, and the corresponding tag weight is set according to the proportion.

[0092] S43. When the recommendation information carries a preset tag, a recommendation-tag weight matrix is ​​generated according to the weight of the recommendation tag of the recommendation information and the weight of the preset tag; when the recommendation information does not carry a preset tag, a recommendation-tag weight matrix is ​​generated according to the weight of the recommendation tag.

[0093] In another embodiment, step S43 may also be:

[0094] Based on the user tag, the recommendation information associated with the recommendation tag (including preset tags) corresponding to the user tag is extracted from the recommendation library to form a candidate set, and a recommendation-tag weight matrix is ​​constructed according to the recommendation tags associated with each recommendation information in the candidate set.

[0095] S5. Construct a user-recommendation association matrix according to the user-tag weight matrix and the recommendation-tag weight matrix.

[0096] Furthermore, step S5 associates the user tags in the user-tag weight matrix with the recommendation information in the recommendation-tag weight matrix according to the user tags and the recommendation tags, and combines them to generate the user-recommendation association matrix.

[0097] The label weight matrix of user A is:

[0098]

[0099] The recommendation-label weight matrix of recommendation information C is:

[0100]

[0101] Among them, there are 4 recommended information: C1, C2, C3, C4 and C5.

[0102] The weight matrix of C1 is:

[0103]

[0104] The user-recommendation association matrix obtained by combining the label weight matrix of user A with the recommendation-label weight matrix of recommendation information C is:

[0105]

[0106] S6. Calculate the similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix, and generate a recommendation list according to the similarity.

[0107] The recommendation list may adopt an integrated recommendation mode, a multi-row parallel recommendation mode or a pipeline recommendation mode.

[0108] In actual applications, the integrated recommendation mode can be applied to website recommendation column push, the multi-line parallel recommendation mode is mainly applicable to email push that requires filling in information or answering questionnaires, and the assembly line recommendation mode can be applied to public account push.

[0109] In this embodiment, the similarity between the recommendation tag of each piece of promotion information and the user tag is calculated, and a recommendation list consisting of the recommendation information is generated in order of the similarity.

[0110] Further, see Figure 5 As shown, step S6 may include:

[0111] S61. Calculate the value evaluation of the recommendation tag corresponding to each recommendation information in the user-recommendation association matrix.

[0112] In this embodiment, the value evaluation quantity of the recommendation tag and the user tag of each piece of advancement information is calculated as well as the recommendation weight value.

[0113] S62. Using the value assessment amount as the similarity of the recommended tag weight.

[0114] S63. Sort the recommendation information corresponding to the recommendation tag weights according to the order of the similarities from large to small, and generate the recommendation list.

[0115] As an example but not a limitation, take the application of the information recommendation method to the "advice" push of the leasing portal website to the internal employees of the enterprise as an example. In the work area of ​​the portal website, the system will intelligently push different adverbs for each employee. The adverbs are stored in the recommendation library, and each adverb has a preset recommendation tag. Considering that the adverbs pushed to each employee are relatively independent, after collecting the employee's employee number, the basic information of the user is obtained according to the employee number, the feature data in the basic information is converted into a first feature tag, and a user-tag weight matrix is ​​constructed according to the first feature tag. A user-recommendation association matrix is ​​constructed according to the user-tag weight matrix and the recommendation-tag weight matrix of the recommendation library, and the similarity between the employee tag and the recommendation tag of each employee is calculated, and the adverbs corresponding to the recommendation tag with the greatest similarity are pushed to the corresponding employee.

[0116] The information recommendation method can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld or portable devices, tablet devices, multiprocessor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0117] In this embodiment, the information recommendation method can obtain the first feature tag according to the user's identification data, and construct a user-tag weight matrix associated with the user's interests and basic information according to the first feature tag and the second feature tag input by the user; considering that one recommended information may correspond to multiple recommended tags, in order to improve the accuracy of the recommendation, the recommended tags and corresponding tag weights of each recommended information are obtained from the recommendation library to generate a recommendation-tag weight matrix; a user-recommendation association matrix associated with user tags and recommended tags is constructed according to the user-tag weight matrix and the recommendation-tag weight matrix, and a recommendation list that satisfies the user's interests and basic information is generated by calculating the similarity between the user tag and the recommended tag weight in the user-recommendation association matrix, thereby achieving the purpose of accurate recommendation. The information recommendation method can provide multi-scenario and multi-dimensional technical support for pushing information, truly realizing the portal website in various scenarios for thousands of people, and can be applied to the advertising column, employee work area, latest information column and other scenarios of the rental portal website.

[0118] Embodiment 2

[0119] See also Figure 6 An information recommendation device 1 of this embodiment includes: an acquisition unit 11, a receiving unit 12, a first construction unit 13, a second construction unit 14, a third construction unit 15 and a generation unit 16.

[0120] The acquisition unit 11 is used to collect identification data of the user and acquire a first feature tag of the user according to the identification data.

[0121] It should be noted that the first feature tag may include at least one of a department tag, a position tag, an interest tag, a birthday tag, a zodiac tag, an address tag, a gender tag, and an age tag.

[0122] In this embodiment, the information recommendation device 1 is mainly used to recommend information to employees within the enterprise, and the users are mainly employees of the enterprise. Identification data refers to data used to identify user information, such as employee number, ID number, etc. The information recommendation device 1 can be applied to a server, which can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDN), and big data and artificial intelligence platforms.

[0123] Furthermore, the acquisition unit 11 is used to collect identification data of the user, obtain basic information of the user from the human system according to the identification data, extract feature data from the basic information, and convert the feature data into a first feature tag.

[0124] In this embodiment, the basic information may include the user's resume information, department information, position information, etc. The human resources system stores the user's basic information. The human resources system of the enterprise is queried according to the identification data to obtain the basic information of the user associated with the identification data. The acquisition unit 11 can extract corresponding feature data according to the labels in the basic information. For example, the corresponding feature data can be extracted according to the labels of each item in the resume information to obtain the user's age, birthday, zodiac sign, residential address, gender, major, and interests; the user's department information, position information, and other information can also be directly obtained from the basic information, and the extracted feature data can be matched with the preset label to determine the first feature label that matches it, thereby obtaining the first feature label in a unified format.

[0125] The receiving unit 12 is configured to receive a second feature tag input by the user.

[0126] It should be noted that there can be multiple second feature tags. The second feature tag is the interest and hobby tag entered by the user, and can also be empty. If the user has no hobbies, no data needs to be entered, and the second feature tag is set to empty.

[0127] The first constructing unit 13 is used to construct a user-label weight matrix according to the first feature label and the second feature label.

[0128] Further, see Figure 7 As shown, the first construction unit 13 may include: a collection module 131 , a matching module 132 , a determination module 133 and a generation module 134 .

[0129] The acquisition module 131 is used to acquire current information.

[0130] In this embodiment, the current information can be obtained through the user's client.

[0131] As an example but not limitation, the current information may include at least one of current time data, current address data, current weather data, and current temperature data.

[0132] The matching module 132 is used to match the current information with the first feature tag to obtain matching data.

[0133] In this embodiment, the current information is matched with the first feature tag to determine whether there is data belonging to the current information in the first feature tag. If there is data that can be highlighted in a targeted manner, the information matching the current information is generally selected in the recommended information to improve the real-time nature of the recommendation. For example, if the date in the current information is the user's birthday, that is, the current time data matches the user's birthday tag (the day is the user's birthday), the birthday tag is extracted, and the birthday tag is used as a high-priority tag.

[0134] The determination module 133 is used to determine a tag weight according to the matching data and the second feature tag.

[0135] In this embodiment, if there is matching data, the weight of each label in the first feature label is adjusted according to the preset rules, the weight of the matched first feature label is increased, and the weight of the unmatched first feature label is reduced. The preset rules may include: when the feature labels in the first feature label and the second feature label involve group labels, it is judged whether the weight value of the group label is less than the third weight threshold, if so, the third weight threshold is used as the weight value of the group label, if not, the weight value of the group label is kept unchanged; when the feature labels in the first feature label and the second feature label involve interest and hobby labels, it is judged whether the weight value of the interest and hobby label is less than the third weight threshold, if so, the third weight threshold is used as the weight value of the interest and hobby label, if not, the weight value of the interest and hobby label is kept unchanged.

[0136] As an example but not a limitation, considering that some users have common or similar behavioral hobbies (preferences), interest tags with the same hobbies can be classified as group tags, and it is determined whether the weight value of the group tag is less than the third weight threshold. If so, the weight value of the group tag is increased, and the third weight threshold is used as the weight value of the group tag. If the user's preference types are wide and have low repetition in the group user data, the type of recommended information that the user liked in the past, that is, the recommended tag, is extracted as the user tag, and the weight value of the user tag is increased; if the user provides personal information / personalized settings (interests, hobbies, specialties, etc.), these tags can be used as the most important evaluation data, and the data provided by the user is given priority, and the weight value of this type of tag is increased.

[0137] Further, when the second feature tag is not empty, the determination module 133 may match the second feature tag with the first feature tag. If they do not match, the weight of the second feature tag is set to the first weight threshold according to the total number of user feature tags; if they match, the weight of the second feature tag is increased to the second weight threshold (the second weight threshold is greater than the first weight threshold).

[0138] The generating module 134 is used to generate the user-tag weight matrix based on the tag weights.

[0139] In this embodiment, a user-label weight matrix is ​​constructed according to the weight values ​​of each first feature label, the weight values ​​of the second feature label, and the preset label.

[0140] The preset tags are all tags provided in advance, including the first feature tag and the second feature tag.

[0141] The second construction unit 14 is used to obtain the recommendation tags and corresponding tag weights of each recommendation information from the recommendation library, and generate a recommendation-tag weight matrix.

[0142] The recommendation library includes recommendation information with preset tags and recommendation information without preset tags.

[0143] In this embodiment, the recommendation information is stored in the recommendation library. Each piece of recommendation information is associated with at least one recommendation tag. The recommendation information may be pre-tagged or untagged.

[0144] Further, see Figure 8 As shown, the second construction unit 14 may include: a word segmentation module 141 , an analysis module 142 and a processing module 143 .

[0145] The word segmentation module 141 is used to segment the titles of the recommendation information in the recommendation library, and generate recommendation tags corresponding to the recommendation information according to the segmentation.

[0146] In this embodiment, the title of the recommended information is segmented, keywords are extracted, the keywords are matched with preset tags, and the matched tags are used as recommended tags.

[0147] The analysis module 142 is used to determine the weight of the recommended tag and the weight of the preset tag based on the recommended tag ratio of the recommended information and the preset tag ratio of the recommended information.

[0148] Furthermore, when the recommendation information carries a preset tag, the number of recommended tags in the content of the recommendation information and the number of the preset tags in the content are obtained, and the preset tags are matched with the recommended tags. If they match, the matched recommended tags are extracted, the proportion of each recommended tag in the recommendation information is counted, and the tag weight is set according to the proportion (e.g., the weight of the tag is the proportion of the tag); if they do not match, the proportion of each recommended tag and the preset tag in the recommendation information is counted, and the tag weight is set according to the proportion (e.g., a table associating proportion values ​​with tag weights is provided, each tag weight value corresponds to a proportion interval, and the corresponding weight value is selected according to the proportion of the tag). When the recommendation information does not carry a preset tag (i.e., the proportion of the preset tag is zero), the number of recommended tags in the content of the recommendation information is obtained, the proportion of each recommended tag in the recommendation information is counted, and the corresponding tag weight is set according to the proportion.

[0149] Processing module 143 is used to generate a recommendation-tag weight matrix based on the weight of the recommendation tag of the recommendation information and the weight of the preset tag when the recommendation information carries a preset tag; when the recommendation information does not carry a preset tag, generate a recommendation-tag weight matrix based on the weight of the recommendation tag.

[0150] In another embodiment, the processing module 143 can also extract the recommendation information associated with the recommendation tag (including preset tags) corresponding to the user tag from the recommendation library based on the user tag to form a candidate set, and construct a recommendation-tag weight matrix according to the recommendation tags associated with each piece of recommendation information in the candidate set.

[0151] The third constructing unit 15 is used to construct a user-recommendation association matrix according to the user-tag weight matrix and the recommendation-tag weight matrix.

[0152] Furthermore, the third constructing unit 15 may associate the user tags in the user-tag weight matrix with the recommendation information in the recommendation-tag weight matrix according to the user tags and the recommendation tags, and combine them to generate the user-recommendation association matrix.

[0153] The generating unit 16 is used to calculate the similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix, and generate a recommendation list according to the similarity.

[0154] The recommendation list may adopt an integrated recommendation mode, a multi-row parallel recommendation mode or a pipeline recommendation mode.

[0155] In actual applications, the integrated recommendation mode can be applied to website recommendation column push, the multi-line parallel recommendation mode is mainly applicable to email push that requires filling in information or answering questionnaires, and the assembly line recommendation mode can be applied to public account push.

[0156] In this embodiment, the similarity between the recommendation tag of each piece of promotion information and the user tag is calculated, and a recommendation list consisting of the recommendation information is generated in order of the similarity.

[0157] Furthermore, the generating unit 16 is used to calculate the value rating of the recommendation tag corresponding to each recommendation information in the user-recommendation association matrix; use the value rating as the similarity of the recommendation tag weight; sort the recommendation information corresponding to the recommendation tag weight in order of the similarity from large to small, and generate the recommendation list.

[0158] As an example but not a limitation, take the application of the information recommendation device 1 to the rental portal website to push "admonitions" to internal employees of the enterprise as an example. In the work area of ​​the portal website, the system will intelligently push different admonitions to each employee. The admonitions are stored in the recommendation library, and each admonition has a preset recommendation tag. Considering that the admonitions pushed to each employee are relatively independent, after collecting the employee's employee number, the basic information of the user is obtained according to the employee number, the feature data in the basic information is converted into a first feature tag, and a user-tag weight matrix is ​​constructed according to the first feature tag. A user-recommendation association matrix is ​​constructed according to the user-tag weight matrix and the recommendation-tag weight matrix of the recommendation library, and the similarity between the employee tag and the recommendation tag of each employee is calculated, and the admonition corresponding to the recommendation tag with the greatest similarity is pushed to the corresponding employee.

[0159] The information recommendation device 1 can be used in many general or special computer system environments or configurations. For example: personal computers, server computers, handheld devices or portable devices, tablet devices, multi-processor systems, microprocessor-based systems, set-top boxes, programmable consumer electronic devices, network PCs, minicomputers, mainframe computers, distributed computing environments including any of the above systems or devices, etc. The present application can be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. The present application can also be practiced in distributed computing environments, in which tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules can be located in local and remote computer storage media including storage devices.

[0160] In this embodiment, the acquisition unit 11 of the information recommendation device 1 can acquire the first feature tag according to the user's identification data, and construct a user-tag weight matrix associated with the user's interests and basic information according to the first feature tag and the second feature tag input by the user through the first construction unit 13; considering that one recommended information may correspond to multiple recommended tags, in order to improve the accuracy of the recommendation, the second construction unit 14 is used to obtain the recommended tags and corresponding tag weights of each recommended information from the recommendation library to generate a recommendation-tag weight matrix; the third construction unit 15 is used to construct a user-recommendation association matrix associated with the user tag and the recommended tag according to the user-tag weight matrix and the recommendation-tag weight matrix, and the generation unit 16 calculates the similarity between the user tag and the recommended tag weight in the user-recommendation association matrix to generate a recommendation list that meets the user's interests and basic information, thereby achieving the purpose of accurate recommendation. The information recommendation device 1 can provide multi-scenario and multi-dimensional technical support for pushing information, truly realizing the portal website in various scenarios for thousands of people, and can be applied to the advertising column, employee work area, latest information column and other scenarios of the rental portal website.

[0161] Embodiment 3

[0162] To achieve the above purpose, the present invention also provides a computer device 2, which includes multiple computer devices 2. The components of the information recommendation device 1 of the second embodiment can be dispersed in different computer devices 2. The computer device 2 can be a smart phone, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server or cabinet server (including independent servers, or server clusters composed of multiple servers) that execute programs. The computer device 2 of this embodiment includes at least but is not limited to: a memory 21, a processor 23, a network interface 22 and an information recommendation device 1 (refer to Fig. 9 ). It should be pointed out that Fig. 9 Only a computer device 2 having components is shown, but it should be understood that implementing all of the components shown is not a requirement and more or fewer components may alternatively be implemented.

[0163] In this embodiment, the memory 21 includes at least one type of computer-readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 21 can be an internal storage unit of the computer device 2, such as a hard disk or memory of the computer device 2. In other embodiments, the memory 21 can also be an external storage device of the computer device 2, such as a plug-in hard disk equipped on the computer device 2, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card (Flash Card), etc. Of course, the memory 21 can also include both the internal storage unit of the computer device 2 and its external storage device. In this embodiment, the memory 21 is generally used to store the operating system and various application software installed on the computer device 2, such as the program code of the information recommendation method of embodiment 1, etc. In addition, the memory 21 can also be used to temporarily store various types of data that have been output or are to be output.

[0164] The processor 23 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips in some embodiments. The processor 23 is generally used to control the overall operation of the computer device 2, such as executing control and processing related to data interaction or communication with the computer device 2. In this embodiment, the processor 23 is used to run the program code stored in the memory 21 or process data, such as running the information recommendation device 1.

[0165] The network interface 22 may include a wireless network interface or a wired network interface, and the network interface 22 is generally used to establish a communication connection between the computer device 2 and other computer devices 2. For example, the network interface 22 is used to connect the computer device 2 to an external terminal through a network, and to establish a data transmission channel and a communication connection between the computer device 2 and the external terminal. The network may be a wireless or wired network such as an Intranet, the Internet, the Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), a 4G network, a 5G network, Bluetooth, or Wi-Fi.

[0166] It should be pointed out that Fig. 9 Only computer device 2 is shown with components 21 - 23 , but it should be understood that implementation of all of the components shown is not a requirement, and more or fewer components may alternatively be implemented.

[0167] In this embodiment, the information recommendation device 1 stored in the memory 21 can also be divided into one or more program modules, which are stored in the memory 21 and executed by one or more processors (processor 23 in this embodiment) to complete the present invention.

[0168] Embodiment 4

[0169] To achieve the above purpose, the present invention also provides a computer-readable storage medium, which includes multiple storage media, such as flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, server, App application store, etc., on which a computer program is stored, and the program realizes the corresponding function when executed by the processor 23. The computer-readable storage medium of this embodiment is used to store the information recommendation device 1, and when executed by the processor 23, the information recommendation method of embodiment 1 is realized.

[0170] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0171] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method.

[0172] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. An information recommendation method, characterized in that: include: Collecting identification data of the user, and obtaining a first feature tag of the user according to the identification data; Acquire a second feature tag input by the user; Constructing a user-label weight matrix according to the first feature label and the second feature label; wherein constructing a user-label weight matrix according to the first feature label and the second feature label comprises: Collecting current information, wherein the current information includes at least one of current time data, current address data, current weather data, and current temperature data; Matching the current information with the first feature tag to obtain matching data; wherein the matching is used to determine whether there is data belonging to the current information in the first feature tag, and if so, highlighting the corresponding first feature tag and obtaining the matching data; Determine the label weight according to the matching data and the second feature label; wherein, determine the label weight according to the matching data and the second feature label, including: adjust the weight of each label in the first feature label according to a preset rule, increase the weight of the matched first feature label, and reduce the weight of the unmatched first feature label; the preset rule includes: when the feature label in the first feature label and the second feature label involves a group label, determine whether the weight value of the group label is less than a third weight threshold, if so, use the third weight threshold as the weight value of the group label, if not, keep the weight value of the group label unchanged; when the feature label in the first feature label and the second feature label involves an interest and hobby label, determine whether the weight value of the interest and hobby label is less than the third weight threshold, if so, use the third weight threshold as the weight value of the interest and hobby label, if not, keep the weight value of the interest and hobby label; Generate the user-tag weight matrix based on the tag weights; Obtain the recommended labels and corresponding label weights of each recommended information from the recommendation library, and generate a recommendation-label weight matrix; Constructing a user-recommendation association matrix according to the user-tag weight matrix and the recommendation-tag weight matrix; The similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix is ​​calculated, and a recommendation list is generated according to the similarity.

2. The information recommendation method according to claim 1, characterized in that: The collecting identification data of the user and obtaining the first feature tag of the user according to the identification data includes: Collecting identification data of the user, and obtaining basic information of the user from the human resources system according to the identification data; Extracting feature data from the basic information, and converting the feature data into a first feature tag; The first feature tag includes at least one of a department tag, a position tag, an interest tag, a birthday tag, a zodiac tag, an address tag, a gender tag, and an age tag.

3. The information recommendation method according to claim 1, characterized in that: The recommendation library includes recommendation information with preset tags and recommendation information without preset tags; The step of obtaining the recommendation tags and corresponding tag weights of each recommendation information from the recommendation library and generating a recommendation-tag weight matrix includes: Segmenting the titles of the recommended information in the recommendation library respectively, and generating recommendation tags corresponding to the recommended information according to the segmented words; Determining the weight of the recommended tag and the weight of the preset tag based on the proportion of the recommended tag in the recommended information and the proportion of the preset tag in the recommended information; A recommendation-label weight matrix is ​​generated according to the weights of the recommendation labels of each of the recommendation information and the weights of the preset labels.

4. The information recommendation method according to claim 1, characterized in that: The step of constructing a user-recommendation association matrix according to the user-tag weight matrix and the recommendation-tag weight matrix includes: The user tags in the user-tag weight matrix are associated with the recommendation information in the recommendation-tag weight matrix according to the user tags and the recommendation tags, and the user-recommendation association matrix is ​​generated by combining them.

5. The information recommendation method according to claim 1, characterized in that: The calculating the similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix and generating a recommendation list according to the similarity includes: Calculate the value evaluation of the recommendation label corresponding to each recommendation information in the user-recommendation association matrix; Using the value assessment amount as the similarity of the recommended tag weight; The recommendation information corresponding to the recommendation tag weights is sorted according to the order of the similarities from large to small to generate the recommendation list.

6. The information recommendation method according to claim 1 or 5, characterized in that: The recommendation list adopts an integrated recommendation mode, a multi-row parallel recommendation mode or a pipeline recommendation mode.

7. An information recommendation device, characterized in that: include: an acquisition unit, configured to collect identification data of a user, and acquire a first feature tag of the user according to the identification data; A receiving unit, configured to receive a second feature tag input by the user; A first construction unit is used to construct a user-label weight matrix according to the first feature label and the second feature label; wherein, the construction of the user-label weight matrix according to the first feature label and the second feature label includes: collecting current information, the current information includes at least one of current time data, current address data, current weather data, and current temperature data; matching the current information with the first feature label to obtain matching data; determining label weights according to the matching data and the second feature label; wherein, determining label weights according to the matching data and the second feature label includes: adjusting the weights of each label in the first feature label according to a preset rule, increasing the weight of the matching first feature label; The weight of the tag is determined by the user, and the weight of the unmatched first feature tag is reduced; the preset rules include: when the feature tag in the first feature tag and the second feature tag involves a group tag, judging whether the weight value of the group tag is less than the third weight threshold, if so, using the third weight threshold as the weight value of the group tag, if not, keeping the weight value of the group tag unchanged; when the feature tag in the first feature tag and the second feature tag involves an interest and hobby tag, judging whether the weight value of the interest and hobby tag is less than the third weight threshold, if so, using the third weight threshold as the weight value of the interest and hobby tag, if not, keeping the weight value of the interest and hobby tag; generating the user-tag weight matrix based on the tag weight; The second construction unit is used to obtain the recommendation tags and corresponding tag weights of each recommendation information from the recommendation library, and generate a recommendation-tag weight matrix; A third construction unit, configured to construct a user-recommendation association matrix according to the user-label weight matrix and the recommendation-label weight matrix; The generating unit is used to calculate the similarity between the user tag and the recommendation tag weight in the user-recommendation association matrix, and generate a recommendation list according to the similarity.

8. A computer device, characterized in that: The computer device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the method according to any one of claims 1 to 6 when executing the computer program.

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

Images