Information Recommendation Method, Apparatus, Electronic Device and Storage Medium

Abstract

The present invention provides an information recommendation method, apparatus, electronic device and storage medium, which relate to the field of computer technology. The information recommendation method includes: obtaining the personal data sets of each owner in the community from a property information management platform; generating an owner portrait corresponding to the owner based on the personal data sets; the owner portrait includes at least one model tag, and different model tags are used to predict different types of target information that the owner is concerned about; determining the target recommendation strategies corresponding to different types of target information based on the personal data sets, the owner portrait and the weights of each target recommendation strategy in the recommendation strategy set; and sending each target information to the terminal associated with the owner based on the target recommendation strategy corresponding to each type of target information. Through the above method, it is possible to accurately recommend the target information that each owner in the community is interested in and of high quality.

Description

Technical Field

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

[0002] Currently, community-based information recommendation usually adopts a method starting from "weak acquaintance relationships". For example, by recruiting individuals in the community, a community information group is constructed based on the personal private traffic of the individuals, and the individuals can recommend relevant information in the community information group.

[0003] However, when an individual recommends information in the community information group, it can only be based on the individual's cognition and cannot grasp the preferences of all the property owners in the community, resulting in poor quality of the recommended information and inability to perform accurate information recommendation for each property owner.

[0004] Therefore, how to accurately recommend information to each property owner in the community is an urgent problem to be solved at present. Summary of the Invention

[0005] In view of the problems existing in the prior art, embodiments of the present invention provide an information recommendation method, device, electronic device and storage medium.

[0006] The present invention provides an information recommendation method applied to a server, including:

[0007] Obtaining a personal data set of each property owner in the community from a property information management platform; the personal data set includes personal data of the property owner in at least one dimension;

[0008] Generating an owner portrait corresponding to the property owner based on the personal data set; the owner portrait includes at least one model label, and different model labels are used to predict different types of target information that the property owner is interested in;

[0009] Determining a target recommendation strategy corresponding to different types of target information based on the personal data set, the owner portrait and the weight of each target recommendation strategy in a recommendation strategy set; the weight of each target recommendation strategy is used to reflect the usage rate of the property owner receiving the target information by using the target recommendation strategy;

[0010] Sending each piece of target information to a terminal associated with the property owner based on the target recommendation strategy corresponding to each type of target information.

[0011] Optionally, the generating an owner portrait corresponding to the property owner based on the personal data set includes:

[0012] Using the term frequency-inverse document frequency (TF-IDF) algorithm to determine at least one target keyword in the personal data set;

[0013] Match each target keyword with the keywords corresponding to the model label names in the model label database, and use the matching results to determine the model labels corresponding to each of the target keywords.

[0014] Generate the portrait of the property owner based on each of the model labels.

[0015] Optionally, based on the personal dataset, the portrait of the property owner, and the weights of each target recommendation strategy in the recommendation strategy set, use the TF-IDF algorithm to determine the target recommendation strategies corresponding to different types of target information, including:

[0016] Based on the first preset original weight, the first time interval, and the first weight decay coefficient corresponding to each personal data, use the TF-IDF algorithm to determine the first weight corresponding to each personal data; the first time interval is the time interval from setting the first preset original weight to determining the first weight.

[0017] Based on the second preset original weight, the second time interval, and the second weight decay coefficient corresponding to each target keyword, determine the second weight corresponding to each target keyword; the second time interval is the time interval from setting the second preset original weight to determining the second weight.

[0018] Based on the second weight corresponding to each target keyword, determine the third weight corresponding to each model label.

[0019] Based on each of the first weights, each of the third weights, and the weights of each target recommendation strategy, determine the recommendation strategies corresponding to different types of target information.

[0020] Optionally, after determining the target recommendation strategies corresponding to different types of target information, the method further includes:

[0021] Compare the second weight corresponding to each target keyword with the second weights corresponding to the target keywords in other property owner personal datasets, and determine the similar target keywords of the other property owners based on the comparison results.

[0022] In the portrait of the other property owners, determine the target model labels corresponding to the similar target keywords.

[0023] Send the target information corresponding to the target model labels to the terminals associated with the similar property owners.

[0024] Optionally, after generating the portrait of the property owner corresponding to the property owner, the method further includes:

[0025] Generate target information corresponding to each of the model labels based on each of the model labels.

[0026] Optionally, the dimensions of the personal data in the personal dataset include at least one of the following:

[0027] The personal basic information of the property owner;

[0028] The family relationship information of the property owner;

[0029] The historical service information for the property owner;

[0030] The community access information of the property owner;

[0031] The online browsing information of the property owner.

[0032] The present invention also provides an information recommendation device, which is applied to a server and includes:

[0033] An acquisition module, configured to acquire the personal dataset of each property owner in the community from a property information management platform; the personal dataset includes the personal data of the property owner in at least one dimension;

[0034] A first generation module, configured to generate a property owner portrait corresponding to the property owner based on the personal dataset; the property owner portrait includes at least one model label, and different model labels are used to predict different types of target information that the property owner is concerned about;

[0035] A first determination module, configured to determine a target recommendation strategy corresponding to different types of target information based on the personal dataset, the property owner portrait, and the weight of each target recommendation strategy in a recommendation strategy set; the weight of each target recommendation strategy is used to reflect the usage rate of the property owner receiving target information by using the target recommendation strategy;

[0036] A first sending module, configured to send each piece of the target information to a terminal associated with the property owner based on the target recommendation strategy corresponding to each type of target information.

[0037] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where when the processor executes the program, the information recommendation method as described in any one of the above is implemented.

[0038] The present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the information recommendation method as described in any one of the above is implemented.

[0039] The present invention also provides a computer program product, including a computer program, where when the computer program is executed by a processor, the information recommendation method as described in any one of the above is implemented.

[0040] The information recommendation method, device, electronic device, and storage medium provided by the present invention. Since the property information management platform, as the service provider for the owners, has a natural and close relationship with each owner in the community, detailed personal data sets of each owner in the community can be obtained from the property information management platform. Based on the personal data sets of each owner, a complete and accurate owner portrait can be generated for each owner. For each owner, based on different model tags in the owner portrait, different types of target information that the owner is interested in can be predicted. Then, based on the personal data set, the owner portrait, and the weights of each target recommendation strategy in the recommendation strategy set, a corresponding target recommendation strategy can be matched for each type of target information. Based on the target recommendation strategy corresponding to each type of target information, the target information can be sent to the terminal associated with the owner, and accurate recommendation of high-quality target information to each owner in the community can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] To more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 is one of the flow diagrams of the information recommendation method provided by the present invention;

[0043] Figure 2 is the structural diagram of the owner portrait provided by the present invention;

[0044] Figure 3 is the second flow diagram of the information recommendation method provided by the present invention;

[0045] Figure 4 is the structural diagram of the information recommendation device provided by the present invention;

[0046] Figure 5 is the structural diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] To make the objectives, technical solutions, and advantages of the present invention clearer, the following clearly and completely describes the technical solutions in the present invention with reference to the drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art without creative efforts based on the embodiments in the present invention belong to the scope of protection of the present invention.

[0048] To facilitate a clearer understanding of the embodiments of the present application, some relevant background knowledge is introduced as follows.

[0049] Currently, community-based information recommendation usually adopts a method starting from "light acquaintance relationships". Taking commodity marketing information as an example, by recruiting "stay-at-home moms" or "community store managers" in the community, a community information group is constructed based on personal private domain traffic. The "stay-at-home moms" or "community store managers" post and promote group-buying commodities in the community information group, and community customers place orders through a small program or app to complete the transaction.

[0050] However, the existing information recommendation methods rely too much on the personal relationship chain of the information recommenders and have great limitations. Specifically, it includes the following two aspects:

[0051] 1. Limited information recommendation scope

[0052] Since information recommendation is carried out based on personal "light acquaintance relationships", the personal social circle is the information recommendation scope, which surely cannot cover the entire community.

[0053] 2. Lack of complete owner data support

[0054] Currently, the sources of obtaining owner information are all online, and some information is calculated by algorithms, which has a certain error from the actual situation; when information recommenders recommend information, they can only be based on personal cognition and cannot grasp the preferences of all owners in the community. Moreover, multiple batch recommendations will cause some owners to receive disliked information multiple times, causing disgust and resulting in a poor user experience.

[0055] In summary, in view of the above existing technical problems, in order to comprehensively and accurately recommend information to each owner in the community, the present invention provides an information recommendation method, device, electronic device and storage medium.

[0056] Next, in combination with Figures 1 to 3 The information recommendation method provided by the present invention will be specifically described. Figure 1 is one of the flow diagrams of the information recommendation method provided by the present invention. Refer to Figure 1 As shown, this method is applied to a server and includes steps 101 - step 104, where:

[0057] Step 101: Obtain the personal data set of each owner in the community from the property information management platform; the personal data set includes personal data of the owner in at least one dimension.

[0058] First of all, it should be noted that the execution subject of the present invention can be any server capable of implementing information recommendation.

[0059] When an individual recommends information in a community information group, they can only rely on their personal knowledge and cannot grasp the preferences of all the community owners. As a result, the quality of the recommended information is poor, and it is impossible to provide accurate information recommendations for each owner.

[0060] Therefore, in order to comprehensively and accurately recommend information to each owner in the community, in this embodiment, it is first necessary to obtain the personal data sets of each owner in the community from the property information management platform.

[0061] In practical applications, as the service provider for the owners, the property information management platform provides daily property services for each owner in the community. Due to this dependency relationship, the property information management platform has a natural close relationship with the owners, forming a built-in private traffic pool.

[0062] It can be understood that the property information management platform stores multi-dimensional personal data of each owner. That is to say, the property information management platform stores static information and dynamic information of community owners; among them, the static information includes: housing, property, family information; the dynamic information includes: interaction information between the property and the owners.

[0063] Optionally, each dimensional personal data element in the personal data set of the owner is also called a fact tag of the owner. The dimensions of the personal data in the personal data set include at least one of the following:

[0064] a) The personal basic information of the owner.

[0065] Specifically, the property information management platform can collect the personal basic information of the owner through the community access control server or the owner's purchase contract file.

[0066] Among them, the personal basic information of the owner includes the number of family members, names, mobile phone numbers, etc.

[0067] b) The family relationship information of the owner.

[0068] Specifically, community property personnel can obtain the family relationship information of the owner by visiting the owner and upload it to the property information management platform for storage.

[0069] Among them, the family relationship information of the owner includes the number of family members, the names of each family member, whether there are elderly people, children, etc.

[0070] c) The historical service information for the owner.

[0071] Specifically, property personnel can collect data such as the age, height, body type, gender, education level, marital status, occupation, and age of children of each owner through the historical service information for the owner (such as free medical consultations, free haircuts, children's activities, etc.);

[0072] Alternatively, through the historical reporting and repair service for the property owners, the brand preferences of the owners can be collected. For example, if the air conditioners, washing machines, and refrigerators in the owner's home need to be repaired or cleaned, the property management staff can collect the information of the repaired equipment in the owner's home and upload it to the property information management platform after collection.

[0073] d) The community access information of the said owner.

[0074] Specifically, the community access information of the owner includes: the time when the owner enters and exits the community, whether the owner has a parking space, whether there is a car, the registration of visitor information, the name, mobile phone number, license plate number, etc. of the owner's party.

[0075] e) The online browsing information of the said owner.

[0076] Specifically, the property management platform can collect the online browsing information of different owners through data embedding, so as to obtain the preference degree of different owners for different categories of goods, online habits, purchase frequency, and payment methods.

[0077] After collecting the online browsing information, tags can be manually assigned to the owners. The tags are divided into first-level and second-level tags, and automatic tagging is also supported. Tagging rules are set according to different transaction conditions. For example, if there is a purchase in the past 90 days but no purchase in the past 60 days, the owner will be automatically tagged as a lost customer; if the historical cumulative consumption is 100 times, the owner will be automatically tagged as a potential customer; if the cumulative consumption amount reaches a certain value, the owner will be tagged as a high-value customer. Tags can also be combined with goods. For example, if the owner often buys product A, the classification tag of product A will be assigned.

[0078] Step 102: Generate the owner portrait corresponding to the said owner based on the personal data set; the owner portrait includes at least one model tag, and different model tags are used to predict different types of target information that the owner is concerned about.

[0079] In this embodiment, after obtaining the detailed personal data set of each owner, for each owner, the owner portrait corresponding to the owner can be generated based on the personal data set.

[0080] Among them, the owner portrait is also called the user portrait. As an effective tool for depicting target users and connecting user demands with design directions, the user portrait has been widely used in various fields. As a virtual representative of actual users, the user role formed by the user portrait is not constructed outside the product and the market. The formed user role needs to be representative and can represent the main audience and target group of the product.

[0081] The owner portrait includes at least one model tag. The model tags are, for example, the interest and hobby tags, income level tags, purchasing power tags, public opinion tags, brand tags, etc. of the owner.

[0082] Different model tags are used to predict different types of target information that owners are concerned about. For example, for owner A, taking the hobby tag in the model tags as an example, if the hobby tag includes "sports", it can be predicted that owner A is more concerned about target information of the sports type; among them, the target information of the sports type can be, for example, sports product marketing information, sports popular science knowledge, etc.

[0083] For another example, taking the purchasing power tag in the model tags as an example, if the purchasing power tag includes "luxury goods", it can be predicted that owner A is more concerned about target information of the luxury goods type; among them, the target information of the luxury goods type can be, for example, luxury goods marketing information.

[0084] Figure 2 It is a schematic structural diagram of the owner portrait provided by the present invention. Refer to Figure 2 As shown, the owner portrait of each owner includes raw data, fact tags, and model tags.

[0085] Among them, the raw data (i.e., the personal data set mentioned above) includes: owner basic information, owner historical transaction information, customer browsing record information, etc.; the fact tags include: demographic attributes, usage frequency, number of product purchases, etc.; the model tags include: demographic attributes, user value, hobbies, dressing styles, user activity, user evaluations, etc.

[0086] Step 103: Based on the personal data set, the owner portrait, and the weights of each target recommendation strategy in the recommendation strategy set, determine the target recommendation strategies corresponding to different types of target information; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving the target information by using the target recommendation strategy.

[0087] In this embodiment, the recommendation strategy set includes multiple target recommendation strategies, and different types of target information correspond to different target recommendation strategies; among them, the target recommendation strategy is also called the recommendation channel of the target information.

[0088] For example, for the target information of the "automobile" type, the corresponding target recommendation strategy is: display the target information on the equipment related to the owner's parking space;

[0089] For another example, it is known from the model tag "hobby tag" in the portrait of owner A that owner A likes online shopping, but it is known from the historical order information in the personal data set that owner A has not shopped online recently; then for the target information of the "online shopping" type, the corresponding target recommendation strategy is: display online shopping discount information on the pages of the social APP and online shopping APP on the owner's terminal.

[0090] If property owner A likes online shopping and has been frequently shopping online recently, for the target information of the "online shopping" type, the corresponding target recommendation strategy is: simply display online shopping discount information on the page of the online shopping APP.

[0091] Step 104: Based on the target recommendation strategy corresponding to each type of target information, send each piece of the target information to the terminal associated with the property owner.

[0092] For the information recommendation method provided by the present invention, since the property information management platform, as the service provider for property owners, has a natural and close relationship with each property owner in the community, detailed personal data sets of each property owner in the community can be obtained from the property information management platform. Based on the personal data sets of each property owner, a complete and accurate portrait of each property owner can be generated; for each property owner, based on different model tags in the property owner portrait, different types of target information that the property owner is concerned about can be predicted. Then, based on the personal data set, the property owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set, a corresponding target recommendation strategy can be matched for each type of target information. Based on the target recommendation strategy corresponding to each type of target information, sending each piece of the target information to the terminal associated with the property owner can achieve precise recommendation of high-quality target information to each property owner in the community.

[0093] Optionally, the generation of the property owner portrait corresponding to the property owner based on the personal data set can be specifically implemented through the following steps 1) to 3):

[0094] Step 1): Use the term frequency-inverse document frequency (TF-IDF) algorithm to determine at least one target keyword in the personal data set.

[0095] Step 2): Match each target keyword with the keywords corresponding to the names of each model tag in the model tag database to determine the model tags corresponding to each target keyword.

[0096] Step 3): Generate the property owner portrait based on each model tag.

[0097] In this embodiment, the term frequency (TF)-inverse document frequency (IDF) is used to determine the keywords preferred by the property owner in the personal data set.

[0098] Among them, the core idea expressed by TF is that the words that appear repeatedly in a piece of text are more important. The idea of IDF is that the words that appear in all texts are not important, and IDF is used to correct the calculation results represented by TF.

[0099] For example, for homeowner A, online shopping browsing records are retrieved from the personal dataset. Table 1 shows the online shopping browsing records of homeowner A.

[0100] Table 1

[0101]

[0102]

[0103] TF = Number of occurrences of a word in the text / Total number of words in the text; IDF = Log(Total number of texts / (Number of texts in which the word appears + 1)); TF-IDF = TF * IDF.

[0104] Using the TF-IDF algorithm, the target keywords for homeowner A are determined to be "black" and "short sleeves"; then, "black" and "windbreaker" are matched with the keywords corresponding to each model label name in the model label database. Table 2 shows the keywords corresponding to each model label name in the model label database.

[0105] Table 2

[0106] Model label name Keywords Hobbies Table tennis, black, cat Dressing style Trench coat, short sleeves, windbreaker

[0107] As can be seen from Table 2 above, the model label corresponding to the target keyword "black" is "hobbies"; the model label corresponding to the target keyword "windbreaker" is "dressing style".

[0108] Based on each model label, a homeowner portrait of homeowner A can be generated.

[0109] In the above embodiment, by matching each target keyword in the homeowner's personal dataset with the keywords corresponding to each model label name in the model label database, the model label corresponding to each target keyword can be determined; based on each model label, a homeowner portrait can be accurately generated based on the homeowner's preferences.

[0110] Optionally, after generating the homeowner portrait corresponding to the homeowner, it is also necessary to generate target information corresponding to each of the model labels based on each of the model labels.

[0111] For example, taking the model label "demographic attributes" including "elderly" and "children" as an example, the target information can be product promotion activities for nutritional products such as milk and oats. The product promotion activities for nutritional products specifically include activity time, activity location, activity target, activity products, etc.

[0112] Taking the model label "hobby label" including "sports" as an example, the target information can be product promotion activities for sports equipment, community sports competitions, etc.

[0113] In the above embodiments, based on each model label in the owner portrait, target information corresponding to each model label is generated, and personalized information recommendation can be comprehensively and accurately made for each owner in the community.

[0114] Optionally, based on the personal data set, the owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set, the target recommendation strategy corresponding to different types of target information is determined, which is specifically implemented through the following steps [1] to [4]:

[0115] Step [1], based on the first preset original weight, the first time interval, and the first weight decay coefficient corresponding to each personal data, use the TF-IDF algorithm to determine the first weight corresponding to each personal data; the first time interval is the time interval from setting the first preset original weight to determining the first weight;

[0116] Step [2], based on the second preset original weight, the second time interval, and the second weight decay coefficient corresponding to each target keyword, use the TF-IDF algorithm to determine the second weight corresponding to each target keyword; the second time interval is the time interval from setting the second preset original weight to determining the second weight;

[0117] Step [3], based on the second weight corresponding to each target keyword, determine the third weight corresponding to each model label;

[0118] Step [4], based on each of the first weights, each of the third weights, and the weights of each target recommendation strategy, determine the recommendation strategy corresponding to different types of target information.

[0119] In this embodiment, for each owner, in order to match the most suitable target recommendation strategy for different types of target information, it is first necessary to calculate the first weight corresponding to each personal data in the owner's personal data set and the third weight corresponding to each model label in the owner portrait.

[0120] [a], The calculation method of the first weight is as follows:

[0121] Based on the first preset original weight, the first time interval, and the first weight decay coefficient corresponding to each personal data, use the TF-IDF algorithm to determine the first weight corresponding to each personal data. Specifically, it can be represented by formula (1), and formula (1) is as follows:

[0122] First weight = (behavior weight * first weight decay coefficient) * (TF-IDF * number of behavior times)

[0123] Among them, the behavior weight refers to the same behavior of the owner. Due to the different degrees of the behavior, the weights are different. For example, if the user has generated an order for a certain commodity, different weights are set according to the three order statuses of the order not paid, paid but not refunded, and paid and refunded.

[0124] The behavior weight is obtained based on the first preset original weight, the first interval time, and the first weight decay coefficient, and is specifically represented by formula (2). Formula (2) is as follows:

[0125] Behavior weight = First preset original weight * exp(-First weight decay coefficient * Interval time)

[0126] Among them, the first weight decay coefficient, also known as the time decay factor, reflects the process of the heat of the first preset original weight gradually cooling over time. The symbol exp is an exponential function with the natural constant e as the base.

[0127] For example, set the first preset original weight to 1, and the behavior weight after 10 days to 0.2, that is, the weight decays to 0.2 after 9 days. Substitute the known variables into formula (2), and the first weight decay coefficient is obtained through exponential operations.

[0128] [b] The calculation method of the second weight is similar to that of the first weight, and will not be elaborated here.

[0129] After calculating the second weights corresponding to each target keyword, it is necessary to perform a weighted sum of the second weights corresponding to each target keyword to obtain the third weights corresponding to each model label.

[0130] For example, to calculate the third weight of the model label "Hobbies", it is necessary to perform a weighted sum of the second weight of the target keyword "Black" and the second weight of "Sports" in "Hobbies" to obtain the third weight of "Hobbies".

[0131] Finally, based on the first weights, the third weights, and the weights of each target recommendation strategy, use formula (3) to determine the recommendation strategies corresponding to different types of target information. Formula (3) is expressed as follows:

[0132]

[0133] Among them, s represents the matching degree between the model label and the target recommendation strategy; w i represents the weight of the i-th target recommendation strategy; f i represents the first weight corresponding to the i-th personal data and the third weight of the i-th model label.

[0134] It should be noted that f iIt reflects the value of the property owner in a certain activity, commodity / service. The higher the value, the higher the user conversion rate and the greater the value of promotion.

[0135] w i It can be calculated through the following formula (4):

[0136] w i = c * weight decay coefficient (4)

[0137] Among them, c represents the conversion rate of the target recommendation strategy within a certain time period. The conversion rate = the number of paying users within a certain time period / the number of visitors; the weight decay coefficient is the first weight decay coefficient and the second weight decay coefficient mentioned above.

[0138] After calculating the matching degree between the model label and the target recommendation strategy, the target recommendation strategy with the matching degree reaching the preset threshold in the recommendation strategy set is used as the target recommendation strategy corresponding to the model label.

[0139] It should be noted that the matching degree between each model label of each property owner portrait and the same target recommendation strategy may be different. For example, the model label "elderly group" has a higher matching degree with the offline target recommendation strategy; for the model label "car owners", the target recommendation strategy is: display relevant activities on the electronic devices at the garage entrance or the community entrance.

[0140] In the above implementation manner, for each property owner, based on different model labels in the property owner portrait, different types of target information that the property owner is concerned about can be predicted. Then, based on the personal data set, the property owner portrait, and the weights of each target recommendation strategy in the recommendation strategy set, the corresponding target recommendation strategy can be matched for each type of target information. Based on the target recommendation strategy corresponding to each type of target information, each target information is sent to the terminal associated with the property owner, and accurate recommendation of high-quality target information to each property owner in the community can be realized.

[0141] Optionally, after determining the target recommendation strategy corresponding to different types of target information, the target information can also be sent to the terminals associated with other property owners, which is specifically implemented through the following steps (a)-(c):

[0142] Step (a): Compare the second weights corresponding to each of the target keywords with the second weights corresponding to each of the target keywords in the personal data set of other property owners, and determine the similar target keywords of the other property owners based on the comparison results;

[0143] Step (b): In the property owner portrait of the other property owner, determine the target model label corresponding to the similar target keywords;

[0144] Step (c): Send the target information corresponding to the target model label to the terminal associated with the similar property owner.

[0145] For example, based on Table 1, it is known that the online shopping browsing records of property owner A. Property owner A has 3 browsing records, with a total of 17 words after word segmentation. Assume: the total number of users on the online shopping platform = 10,000 people, and there are 500 users whose product titles browsed by property owner A contain the word "black". The base is 2. With a base of 2, the product titles browsed by users are used as the word library after word segmentation and summarization, the total number of users on the platform is the total number of texts, and the number of texts with this word is the number of users with the same browsing behavior. The term frequency TF = 0.12, and the inverse document frequency index = 4.32. The weight of the label "black" for users (i.e., the second weight) TF-IDF can be calculated as 0.52.

[0146] For the target keyword "black", the second weight of property owner A for "black" is 0.52, the second weight of property owner B for "black" is 0.7, and the second weight of property owner C for "black" is 0.4.

[0147] Then, for property owner A, determine property owner B whose second weight is greater than or equal to 0.5 as the similar property owner, determine the model label corresponding to "black" as "hobby", and then send the target information corresponding to property owner A's "hobby" to the terminal associated with property owner B.

[0148] In the above embodiment, by sending the target information corresponding to the target model label to the terminal associated with the similar property owner, the recommended range of the target information can be further expanded.

[0149] Figure 3 It is the second flow diagram of the information recommendation method provided by the present invention. Refer to Figure 3 As shown, the method includes Step 301 - Step 312, where:

[0150] Step 301: Obtain the personal data set of each property owner in the community from the property information management platform; wherein, the personal data set includes personal data of the property owner in at least one dimension.

[0151] Specifically, the dimensions of the personal data in the personal data set include at least one of the following: 1) the personal basic information of the property owner; 2) the family relationship information of the property owner; 3) the historical service information for the property owner; 4) the community access information of the property owner; 5) the online browsing information of the property owner.

[0152] Step 302: Use the TF-IDF algorithm to determine at least one target keyword in the personal data set.

[0153] Step 303: Match each target keyword with the keywords corresponding to each model label name in the model label database to determine the model label corresponding to each target keyword, where different model labels are used to predict different types of target information that the property owner is concerned about.

[0154] Step 304: Generate a property owner portrait based on each model label.

[0155] Step 305: Generate target information corresponding to each model label based on the model labels in the property owner portrait.

[0156] Step 306: Determine the first weight corresponding to each personal data using the TF-IDF algorithm based on the first preset original weight, the first interval time, and the first weight decay coefficient corresponding to each personal data.

[0157] Specifically, the first interval time is the interval time from setting the first preset original weight to determining the first weight.

[0158] Step 307: Determine the second weight corresponding to each target keyword using the TF-IDF algorithm based on the second preset original weight, the second interval time, and the second weight decay coefficient corresponding to each target keyword.

[0159] Specifically, the second interval time is the interval time from setting the second preset original weight to determining the second weight.

[0160] Step 308: Determine the third weight corresponding to each model label based on the second weight corresponding to each target keyword.

[0161] Step 309: Determine the recommendation strategy corresponding to different types of target information based on the first weights, the third weights, and the weights of each target recommendation strategy.

[0162] Step 310: Send each target information to the terminal associated with the property owner based on the target recommendation strategy corresponding to each type of target information.

[0163] Step 311: Compare the second weights corresponding to each target keyword with the second weights corresponding to each target keyword in other property owner personal data sets, and determine the similar target keywords of other property owners based on the comparison results; in the property owner portraits of other property owners, determine the target model labels corresponding to the similar target keywords.

[0164] Step 312: Send the target information corresponding to the target model label to the terminal associated with the similar property owner.

[0165] It should be noted that the execution order of Step 310 and Steps 311 - 312 is not sequential.

[0166] Taking the marketing of each property owner in the community as an example, the information recommendation method provided by the present invention will be further described.

[0167] Step 1: The property information management platform enters the basic information of the property owners through access control, convenience services, etc.

[0168] Step 2: The property management staff registers services such as property owners' reporting of problems, repair requests, cleaning, etc., and further obtains relevant information about the property owners.

[0169] Step 3: The property information management platform sets up points for the browsing and purchasing behaviors of the property owners.

[0170] Step 4: Combining the data stored in the property information management platform, fact tags are given to the property owners.

[0171] Step 5: Based on the fact tags and basic information of the property owners, through TF-IDF modeling, a property owner portrait is formed.

[0172] Step 6: Different types of activities are created according to the property owner portrait, products, event events, etc.

[0173] Step 7: The marketing channels (i.e., the target information recommendation strategies mentioned above) classify groups according to audience characteristics, age, etc.

[0174] Step 8: The matching degree between the property owner portrait in the activity and the marketing channels is calculated, so as to achieve precision marketing.

[0175] The information recommendation device provided by the present invention will be described below. The information recommendation device described below can be mutually corresponding and referred to the information recommendation method described above. Figure 4 It is a schematic structural diagram of the information recommendation device provided by the present invention. As Figure 4 shown, the information recommendation device 400 includes: an acquisition module 401, a first generation module 402, a first determination module 403, and a first sending module 404, where:

[0176] The acquisition module 401 is used to obtain the personal data set of each property owner in the community from the property information management platform; the personal data set includes personal data of the property owner in at least one dimension;

[0177] The first generation module 402 is used to generate the corresponding property owner portrait based on the personal data set; the property owner portrait includes at least one model tag, and different model tags are used to predict different types of target information that the property owner is concerned about;

[0178] The first determination module 403 is configured to determine a target recommendation strategy corresponding to different types of target information based on the personal data set, the owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving the target information by using the target recommendation strategy.

[0179] The first sending module 404 is configured to send each of the target information to a terminal associated with the owner based on the target recommendation strategy corresponding to each type of target information.

[0180] In the information recommendation device provided by the present invention, since the property information management platform is the service provider for the owners and has a natural close relationship with each owner in the community, detailed personal data sets of each owner in the community can be obtained from the property information management platform. Based on the personal data sets of each owner, a complete and accurate owner portrait can be generated for each owner; for each owner, based on different model tags in the owner portrait, different types of target information that the owner is concerned about can be predicted. Then, based on the personal data set, the owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set, a corresponding target recommendation strategy can be matched for each type of target information. Based on the target recommendation strategy corresponding to each type of target information, each target information is sent to a terminal associated with the owner, and accurate recommendation of high-quality target information to each owner in the community can be realized.

[0181] Optionally, the generation module 402 is further configured to:

[0182] Use the term frequency-inverse document frequency (TF-IDF) algorithm to determine at least one target keyword in the personal data set;

[0183] Match each target keyword with the keywords corresponding to the names of each model tag in the model tag database to determine the model tags corresponding to each target keyword;

[0184] Generate the owner portrait based on each of the model tags.

[0185] Optionally, the determination module 403 is further configured to:

[0186] Determine the first weight corresponding to each personal data based on the first preset original weight, the first interval time, and the first weight decay coefficient corresponding to each personal data; the first interval time is the interval time from setting the first preset original weight to determining the first weight;

[0187] Determine the second weight corresponding to each of the target keywords based on the second preset original weight, the second interval time, and the second weight decay coefficient corresponding to each target keyword; the second interval time is the interval time from setting the second preset original weight to determining the second weight;

[0188] Determine the third weight corresponding to each of the model tags based on the second weight corresponding to each target keyword;

[0189] Determine the recommendation strategies corresponding to different types of target information based on the first weights, the third weights, and the weights of the target recommendation strategies.

[0190] Optionally, the device further includes:

[0191] A comparison module, configured to compare the second weights corresponding to each of the target keywords with the second weights corresponding to each target keyword in other owners' personal datasets, and determine the similar target keywords of the other owners based on the comparison result;

[0192] A second determination module, configured to determine the target model tags corresponding to the similar target keywords in the owner portrait of the other owners;

[0193] A second sending module, configured to send the target information corresponding to the target model tags to the terminals associated with the similar owners.

[0194] Optionally, the device further includes:

[0195] A second generation module, configured to generate target information corresponding to each of the model tags based on each of the model tags.

[0196] Optionally, the dimensions of the personal data in the personal dataset include at least one of the following:

[0197] The personal basic information of the owner;

[0198] The family relationship information of the owner;

[0199] The historical service information for the owner;

[0200] The community access information of the owner;

[0201] The online browsing information of the owner.

[0202] Figure 5 It is a schematic structural diagram of the electronic device provided by the present invention, as Figure 5As shown in the figure, the electronic device may include: a processor 510, a communications interface 520, a memory 530, and a communication bus 540. Among them, the processor 510, the communications interface 520, and the memory 530 complete communication with each other through the communication bus 540. The processor 510 may call the logical instructions in the memory 530 to execute an information recommendation method, which includes: obtaining a personal data set of each owner in the community from a property information management platform; the personal data set includes personal data of the owner in at least one dimension; based on the personal data set, generating an owner portrait corresponding to the owner; the owner portrait includes at least one model tag, and different model tags are used to predict different types of target information that the owner is concerned about; based on the personal data set, the owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set, determining the target recommendation strategy corresponding to different types of target information; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving the target information by using the target recommendation strategy; based on the target recommendation strategy corresponding to each type of target information, sending each target information to a terminal associated with the owner.

[0203] In addition, when the logical instructions in the above-mentioned memory 530 can be implemented in the form of software functional units and sold or used as an independent product, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media such as a USB flash drive, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk, or an optical disc that can store program codes.

[0204] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the information recommendation method provided by each of the above methods. The method includes: obtaining a personal data set of each owner in the community from a property information management platform; the personal data set includes personal data of the owner in at least one dimension; based on the personal data set, generating an owner portrait corresponding to the owner; the owner portrait includes at least one model label, and different model labels are used to predict different types of target information that the owner is concerned about; based on the personal data set, the owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set, determining a target recommendation strategy corresponding to different types of target information; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving target information by using the target recommendation strategy; based on the target recommendation strategy corresponding to each type of target information, sending each target information to a terminal associated with the owner.

[0205] In another aspect, the present invention also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it is implemented to execute the information recommendation method provided by each of the above methods. The method includes: obtaining a personal data set of each owner in the community from a property information management platform; the personal data set includes personal data of the owner in at least one dimension; based on the personal data set, generating an owner portrait corresponding to the owner; the owner portrait includes at least one model label, and different model labels are used to predict different types of target information that the owner is concerned about; based on the personal data set, the owner portrait, and the weight of each target recommendation strategy in the recommendation strategy set, determining a target recommendation strategy corresponding to different types of target information; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving target information by using the target recommendation strategy; based on the target recommendation strategy corresponding to each type of target information, sending each target information to a terminal associated with the owner.

[0206] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative efforts.

[0207] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0208] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An information recommendation method, characterized in that, Applied to a server, including: Obtain the personal data set of each owner in the community from the property information management platform; the personal data set includes the personal data of the owner in at least one dimension; Based on the personal data set, generate an owner portrait corresponding to the owner; the owner portrait includes at least one model tag, and different model tags are used to predict different types of target information that the owner is concerned about; Based on the personal data set, the owner portrait, and the weights of each target recommendation strategy in the recommendation strategy set, determine the target recommendation strategies corresponding to different types of target information; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving target information using the target recommendation strategy; the determining the target recommendation strategies corresponding to different types of target information based on the personal data set, the owner portrait, and the weights of each target recommendation strategy in the recommendation strategy set includes: based on the first preset original weight, the first interval time, and the first weight decay coefficient corresponding to each personal data, use the TF-IDF algorithm to determine the first weight corresponding to each personal data; the first interval time is the interval time from setting the first preset original weight to determining the first weight; based on the second preset original weight, the second interval time, and the second weight decay coefficient corresponding to each target keyword, use the TF-IDF algorithm to determine the second weight corresponding to each target keyword; the second interval time is the interval time from setting the second preset original weight to determining the second weight; based on the second weight corresponding to each target keyword, determine the third weight corresponding to each model tag; based on each first weight, each third weight, and the weights of each target recommendation strategy, determine the target recommendation strategies corresponding to different types of target information; Based on the target recommendation strategies corresponding to each type of target information, send each target information to the terminal associated with the owner.

2. The information recommendation method according to claim 1, wherein The generating an owner portrait corresponding to the owner based on the personal data set includes: Use the term frequency-inverse document frequency TF-IDF algorithm to determine at least one target keyword in the personal data set; Match each target keyword with the keywords corresponding to each model tag name in the model tag database to determine the model tag corresponding to each target keyword; Based on each model tag, generate the owner portrait.

3. The information recommendation method according to claim 1, wherein After determining the target recommendation strategies corresponding to different types of target information, the method further includes: Compare the second weights corresponding to each target keyword with the second weights corresponding to each target keyword in the personal data sets of other owners, and determine the similar target keywords of the other owners based on the comparison result; In the owner portraits of the other owners, determine the target model tags corresponding to the similar target keywords; Send the target information corresponding to the target model tag to the terminal associated with the similar owner.

4. The information recommendation method according to claim 1, wherein After generating the owner portrait corresponding to the owner, the method further includes: Generate target information corresponding to each of the model tags based on each of the model tags.

5. The information recommendation method according to claim 1, wherein The dimensions of the personal data in the personal dataset include at least one of the following: The personal basic information of the owner; The family relationship information of the owner; The historical service information for the owner; The community access information of the owner; The online browsing information of the owner.

6. An information recommendation device, characterized in that, Applied to a server, including: An acquisition module for acquiring a personal dataset of each owner in the community from a property information management platform; the personal dataset includes personal data of at least one dimension of the owner; A first generation module for generating an owner profile corresponding to the owner based on the personal dataset; the owner profile includes at least one model tag, and different model tags are used to predict different types of target information that the owner is concerned about; A first determination module for determining a target recommendation strategy corresponding to different types of target information based on the personal dataset, the owner profile, and the weight of each target recommendation strategy in the recommendation strategy set; the weight of each target recommendation strategy is used to reflect the usage rate of the owner receiving target information using the target recommendation strategy; the determining a target recommendation strategy corresponding to different types of target information based on the personal dataset, the owner profile, and the weight of each target recommendation strategy in the recommendation strategy set includes: determining the first weight corresponding to each personal data using the TF-IDF algorithm based on the first preset original weight, the first interval time, and the first weight decay coefficient corresponding to each personal data; the first interval time is the interval time from setting the first preset original weight to determining the first weight; determining the second weight corresponding to each target keyword using the TF-IDF algorithm based on the second preset original weight, the second interval time, and the second weight decay coefficient corresponding to each target keyword; the second interval time is the interval time from setting the second preset original weight to determining the second weight; determining the third weight corresponding to each model tag based on the second weight corresponding to each target keyword; determining a target recommendation strategy corresponding to different types of target information based on each first weight, each third weight, and the weight of each target recommendation strategy; A first sending module for sending each target information to a terminal associated with the owner based on the target recommendation strategy corresponding to each type of target information.

7. An electronic device, comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the information recommendation method according to any one of claims 1 to 5.

8. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the information recommendation method according to any one of claims 1 to 5.

9. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the information recommendation method according to any one of claims 1 to 5.

Images