House resource recommendation method and device based on multi-dimensional behavior data

Abstract

The invention provides a house resource recommendation method and device based on multi-dimensional behavior data, and relates to the technical field of data processing, and the method comprises the steps: obtaining user behavior data of a plurality of data sources; inputting the user behavior data of the plurality of data sources into a pre-constructed strategy model to obtain a first recommendation list, and performing calculation through a pre-constructed similar model based on the user behavior data of the plurality of data sources to obtain a second recommendation list; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; the similarity model is used for calculating the similarity between the users; and performing fusion processing on the first recommendation list and the second recommendation list to obtain a target recommendation list. Through the method provided by the invention, more accurate and personalized recommendation can be realized, brokers can be helped to better understand customer demands, and house source options better conforming to customer expectations are provided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and in particular to a house source recommendation method and device based on multi-dimensional behavior data. BACKGROUND

[0002] In the current real estate market environment, the supply of house sources is significantly greater than the demand of customers, which leads to the market showing the characteristics of "more houses and fewer customers". At the same time, with the improvement of the transportation network and the change of people's concept of buying a house, more and more customers begin to accept cross-regional house purchase, thereby gradually increasing the proportion of cross-regional transactions. This trend not only increases the range of customers' house selection, but also brings new challenges to real estate brokers. Under this background, the traditional house customer matching method seems to be somewhat inadequate. In the past, many matching models mainly relied on the behavior data of customers online browsing, and this method could capture part of the potential demand, but in the era of information explosion, the browsing record alone was not enough to fully understand the real preferences of customers. In addition, since the data of old product models of new houses and second-hand houses are isolated from each other, this further limits the matching accuracy of the model.

[0003] How to realize more accurate and personalized house source recommendation and help brokers better understand customer demand is a technical problem to be solved at present. SUMMARY

[0004] The present application provides a house source recommendation method and device based on multi-dimensional behavior data to solve the defects in the prior art.

[0005] The present application provides a house source recommendation method based on multi-dimensional behavior data, comprising the following steps: Obtaining user behavior data of a plurality of data sources; Inputting the user behavior data of the plurality of data sources into a pre-constructed strategy model to obtain a first recommendation list, and calculating a second recommendation list based on the user behavior data of the plurality of data sources through a pre-constructed similarity model; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users; Fusing the first recommendation list and the second recommendation list to obtain a target recommendation list.

[0006] According to the house source recommendation method based on multi-dimensional behavior data provided by the present application, after obtaining the user behavior data of the plurality of data sources, the method further comprises: Data preprocessing is performed on the user behavior data of the plurality of data sources; The data preprocessing comprises data cleaning, data standardization and feature extraction.

[0007] The application provides a house source recommendation method based on multi-dimensional behavior data. Obtain historical user behavior data of multiple data sources. Extract behavior characteristics from the historical user behavior data of the multiple data sources to obtain historical user behavior characteristics. Train a reinforcement learning model based on the historical user behavior characteristics to obtain an initial strategy model. Adjust and optimize the initial strategy model according to user feedback to obtain the strategy model, wherein the user feedback is obtained based on user feedback on a result generated by the initial strategy model.

[0008] The application provides a house source recommendation method based on multi-dimensional behavior data. Calculate the similarity between users based on a pre-constructed user house source scoring matrix, and determine target similar users based on the similarity between the users, wherein the target similar users are users with the highest similarity to the current user, and the user house source scoring matrix is constructed based on historical user behavior data of the multiple data sources and is used to record the scores of users on house sources. Obtain the second recommendation list according to the recommendation list of the target similar users.

[0009] The application provides a house source recommendation method based on multi-dimensional behavior data. Obtain user feedback on the second list. Adjust and optimize the similar model based on the user feedback on the second list.

[0010] The application provides a house source recommendation method based on multi-dimensional behavior data. After the first recommendation list and the second recommendation list are fused to obtain the target recommendation list, the method further comprises: Display the target recommendation list and the recommendation reasons of each target recommended house source in the at least one target recommended house source to a user interface.

[0011] The application further provides a house source recommendation device based on multi-dimensional behavior data, comprising the following modules. An obtaining module is configured to obtain user behavior data of multiple data sources. a recommendation module configured to input the user behavior data of the multiple data sources into a pre-constructed strategy model to obtain a first recommendation list, and to calculate a second recommendation list based on the user behavior data of the multiple data sources by using a pre-constructed similarity model, wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data, and the similarity model is configured to calculate the similarity between users; a fusion module configured to fuse the first recommendation list and the second recommendation list to obtain a target recommendation list.

[0012] The application further provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for recommending a house source based on multi-dimensional behavior data according to any one of the above embodiments when executing the computer program.

[0013] The application further provides a non-transitory computer-readable storage medium having a computer program stored thereon, wherein the computer program is executable on a processor to implement the method for recommending a house source based on multi-dimensional behavior data according to any one of the above embodiments.

[0014] The application further provides a computer program product comprising a computer program, wherein the computer program is executable on a processor to implement the method for recommending a house source based on multi-dimensional behavior data according to any one of the above embodiments.

[0015] The application provides a method and device for recommending a house source based on multi-dimensional behavior data, which comprises the following steps: obtaining user behavior data of multiple data sources; inputting the user behavior data of the multiple data sources into a pre-constructed strategy model to obtain a first recommendation list, and calculating a second recommendation list based on the user behavior data of the multiple data sources by using a pre-constructed similarity model, wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data, and the similarity model is configured to calculate the similarity between users; and fusing the first recommendation list and the second recommendation list to obtain a target recommendation list. Therefore, the application can realize more accurate and personalized recommendation by constructing an intelligent matching system for multi-source data fusion, which helps real estate agents better understand the needs of customers and provides more suitable house source options for customers. BRIEF DESCRIPTION OF DRAWINGS

[0016] In order to more clearly illustrate the technical solutions of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0017] Figure 1 4 is a flow chart of the housing recommendation method based on multi-dimensional behavioral data provided by the present invention.

[0018] Figure 2 2. It is a data processing flow diagram of the housing recommendation method based on multi-dimensional behavioral data provided by the present invention.

[0019] Figure 3 Schematic diagram of the model architecture of the housing recommendation method based on multi-dimensional behavioral data provided by the present invention.

[0020] Figure 4 This is a product flow diagram of the housing recommendation method based on multi-dimensional behavioral data provided by the present invention.

[0021] Figure 5 This is a system architecture diagram of the housing recommendation method based on multi-dimensional behavioral data provided by the present invention.

[0022] Figure 6 2 is a schematic diagram of the structure of the housing recommendation device based on multi-dimensional behavioral data provided by the present invention.

[0023] Figure 7 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0024] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0025] The following combination Figures 1-7 The present invention describes a housing recommendation method and device based on multi-dimensional behavioral data.

[0026] Figure 1 is a flow chart of the housing recommendation method based on multi-dimensional behavior data provided by the present invention, such as Figure 1 As shown, the method includes the following: Step 100: Obtain user behavior data from multiple data sources.

[0027] It should be noted that the embodiments of the present invention are applicable to various real estate information service platforms, such as online rental platforms, second-hand housing transaction platforms, etc. By providing accurate housing recommendations, these platforms can better meet user needs and improve user satisfaction and loyalty.

[0028] Figure 2is a data processing flow schematic diagram of a house source recommendation method based on multi-dimensional behavior data provided by the application, as shown Figure 2 First, the data source is defined, which can specifically include: browsing behavior, recording the user's online browsing behavior of real estate information; attention action, tracking the user's attention to specific house sources; business opportunity information, collecting information about the user's intention to buy a house; and viewing records, recording the user's on-site house viewing. By integrating user behavior data from multiple data sources, the user's needs and preferences can be better understood, thereby improving the accuracy of the recommendations.

[0029] Further, after step 100 obtains user behavior data from multiple data sources, the method further comprises: data preprocessing of the user behavior data from the multiple data sources; The data preprocessing includes data cleaning, data standardization and feature extraction.

[0030] Specifically, data cleaning is used to remove invalid customer data and outliers; data standardization is used to unify the data format; and feature extraction is used to extract useful features from the original data.

[0031] Step 200, inputting the user behavior data from the multiple data sources into a pre-constructed strategy model to obtain a first recommendation list, and calculating a second recommendation list based on the user behavior data from the multiple data sources through a pre-constructed similarity model; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users.

[0032] It should be noted that the strategy model is a reinforcement learning (Reinforcement Learning, RL) model that learns how to recommend house sources to maximize user satisfaction or a certain business goal (such as click-through rate, conversion rate, etc.) by analyzing historical user behavior data. The strategy model is mainly used to dynamically adjust the recommendation strategy according to the actual behavior and preferences of the user. It can adjust the recommendation list according to market changes and personal preferences to make the recommendation more targeted. After inputting the user behavior data, the strategy model generates a preliminary recommended house source list, i.e. the first recommended list. The strategy model uses reinforcement learning to continuously optimize the recommendation strategy based on user feedback, achieving dynamic adjustment and personalized recommendation.

[0033] A similarity model is used to calculate the similarity between users, possibly based on multiple dimensions such as user interests, behavior patterns, preferences, etc. By finding other users similar to the target user and analyzing the behavior data of these similar users, the similarity model can generate another list of recommended housing sources, i.e. a second recommended list. The similarity model based on collaborative filtering focuses on discovering similarities between users, thereby recommending housing sources that similar users like. This method can discover the potential interests of users and provide novel recommendations.

[0034] Step 300, fusing the first recommended list and the second recommended list to obtain a target recommended list.

[0035] Specifically, the first recommended list and the second recommended list are fused, for example, using weighted averaging, cross-validation, ranking fusion, etc. to obtain the final recommended result, i.e. the target recommended list. Fusing the recommended results of the strategy model and the similarity model can fully utilize the advantages of both and further improve the diversity and accuracy of the recommendations.

[0036] The above is a step-by-step description of the housing source recommendation method based on multi-dimensional behavior data provided by the present application. From the above description of the steps, it can be seen that according to the housing source recommendation method based on multi-dimensional behavior data provided by the present application, user behavior data from multiple data sources is obtained; the user behavior data from the multiple data sources is input into a pre-constructed strategy model to obtain a first recommended list, and a second recommended list is obtained by calculating based on the user behavior data from the multiple data sources through a pre-constructed similarity model; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; the similarity model is used to calculate the similarity between users; the first recommended list and the second recommended list are fused to obtain a target recommended list. Therefore, by constructing an intelligent matching system for multi-source data fusion, the present application can achieve more accurate and personalized recommendations, helping brokers better understand customer needs and providing more customer-expected housing options.

[0037] Based on the above embodiment, in the present embodiment, the construction process of the strategy model comprises: Step 210, obtaining historical user behavior data from multiple data sources.

[0038] Step 220, extracting behavior features from the historical user behavior data of the multiple data sources to obtain historical user behavior features.

[0039] Step 230, training a reinforcement learning model based on the historical user behavior features to obtain an initial strategy model.

[0040] Step 240, adjusting and optimizing the initial strategy model according to user feedback results to obtain the strategy model; wherein the user feedback results are obtained based on user feedback on the results generated by the initial strategy model.

[0041] Figure 3 is the model architecture schematic diagram of the house source recommendation method based on multi-dimensional behavior data provided by the application, combined with Figure 3 The construction process of the strategy model is described in detail.

[0042] Specifically, extracting historical user behavior features includes: extracting user browsing habits (such as frequently browsed areas, time periods) from browsing behavior. Extracting user interest preferences (such as house types, area ranges) from attention actions. Extracting user purchase intentions (such as budget, housing purchase purposes) from business information. Extracting user on-site viewing conditions and feedback (such as viewed house sources and viewing evaluations) from viewing records.

[0043] Further, generating a recommendation strategy according to the extracted historical user behavior features, such as preferentially recommending house sources that meet user preferences. Using reinforcement learning technology, the model continuously optimizes the recommendation strategy through trial and error. Apply the generated strategy to the recommendation system and adjust the strategy according to user feedback. A / B testing, also known as split testing or bucket testing, is a statistical method commonly used in marketing and product development, which compares two or more versions of web pages, application functions, advertisements or other elements to observe which version can bring better results. By comparing the effects of different strategies through A / B testing, the best solution is selected.

[0044] The house source recommendation method based on multi-dimensional behavior data provided by the embodiment dynamically adjusts the recommendation strategy according to the actual behavior and preferences of the user through the strategy model, making the recommendation more targeted.

[0045] Based on the above embodiment, in this embodiment, the user behavior data based on the plurality of data sources in step 200 is calculated by a pre-constructed similarity model to obtain a second recommendation list, including: Step 250, calculating the similarity between users based on a pre-constructed user house source scoring matrix, and determining target similar users based on the similarity between the users; wherein the target similar users are the users with the highest similarity to the current user; the user house source scoring matrix is constructed based on the historical user behavior data of the plurality of data sources, and the user house source scoring matrix is used to record the scores of the users on the house sources.

[0046] Step 260, obtaining the second recommendation list according to the recommendation list of the target similar user.

[0047] Continue to participate Figure 3The construction process of the similar model is described in detail.

[0048] Specifically, first, a user-house source score matrix is constructed to record the scores of users on house sources (such as the number of times of browsing, the number of times of collecting, etc.), and the similarity between users is calculated based on the user-house source score matrix using methods such as cosine similarity and Pearson correlation coefficient. Second, for each target user, the most similar other users are found, and a recommendation list is generated according to the preferences of the similar users. Finally, the results of the strategy model are fused to generate a final recommendation list.

[0049] Further, after step 260 obtains the second recommendation list according to the recommendation list of the target similar user, the method further comprises: obtaining the feedback result of the user on the second list; adjusting and optimizing the similar model based on the feedback result of the user on the second list.

[0050] Specifically, the feedback of the user on the recommended house source is collected for further optimization of the model.

[0051] The house source recommendation method based on multi-dimensional behavior data provided in the embodiment not only recommends house sources according to strategies, but also recommends similar house sources for customers based on a collaborative filtering algorithm, so that the viewing needs of customers are met to a greater extent.

[0052] Based on the above embodiment, in the present embodiment, the target recommendation list comprises at least one target recommended house source. After step 400 fuses the first recommendation list and the second recommendation list to obtain a target recommendation list, the method further comprises: displaying the target recommendation list and the recommendation reason of each target recommended house source in the at least one target recommended house source to a user interface.

[0053] Specifically, Figure 4 is a product flowchart of the house source recommendation method based on multi-dimensional behavior data provided by the present application, as Figure 4 shown, a user product interface is designed to display recommended house sources and their recommendation reasons. Each recommended house source should contain necessary information, such as house source name, location, price, area, etc., so that the user can quickly understand. For each house source in the target recommendation list, the system should generate and display one or more recommendation reasons. These reasons can be based on a variety of factors, such as the user's historical behavior data, the popularity of the house source, the price advantage, the convenience of the geographical location, the specific selling points of the house source (such as school district house, fine decoration, sea view house, etc.), etc.

[0054] The multi-dimensional behavior data-based house source recommendation method provided by the embodiment can display the target recommendation list and the recommendation reason of each target recommended house source to the user interface, so that the customer can intuitively and clearly view the house source information, and the user satisfaction is improved.

[0055] Figure 5 The system architecture diagram of the multi-dimensional behavior data-based house source recommendation method provided by the application, in combination with Figure 4 and Figure 5 The multi-dimensional behavior data-based house source recommendation method provided by the embodiment is described.

[0056] 1. The broker can enter the house source recommendation system through various channels, including but not limited to IM (instant messaging software), enterprise WeChat, and process network. These channels provide flexible choices for customers, so that they can access the system according to their own preferences and habits.

[0057] 2. After the broker enters the system, the broker can click the "select customer" button to select their target customers. For example, according to customer needs such as housing budget, geographical location preference, etc., to filter and determine potential target customers.

[0058] 3. The broker can click the "select house" button to browse the house source. The system will generate a personalized house source recommendation list based on the model according to the customer's preferences and choices. Each house source will be accompanied by AI-generated recommendation reasons, which aim to help customers better understand the characteristics and advantages of the house source. The broker can check the suitable house source according to their own needs for further operation.

[0059] 4. The broker can choose to share the selected house source with the customer. For example, social media platforms such as WeChat, as well as internal communication tools such as IM, enterprise WeChat, and process network.

[0060] 5. The customer can view the house source shared by the broker through the "view shared house source function", and the system can track their browsing data, such as the number of views, likes, and comments, and other interactive information.

[0061] 6. If the customer is interested in a certain house source, they can express their opinions and needs through the feedback channel provided by the system. In addition, the customer can also choose the "intention house source initiates a reservation to see the house" function to reserve the time for real estate viewing.

[0062] The method for recommending a house source based on multi-dimensional behavior data provided by the embodiment of the present application makes strategy weighting calculation based on multiple data sources of customer browsing, key attention, business opportunity and viewing conditions, makes up for the deficiency of single browsing behavior data, and can comprehensively understand customer demand and preference through comprehensive analysis of the multiple data; in addition to recommending a house source according to a strategy, a collaborative filtering algorithm is added to recommend similar house sources for customers, so that the viewing demand of customers can be met to a greater extent; and the operation mode is more in line with the existing division of broker posts.

[0063] The device for recommending a house source based on multi-dimensional behavior data provided by the present application is described below, and the device for recommending a house source based on multi-dimensional behavior data described below can be correspondingly referred to the method for recommending a house source based on multi-dimensional behavior data described above.

[0064] Figure 6 The device for recommending a house source based on multi-dimensional behavior data provided by the present application is described below, and the device for recommending a house source based on multi-dimensional behavior data described below can be correspondingly referred to the method for recommending a house source based on multi-dimensional behavior data described above. Figure 6 The device for recommending a house source based on multi-dimensional behavior data provided by the present application is described below, and the device for recommending a house source based on multi-dimensional behavior data described below can be correspondingly referred to the method for recommending a house source based on multi-dimensional behavior data described above. The acquisition module 601 is configured to acquire user behavior data of multiple data sources. The recommendation module 602 is configured to input the user behavior data of the multiple data sources into a pre-constructed strategy model to obtain a first recommendation list, and perform calculation based on the user behavior data of the multiple data sources through a pre-constructed similarity model to obtain a second recommendation list; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users. The fusion module 603 is configured to perform fusion processing on the first recommendation list and the second recommendation list to obtain a target recommendation list.

[0065] The device for recommending a house source based on multi-dimensional behavior data provided by the present application acquires user behavior data of multiple data sources, inputs the user behavior data of the multiple data sources into a pre-constructed strategy model to obtain a first recommendation list, and performs calculation based on the user behavior data of the multiple data sources through a pre-constructed similarity model to obtain a second recommendation list; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users; and the first recommendation list and the second recommendation list are fused to obtain a target recommendation list. Therefore, the present application can realize more accurate and personalized recommendation by constructing an intelligent matching system of multi-source data fusion, help brokers better understand customer demand, and provide house source options more in line with customer expectations.

[0066] Based on the above embodiment, in this embodiment, the device further comprises a preprocessing module, specifically for: After obtaining the user behavior data of the plurality of data sources, the user behavior data of the plurality of data sources is preprocessed. The data preprocessing includes data cleaning, data standardization and feature extraction.

[0067] Based on the above embodiment, in this embodiment, the device further comprises a construction module, specifically for: Obtain historical user behavior data of a plurality of data sources; Extract behavior features from the historical user behavior data of the plurality of data sources to obtain historical user behavior features; Train a reinforcement learning model based on the historical user behavior features to obtain an initial strategy model; Adjust and optimize the initial strategy model according to user feedback to obtain the strategy model; wherein the user feedback is obtained based on user feedback on the results generated by the initial strategy model.

[0068] Based on the above embodiment, in this embodiment, the recommendation module 602, specifically for: Calculate the similarity between users based on a pre-constructed user housing source score matrix, and determine target similar users based on the similarity between users; wherein the target similar user is the user with the highest similarity to the current user; the user housing source score matrix is constructed based on the historical user behavior data of the plurality of data sources, and the user housing source score matrix is used to record the scores of users on housing sources; Obtain the second recommendation list according to the recommendation list of the target similar user.

[0069] Based on the above embodiment, in this embodiment, the device further comprises an adjustment module, specifically for: After obtaining the second recommendation list according to the recommendation list of the target similar user, obtain the feedback results of the user on the second list; Adjust and optimize the similar model based on the feedback results of the user on the second list.

[0070] Based on the above embodiment, in this embodiment, the target recommendation list includes at least one target recommended housing source; The device further comprises a display module, specifically for: After fusing the first recommendation list and the second recommendation list to obtain a target recommendation list, display the target recommendation list and the recommendation reason of each target recommended housing source in the at least one target recommended housing source to the user interface.

[0071] Figure 7 An example of a schematic diagram of a physical structure of an electronic device is shown in Figure 7 The electronic device can be a robot or other electronic device, which can include a processor 710, a communications interface 720, a memory 730, and a communications bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other through the communications bus 740. The processor 710 can invoke the logic instructions in the memory 730 to execute a house source recommendation method based on multi-dimensional behavior data, including: obtaining user behavior data of a plurality of data sources; inputting the user behavior data of the plurality of data sources into a pre-constructed strategy model to obtain a first recommendation list, and calculating a second recommendation list based on the user behavior data of the plurality of data sources through a pre-constructed similarity model; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users; fusing the first recommendation list and the second recommendation list to obtain a target recommendation list.

[0072] In addition, the logic instructions in the memory 730 described above can be implemented in the form of a software functional unit and sold or used as an independent product, which can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the part of the technical solutions that essentially contribute to the prior art can be embodied in the form of a software product, which is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in the various embodiments of the present application. The aforementioned storage medium includes a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0073] On the other hand, the present application also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer readable storage medium, and the computer program can be executed by a processor to enable a computer to execute the house source recommendation method based on multi-dimensional behavior data provided by the above-mentioned methods, including: obtaining user behavior data of a plurality of data sources; input the user behavior data of the plurality of data sources into a pre-constructed strategy model to obtain a first recommendation list, and perform calculation based on the user behavior data of the plurality of data sources through a pre-constructed similarity model to obtain a second recommendation list; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users; fuse the first recommendation list and the second recommendation list to obtain a target recommendation list.

[0074] In another aspect, the present application also provides a non-transitory computer readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the multi-dimensional behavior data-based house source recommendation method provided by the above method, comprising: obtaining user behavior data of a plurality of data sources; inputting the user behavior data of the plurality of data sources into a pre-constructed strategy model to obtain a first recommendation list, and performing calculation based on the user behavior data of the plurality of data sources through a pre-constructed similarity model to obtain a second recommendation list; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate the similarity between users; fusing the first recommendation list and the second recommendation list to obtain a target recommendation list.

[0075] The device embodiments described above are only schematic, wherein the units shown as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement without creative labor.

[0076] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software plus necessary general hardware platforms, and of course can also be realized by hardware. Based on such understanding, the above technical solutions, essentially or in terms of the contribution to the prior art, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0077] It should be pointed out finally that the above embodiments are only used to illustrate the technical solutions of the present application, but not to limit the same; and although the present application has been described in detail with reference to the foregoing embodiments, it should be appreciated by those skilled in the art that the technical solutions recorded in the foregoing embodiments can be modified, or some technical features thereof can be replaced equivalently; 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 application.

Claims

1. A housing recommendation method based on multi-dimensional behavioral data, characterized in that: include: Obtain user behavior data from multiple data sources; Inputting the user behavior data from the multiple data sources into a pre-built strategy model to obtain a first recommendation list, and performing calculations based on the user behavior data from the multiple data sources using a pre-built similarity model to obtain a second recommendation list; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate similarities between users; The first recommendation list and the second recommendation list are merged to obtain a target recommendation list.

2. The housing recommendation method based on multi-dimensional behavioral data according to claim 1, characterized in that: After obtaining the user behavior data from multiple data sources, the method further includes: Performing data preprocessing on the user behavior data from the multiple data sources; The data preprocessing includes data cleaning, data standardization and feature extraction.

3. The housing recommendation method based on multi-dimensional behavioral data according to claim 1, characterized in that: The process of constructing the policy model includes: Obtain historical user behavior data from multiple data sources; Extracting behavior features from the historical user behavior data of the multiple data sources to obtain historical user behavior features; Training the reinforcement learning model based on the historical user behavior characteristics to obtain an initial strategy model; The initial policy model is adjusted and optimized according to the user feedback result to obtain the policy model; wherein the user feedback result is obtained based on the user's feedback on the result generated by the initial policy model.

4. The housing recommendation method based on multi-dimensional behavioral data according to claim 3, characterized in that: The user behavior data based on the multiple data sources is calculated using a pre-built similarity model to obtain a second recommendation list, including: Calculating the similarity between users based on a pre-constructed user-property rating matrix, and determining target similar users based on the similarity between users; wherein the target similar user is: the user with the highest similarity to the current user; the user-property rating matrix is ​​constructed based on historical user behavior data from the multiple data sources, and is used to record users' ratings of properties; The second recommendation list is obtained according to the recommendation list of the target similar user.

5. The housing recommendation method based on multi-dimensional behavioral data according to claim 4, characterized in that: After obtaining the second recommendation list according to the recommendation list of the target similar user, the method further includes: Obtaining user feedback on the second list; The similarity model is adjusted and optimized based on the user's feedback result on the second list.

6. The housing recommendation method based on multi-dimensional behavioral data according to claim 1, characterized in that: The target recommendation list includes: at least one target recommended property; After fusing the first recommendation list and the second recommendation list to obtain a target recommendation list, the method further includes: The target recommendation list and the recommendation reason for each target recommended property in the at least one target recommended property are displayed on a user interface.

7. A housing recommendation device based on multi-dimensional behavioral data, characterized in that: include: Acquisition module, used to obtain user behavior data from multiple data sources; a recommendation module, configured to input the user behavior data from the multiple data sources into a pre-built strategy model to obtain a first recommendation list, and to calculate a second recommendation list based on the user behavior data from the multiple data sources using a pre-built similarity model; wherein the strategy model is obtained by training a reinforcement learning model based on historical user behavior data; and the similarity model is used to calculate similarity between users; The fusion module is used to fuse the first recommendation list and the second recommendation list to obtain a target recommendation list.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the housing recommendation method based on multi-dimensional behavioral data as described in any one of claims 1 to 6 is implemented.

9. 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, the method for recommending housing based on multi-dimensional behavioral data as claimed in any one of claims 1 to 6 is implemented.

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the method for recommending housing based on multi-dimensional behavioral data as claimed in any one of claims 1 to 6 is implemented.

Images