Furniture style intelligent recommendation system and method

Abstract

The present application relates to the technical field of user preference profiling, in particular to a furniture style intelligent recommendation system and method, comprising a user preference event collection module, a furniture style feature library, a user preference profile generation module, a scene demand constraint module, a recommendation matching module and a feedback correction module, the user preference event collection module converts user preference related data into preference event data, the user preference profile generation module generates a user preference profile according to the mapping result of the preference event data and the furniture style hierarchical label, the scene demand constraint module forms scene constraint data, the recommendation matching module outputs according to the constraint matching relationship, filters or rearranges the candidate furniture style set, and the feedback correction module corrects the user preference profile or the recommendation sorting basis according to the interaction feedback. Through preference event mapping, scene constraint matching and feedback correction, the furniture style recommendation information is made to correspond to the current scene demand and user feedback.

Description

Technical Field

[0001] This invention relates to the field of user preference profiling technology, and in particular to a furniture style intelligent recommendation system and method. Background Technology

[0002] The field of user preference profiling technology involves computer information processing that collects, organizes, labels, and models user basic information, behavioral data, interest characteristics, selection tendencies, and scenario requirements. This technology is commonly used in industries such as e-commerce, content recommendation, home design, smart shopping guides, and personalized services to support systems in identifying, classifying, and matching user needs. Specifically, a traditional furniture style intelligent recommendation system refers to a computer system that, for furniture style selection scenarios, organizes and processes furniture style data, user preference data, and scenario requirement data based on user browsing, selection, collection, purchase, evaluation, and active input to provide users with furniture style-related recommendations.

[0003] Traditional furniture style recommendation systems often rely on user behavior records and actively input information to generate recommendation results. The correspondence between user behavior and furniture style features lacks a traceable data foundation. Historical preferences can easily directly affect the current recommendation process. The current scenario requirements do not adequately constrain candidate furniture styles. The output, filtering, or rearrangement of candidate results lacks a unified judgment basis. Interactive feedback is difficult to reflect in subsequent recommendation processes, leading to a deviation between furniture style recommendation information and the user's current selection preferences. Summary of the Invention

[0004] The purpose of this invention is to provide a furniture style intelligent recommendation system and method. By associating user preference-related data, furniture style hierarchical tags, current scenario demand data, and interactive feedback, a traceable data relationship is formed between user behavior and furniture style characteristics. The furniture style recommendation process can generate recommendation information by combining current scenario demands and feedback correction results.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A furniture style intelligent recommendation system includes a user preference event collection module, a furniture style feature library, a user preference profile generation module, a scene requirement constraint module, a recommendation matching module, and a feedback correction module.

[0007] The user preference event collection module collects user preference-related data and converts it into preference event data, which includes behavioral attribute information. The user preference profile generation module reads furniture style level tags from the furniture style feature library and maps the preference event data to these tags to obtain a mapping result. The user preference profile generation module also generates a user preference profile based on the mapping result. The scene requirement constraint module obtains current scene requirement data and converts it into scene constraint data. The recommendation matching module generates a candidate furniture style set based on the user preference profile. When a candidate furniture style matches the scene constraint data, furniture style recommendation information is output; otherwise, the set is filtered or rearranged. The feedback correction module collects interactive feedback on the furniture style recommendation information and corrects the user preference profile or recommendation ranking criteria based on this feedback to obtain updated furniture style recommendation information.

[0008] Furthermore, the process of collecting user preference-related data includes the following steps: the user preference event collection module receives user interaction data and input data during the furniture style selection process, writes the interaction data into the user behavior record according to the generation time, writes the input data into the user demand record according to the input fields, and associates the user behavior record and the user demand record with the same user identifier to form the user preference-related data.

[0009] Furthermore, the process of converting the user preference-related data into the preference event data includes: extracting the behavior occurrence time, the object of the behavior, and the source of the behavior from the user behavior record; extracting the requirement input content and the corresponding scenario from the user requirement record; writing the behavior occurrence time, the object of the behavior, the source of the behavior, the requirement input content, and the corresponding scenario into the same event record; when the event record has the object of the behavior or the corresponding scenario, the event record is identified as the preference event data; when the event record lacks both the object of the behavior and the corresponding scenario, writing the event record into the preference event data is stopped.

[0010] Further, the process of mapping the preference event data to the furniture style hierarchy tags includes: when the preference event data has the behavior object, reading the object tag corresponding to the behavior object in the furniture style feature library, and splitting the object tag into a main style tag and a subordinate style tag according to the style hierarchy relationship; when the preference event data lacks the behavior object but has the corresponding scenario of the need, reading the scene tag corresponding to the corresponding scenario of the need in the furniture style feature library, and splitting the scene tag into a main style tag and a subordinate style tag according to the style hierarchy relationship; when the preference event data has the corresponding scenario, reading the scene tag corresponding to the corresponding scenario ... in the furniture style feature library, reading the scene tag corresponding to the corresponding scenario in the furniture style feature library, and splitting the scene tag into a main style tag and a subordinate style tag according to the style hierarchy relationship; when the preference event data has the corresponding scenario in the furniture style feature library, reading the scene tag corresponding to the corresponding scenario in the furniture style feature library, and splitting the scene tag into a main style tag and a subordinate style tag according to the style hierarchy relationship; when the preference event data has the corresponding scenario in the furniture style feature library, reading the scene tag corresponding to the corresponding scenario in the furniture When describing the source of a behavior, a first mapping relationship is established between the primary style tag and the source of the behavior. When the preference event data lacks the source of the behavior but has a corresponding scenario for the demand, a first mapping relationship is established between the primary style tag and the corresponding scenario for the demand. When the preference event data has a corresponding scenario for the demand, a second mapping relationship is established between the subordinate style tag and the corresponding scenario for the demand. When the preference event data lacks the corresponding scenario for the demand but has the object of the behavior, a second mapping relationship is established between the subordinate style tag and the object of the behavior. The first mapping relationship and the second mapping relationship are merged into the mapping result.

[0011] Furthermore, the furniture style feature library stores furniture style data according to the furniture style hierarchical tags, whereby the furniture style hierarchical tags include a main style category tag, style attribute tags, and scene adaptation tags. When the user preference profile generation module reads the furniture style hierarchical tags, it first determines the range of candidate tags according to the main style category tag, then narrows the range of candidate tags according to the style attribute tags and the scene adaptation tags, and uses the narrowed range of candidate tags as the tag range to map to the preference event data.

[0012] Furthermore, the process by which the user preference profile generation module generates the user preference profile based on the mapping result includes: determining the behavioral attribute category corresponding to the mapping result according to the behavioral attribute information in the preference event data; aggregating the mapping results belonging to the same furniture style level label under the same user identifier; setting a profile weight for the aggregated mapping results based on the behavioral attribute category; and writing the furniture style level label and the profile weight into the same profile record to form the user preference profile.

[0013] Furthermore, the process of generating the scene constraint data includes: obtaining space usage information and style restriction information from the current scene requirement data; converting the space usage information into scene adaptation constraints; converting the style restriction information into style exclusion constraints; writing the scene adaptation constraints and the style exclusion constraints into the same scene constraint record; and using the scene constraint record as the scene constraint data that constrains the candidate furniture style set.

[0014] Further, the process by which the recommendation matching module generates the candidate furniture style set based on the user preference profile includes: sorting the furniture style level tags according to the profile weights in the user preference profile; using the top-ranked furniture style level tags as candidate recall tags; and extracting corresponding furniture style data from the furniture style feature library based on the candidate recall tags to form the candidate furniture style set. The recommendation matching module compares each candidate in the candidate furniture style set with the scene constraint data. When the furniture style level tag corresponding to a candidate meets the constraints in the scene constraint data, the candidate is determined as an output object; when the furniture style level tag corresponding to a candidate does not meet the constraints in the scene constraint data, the candidate is determined as a pending object, and filtering or rearranging is performed on the pending object.

[0015] Furthermore, the feedback correction module collects interactive feedback regarding the furniture style recommendation information, and the process of correcting the user preference profile or the recommendation ranking criteria based on the interactive feedback includes: recording the user's selection results for the furniture style recommendation information; associating the selection results with the furniture style level tags corresponding to the output objects to form a feedback association record; when the feedback association record matches the furniture style level tags in the user preference profile, updating the recommendation ranking criteria according to the feedback association record; when the feedback association record does not match the furniture style level tags in the user preference profile, correcting the user preference profile according to the feedback association record, and generating updated furniture style recommendation information based on the corrected user preference profile.

[0016] Furthermore, a method for intelligently recommending furniture styles is provided. This method is applied to an intelligent furniture style recommendation system and includes the following steps:

[0017] S1: Collect user preference-related data and convert it into preference event data, wherein the preference event data includes behavioral attribute information;

[0018] S2: Read the furniture style hierarchy tags from the furniture style feature library, and map the preference event data with the furniture style hierarchy tags to obtain the mapping result;

[0019] S3: Generate a user preference profile based on the mapping result, obtain current scenario demand data, and convert the current scenario demand data into scenario constraint data;

[0020] S4: Generate a set of candidate furniture styles based on the user preference profile. When the candidate furniture styles in the set form a constraint matching relationship with the scene constraint data, output furniture style recommendation information. When the candidate furniture styles in the set do not form a constraint matching relationship with the scene constraint data, filter or rearrange the set of candidate furniture styles.

[0021] S5: Collect interactive feedback on the furniture style recommendation information, and adjust the user preference profile or the recommendation ranking criteria formed by the recommendation matching process based on the interactive feedback to obtain updated furniture style recommendation information.

[0022] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0023] User preference-related data is used to generate preference event data carrying behavioral attribute information. Based on the mapping results between preference event data and furniture style hierarchy tags, user preference profiles are generated, establishing a traceable link between user behavior and furniture style characteristics. Scene constraint data is generated according to current scenario requirements and applied to the candidate furniture style set, ensuring that the influence of historical preferences on recommendation results is constrained by current scenario requirements. Based on the constraint matching relationship between candidate styles and scene constraint data, the candidate furniture style set is filtered or rearranged to ensure that furniture style recommendations correspond to current scenario conditions. User preference profiles or recommendation ranking criteria are corrected based on interactive feedback, ensuring that subsequent recommendation processes reflect past user feedback and reduce the deviation between furniture style recommendations and users' current selection tendencies. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 This is the main flowchart of the system of the present invention;

[0026] Figure 2 This is a schematic diagram illustrating the formation of user preference-related data in this invention;

[0027] Figure 3 This is a schematic diagram of the preference event data mapping relationship of the present invention;

[0028] Figure 4 This is a schematic diagram illustrating the collaborative recommendation between user preference profiles and scenario constraint data in this invention.

[0029] Figure 5 This is a schematic diagram of the interactive feedback correction of the present invention. Detailed Implementation

[0030] The technical solutions described in this invention will be further described below with reference to the accompanying drawings. In subsequent embodiments, the user preference event collection module, furniture style feature library, user preference profile generation module, scene requirement constraint module, recommendation matching module, and feedback correction module shall all use the names already determined in the protection scope text and the invention content. User preference related data, preference event data, behavioral attribute information, furniture style hierarchical tags, mapping results, user preference profiles, current scene requirement data, scene constraint data, candidate furniture style set, furniture style recommendation information, interactive feedback, recommendation ranking criteria, and updated furniture style recommendation information all represent the corresponding data objects or processing results in this invention.

[0031] Please see Figures 1 to 5 This embodiment provides a furniture style intelligent recommendation system. This system correlates user preference data, furniture style hierarchy tags, and current scenario requirements during the furniture style selection process, and corrects subsequent recommendation processes based on interactive feedback. The furniture style intelligent recommendation system includes a user preference event collection module, a furniture style feature library, a user preference profile generation module, a scenario requirement constraint module, a recommendation matching module, and a feedback correction module.

[0032] The user preference event collection module serves as the data entry point for the intelligent furniture style recommendation system, collecting user preference-related data. This data originates from user interaction and input data during the furniture style selection process. Interaction data refers to recordable operational data generated by users during furniture style selection, while input data refers to user-initiated demand-related data. After receiving the interaction and input data, the user preference event collection module writes the interaction data into the user behavior record according to its occurrence time and the input data into the user demand record according to the input fields. The user behavior and demand records are then associated using the same user identifier. These user behavior and demand records, linked by the same user identifier, together form the user preference-related data, enabling subsequent processing to retrieve the source of behavior and demand from the same user perspective.

[0033] The user preference event collection module converts user preference-related data into preference event data. The conversion process uses user behavior records and user request records as input. From user behavior records, it extracts the behavior occurrence time, the object affected by the behavior, and the source of the behavior. From user request records, it extracts the request input content and the corresponding scenario. The behavior occurrence time represents the time when the interaction data was generated; the object affected by the behavior represents the furniture style object or related object that the user interaction affected; the source of the behavior represents the type of source from which the behavior occurred; the request input content carries the user's actively input request content; and the corresponding scenario represents the furniture style selection scenario corresponding to the request input content.

[0034] The user preference event collection module writes the behavior occurrence time, the object affected by the behavior, the source of the behavior, the input content of the requirement, and the corresponding scenario of the requirement into the same event record. The event record is used to carry both behavior-side fields and requirement-side fields within the same data structure. When an event record has an object affected by the behavior or a corresponding scenario, the event record is identified as preference event data; when an event record lacks both an object affected by the behavior and a corresponding scenario, it is no longer written into the preference event data. Therefore, preference event data entering the subsequent mapping process must at least have an object affected by the behavior or a corresponding scenario that can be associated with the furniture style level tag, preventing data lacking a basis for association from entering the mapping chain.

[0035] The furniture style feature library stores furniture style data and organizes it according to furniture style hierarchy tags. These tags include a main style category tag, style attribute tags, and scene adaptation tags. The main style category tag identifies the higher-level style category in the furniture style data; the style attribute tag identifies the attribute content related to the style features; and the scene adaptation tag identifies the adaptation relationship between the furniture style data and scene requirements. The furniture style feature library provides readable furniture style hierarchy tags to the user preference profile generation module, enabling a tagged correspondence between preference event data and furniture style data.

[0036] The user preference profile generation module reads furniture style hierarchical tags from the furniture style feature library. During this process, the module first determines the range of candidate tags based on the main style category tag, then narrows this range down to style attribute tags and scene adaptation tags. This narrowed range is then used as the tag range for mapping with preference event data. By first determining the range of higher-level tags and then narrowing down the attributes and scenes, the mapping between preference event data and furniture style hierarchical tags is limited to data related to furniture style.

[0037] When mapping preference event data to furniture style hierarchy tags, the user preference profile generation module determines the tag reading path based on whether the preference event data contains a behavioral object and a corresponding demand scenario. When the preference event data contains a behavioral object, the user preference profile generation module reads the object tags corresponding to the behavioral object from the furniture style feature library and splits these object tags into primary style tags and subordinate style tags according to style hierarchy. Object tags represent the furniture style tag content derived from the behavioral object; primary style tags establish a higher-level mapping relationship related to the source of the behavior; and subordinate style tags establish a lower-level mapping relationship related to the corresponding demand scenario or the behavioral object.

[0038] When preference event data lacks a behavioral target but has a corresponding scenario, the user preference profile generation module reads the scenario tags corresponding to the scenario from the furniture style feature library and splits the scenario tags into primary style tags and subordinate style tags according to style hierarchy. Scenario tags are used to represent the furniture style tag content derived from the scenario, enabling preference event data lacking a behavioral target but having a corresponding scenario to still enter the furniture style hierarchy tag mapping process.

[0039] When establishing the first mapping relationship, if the preference event data has a behavioral source, the user preference profile generation module establishes a first mapping relationship between the main style label and the behavioral source; if the preference event data lacks a behavioral source but has a corresponding demand scenario, the user preference profile generation module establishes a first mapping relationship between the main style label and the corresponding demand scenario. The first mapping relationship is used to represent the correspondence between the main style label and the higher-level preference source in the preference event data.

[0040] When establishing the second mapping relationship, if the preference event data has a corresponding demand scenario, the user preference profile generation module establishes a second mapping relationship between the subordinate style tag and the corresponding demand scenario; if the preference event data lacks a corresponding demand scenario but has a behavioral object, the user preference profile generation module establishes a second mapping relationship between the subordinate style tag and the behavioral object. The second mapping relationship is used to represent the correspondence between the subordinate style tag and the specific scenario or specific object in the preference event data. The user preference profile generation module merges the first and second mapping relationships into a mapping result, which serves as the input for generating the user preference profile.

[0041] The user preference profile generation module generates user preference profiles based on the mapping results. It determines the behavioral attribute categories corresponding to the mapping results according to the behavioral attribute information in the preference event data. These behavioral attribute categories distinguish the sources of influence of different preference events on furniture style level tags. The module aggregates mapping results belonging to the same furniture style level tag under the same user identifier and assigns profile weights to the aggregated mapping results based on the behavioral attribute categories. These profile weights represent the ranking and recall effects of the corresponding furniture style level tag within the user preference profile. The module writes the furniture style level tag and profile weight into the same profile record, forming the user preference profile. This profile record provides the recommendation matching module with the tag basis for generating a set of candidate furniture styles.

[0042] The scene requirement constraint module acquires current scene requirement data and converts it into scene constraint data. This data includes space usage information and style restriction information. Space usage information represents the space intended for the current furniture style selection, while style restriction information represents the style conditions that need to be excluded or restricted in the selection. The scene requirement constraint module converts the space usage information into scene adaptation constraints and the style restriction information into style exclusion constraints, writing both into the same scene constraint record. This scene constraint record serves as the scene constraint data for the set of candidate furniture styles and is provided to the recommendation matching module for constraint matching.

[0043] The recommendation matching module receives user preference profiles and scene constraint data. Based on the user preference profiles, the module generates a set of candidate furniture styles. When generating the candidate furniture style set, the module sorts the furniture style hierarchical tags according to the profile weights in the user preference profiles, using the top-ranked furniture style hierarchical tags as candidate recall tags. Based on these candidate recall tags, the module extracts corresponding furniture style data from the furniture style feature library to form the candidate furniture style set. The candidate furniture styles in the set correspond to the furniture style hierarchical tags, enabling the use of scene constraint data in subsequent constraint matching.

[0044] The recommendation matching module compares each candidate furniture style in the set with the scene constraint data. When the furniture style level label corresponding to a candidate style meets the constraints in the scene constraint data, the recommendation matching module determines the candidate style as an output object and outputs furniture style recommendation information based on the output object. When the furniture style level label corresponding to a candidate style does not meet the constraints in the scene constraint data, the recommendation matching module determines the candidate style as an object to be processed and performs filtering or rearrangement on the objects to be processed. Filtering is used to prevent candidates that do not meet the constraints in the scene constraint data from entering the output range, and rearrangement is used to change the output order of the objects to be processed in the candidate furniture style set. Thus, the furniture style recommendation information is formed by the combined effect of user preference profiles and scene constraint data.

[0045] The feedback correction module collects interactive feedback on furniture style recommendations and adjusts user preference profiles or recommendation ranking criteria based on this feedback. Interactive feedback originates from user selections related to furniture style recommendations. The feedback correction module records these selections and associates them with the corresponding furniture style level tags for output objects, forming a feedback association record. This record represents the correspondence between the user's actual feedback on the output furniture style recommendations and the corresponding furniture style level tags.

[0046] When the feedback association record matches the furniture style hierarchy tags in the user preference profile, the feedback correction module updates the recommendation ranking criteria according to the feedback association record. The recommendation ranking criteria influence the ranking of candidate options in the subsequent candidate furniture style set. When the feedback association record does not match the furniture style hierarchy tags in the user preference profile, the feedback correction module corrects the user preference profile according to the feedback association record and generates updated furniture style recommendation information based on the corrected user preference profile. By correcting the user preference profile or recommendation ranking criteria through the feedback association record, interactive feedback can enter the subsequent recommendation processing chain, ensuring that the updated furniture style recommendation information output later reflects the feedback results.

[0047] During the operation of the intelligent furniture style recommendation system, the preference event data output by the user preference event collection module serves as the input to the user preference profile generation module, the furniture style hierarchical tags output by the furniture style feature library serve as the mapping basis, the user preference profile output by the user preference profile generation module serves as the recall basis for the recommendation matching module, the scene constraint data output by the scene requirement constraint module serves as the constraint basis for the recommendation matching module, the furniture style recommendation information output by the recommendation matching module serves as the object of the feedback correction module's interactive feedback, and the updated furniture style recommendation information output by the feedback correction module serves as the output result of the subsequent recommendation process.

[0048] This embodiment also provides a method for intelligent furniture style recommendation, which is used to implement the above-mentioned intelligent furniture style recommendation system, including:

[0049] S1. Collect user preference-related data and convert it into preference event data, which includes behavioral attribute information.

[0050] When executing S1, the system receives user interaction and input data during the furniture style selection process. Interaction data is written to the user behavior record according to its occurrence time, and input data is written to the user requirement record according to its input fields. The user behavior record and user requirement record are associated with the same user identifier to form user preference-related data. Subsequently, the system extracts the behavior occurrence time, the object of the behavior, and the source of the behavior from the user behavior record, and extracts the requirement input content and the corresponding scenario from the user requirement record. These fields are then written to the same event record. When an event record has an object of the behavior or a corresponding scenario, the event record is identified as preference event data; when an event record lacks both an object of the behavior and a corresponding scenario, the writing of the event record to preference event data is stopped.

[0051] S2, read the furniture style level labels from the furniture style feature library, and map the preference event data with the furniture style level labels to obtain the mapping result.

[0052] When executing S2, the furniture style data stored in the furniture style feature library according to furniture style hierarchy tags is first read. Then, based on the behavior's target or the corresponding scenario in the preference event data, object tags or scenario tags are determined. Object tags or scenario tags are split into primary style tags and subordinate style tags according to style hierarchy relationships. A first mapping relationship and a second mapping relationship are established based on the behavior's source, the corresponding scenario, or the behavior's target in the preference event data. The first and second mapping relationships are then merged to form the mapping result.

[0053] S3 generates a user preference profile based on the mapping results, obtains the current scenario requirement data, and converts the current scenario requirement data into scenario constraint data.

[0054] When executing S3, the behavioral attribute category corresponding to the mapping result is determined according to the behavioral attribute information in the preference event data. The mapping results belonging to the same furniture style level label under the same user identifier are aggregated. The profile weight is set according to the behavioral attribute category of the aggregated mapping results, and the furniture style level label and profile weight are written into the same profile record to form a user preference profile. At the same time, the space usage information and style restriction information in the current scene requirement data are obtained. The space usage information is converted into scene adaptation constraints, and the style restriction information is converted into style exclusion constraints. The scene adaptation constraints and style exclusion constraints are written into the same scene constraint record to form scene constraint data.

[0055] S4 generates a set of candidate furniture styles based on user preference profiles. When the candidate furniture styles in the set form a constraint match with the scene constraint data, the furniture style recommendation information is output. When the candidate furniture styles in the set do not form a constraint match with the scene constraint data, the set of candidate furniture styles is filtered or rearranged.

[0056] When executing S4, the furniture style level tags are sorted according to the profile weights in the user preference profile. The top-ranked furniture style level tags are used as candidate recall tags, and corresponding furniture style data is extracted from the furniture style feature library based on the candidate recall tags to form a candidate furniture style set. Subsequently, the candidate furniture styles in the set are compared item by item with the scene constraint data. Based on whether the furniture style level tag corresponding to the candidate meets the constraints in the scene constraint data, the candidate is determined to be an output object or an object to be processed, and output, filtered, or rearranged accordingly.

[0057] S5 collects interactive feedback on furniture style recommendations and adjusts the user preference profile or the recommendation ranking criteria formed by the recommendation matching process based on the interactive feedback to obtain updated furniture style recommendations.

[0058] When executing S5, the user's selection results for furniture style recommendations are recorded. These selection results are then associated with the furniture style hierarchy tags corresponding to the output objects, forming a feedback association record. When the feedback association record matches the furniture style hierarchy tags in the user preference profile, the recommendation ranking criteria are updated according to the feedback association record. When the feedback association record does not match the furniture style hierarchy tags in the user preference profile, the user preference profile is corrected according to the feedback association record, and updated furniture style recommendations are generated based on the corrected user preference profile.

[0059] In the above system implementation, the user preference-related data, the preference event data, the furniture style hierarchy tags, the mapping results, the user preference profile, the scene constraint data, the candidate furniture style set, the furniture style recommendation information, the interactive feedback, the recommendation sorting criteria, and the updated furniture style recommendation information have a sequential relationship. The specific processes, field states, judgment conditions, data sources, module collaboration, and output results are used to explain the possible implementations of the present invention and should not limit the present invention to the specific implementations listed. Without departing from the technical solutions described in the present invention and the scope of the original disclosure, any equivalent substitutions, equivalent modifications, equivalent combinations, order adjustments, module corresponding replacements, equivalent changes in field names, equivalent succession of execution subjects, or equivalent changes in carrier forms that can be conceived by those skilled in the art should fall within the protection scope of this application; however, the above changes should not extend to unclaimed topics, nor should they change the substantive correspondence of technical objects through name changes, and they should not replace, delete, or weaken the data relationships, module collaboration relationships, judgment conditions, temporal relationships, and processing result succession relationships necessary to achieve the technical effect.

Claims

1. A furniture style intelligent recommendation system, characterized in that, It includes a user preference event collection module, a furniture style feature library, a user preference profile generation module, a scenario requirement constraint module, a recommendation matching module, and a feedback correction module; The user preference event collection module collects user preference-related data and converts the user preference-related data into preference event data, which includes behavioral attribute information. The process of collecting user preference-related data includes: The user preference event collection module receives user interaction data and input data during the furniture style selection process, writes the interaction data into the user behavior record according to the generation time, writes the input data into the user demand record according to the input fields, and associates the user behavior record and the user demand record according to the same user identifier to form the user preference related data. The user preference profile generation module reads the furniture style hierarchy tags in the furniture style feature library and maps the preference event data to the furniture style hierarchy tags to obtain the mapping result; The process of mapping the preference event data to the furniture style hierarchy tags includes: When the preference event data has the object of the behavior, read the object tag corresponding to the object of the behavior in the furniture style feature library, and split the object tag into a main style tag and a subordinate style tag according to the style hierarchy relationship; When the preference event data lacks the object of the behavior but has the scenario corresponding to the demand, read the scenario tag corresponding to the demand in the furniture style feature library, and split the scenario tag into a main style tag and a subordinate style tag according to the style hierarchy relationship; When the preference event data has the behavior source, a first mapping relationship is established between the main style tag and the behavior source; when the preference event data lacks the behavior source but has the corresponding scenario of the demand, a first mapping relationship is established between the main style tag and the corresponding scenario of the demand. When the preference event data has a scenario corresponding to the demand, a second mapping relationship is established between the subordinate style tag and the scenario corresponding to the demand; when the preference event data lacks a scenario corresponding to the demand but has the object of the behavior, a second mapping relationship is established between the subordinate style tag and the object of the behavior. The first mapping relationship and the second mapping relationship are combined into the mapping result; The user preference profile generation module generates a user preference profile based on the mapping result, and the scenario requirement constraint module obtains the current scenario requirement data and converts the current scenario requirement data into scenario constraint data. The recommendation matching module generates a set of candidate furniture styles based on the user preference profile. When the candidate furniture styles in the set form a constraint matching relationship with the scene constraint data, the module outputs furniture style recommendation information. When the candidate furniture styles in the set do not form a constraint matching relationship with the scene constraint data, the module filters or rearranges the set of candidate furniture styles. The feedback correction module collects interactive feedback on the furniture style recommendation information and corrects the user preference profile or recommendation ranking criteria based on the interactive feedback to obtain updated furniture style recommendation information.

2. The intelligent furniture style recommendation system according to claim 1, characterized in that, The process of converting the user preference-related data into the preference event data includes: Extract the behavior occurrence time, the object of the behavior, and the source of the behavior from the user behavior record; extract the requirement input content and the corresponding scenario from the user requirement record; write the behavior occurrence time, the object of the behavior, the source of the behavior, the requirement input content, and the corresponding scenario into the same event record; when the event record has the object of the behavior or the corresponding scenario, the event record is identified as the preference event data; when the event record lacks both the object of the behavior and the corresponding scenario, stop writing the event record into the preference event data.

3. The intelligent furniture style recommendation system according to claim 1, characterized in that, The furniture style feature library stores furniture style data according to the furniture style hierarchical tags. The furniture style hierarchical tags have style main category tags, style attribute tags, and scene adaptation tags. When the user preference profile generation module reads the furniture style hierarchical tags, it first determines the candidate tag range according to the style main category tags, then narrows the candidate tag range according to the style attribute tags and the scene adaptation tags, and uses the narrowed candidate tag range as the tag range to map with the preference event data.

4. The intelligent furniture style recommendation system according to claim 1, characterized in that, The process by which the user preference profile generation module generates the user preference profile based on the mapping result includes: The behavior attribute category corresponding to the mapping result is determined according to the behavior attribute information in the preference event data. The mapping results belonging to the same furniture style level label under the same user identifier are aggregated. The profile weight is set according to the behavior attribute category. The furniture style level label and the profile weight are written into the same profile record to form the user preference profile.

5. The intelligent furniture style recommendation system according to claim 4, characterized in that, The process of generating the scenario constraint data includes: Obtain the space usage information and style restriction information from the current scene requirement data, convert the space usage information into scene adaptation constraints, convert the style restriction information into style exclusion constraints, write the scene adaptation constraints and the style exclusion constraints into the same scene constraint record, and use the scene constraint record as the scene constraint data that constrains the candidate furniture style set.

6. The intelligent furniture style recommendation system according to claim 5, characterized in that, The process by which the recommendation matching module generates the candidate furniture style set based on the user preference profile includes: The furniture style level tags are sorted according to the image weight in the user preference profile. The furniture style level tags with the highest ranking are used as candidate recall tags. The corresponding furniture style data is extracted from the furniture style feature library according to the candidate recall tags to form the candidate furniture style set. The recommendation matching module compares the candidate furniture styles in the candidate furniture style set with the scene constraint data item by item. When the furniture style level label corresponding to the candidate furniture style meets the constraint conditions in the scene constraint data, the candidate furniture style is determined as an output object. When the furniture style level label corresponding to the candidate furniture style does not meet the constraint conditions in the scene constraint data, the candidate furniture style is determined as an object to be processed, and the object to be processed is filtered or rearranged.

7. The intelligent furniture style recommendation system according to claim 6, characterized in that, The feedback correction module collects interactive feedback regarding the furniture style recommendation information, and the process of correcting the user preference profile or the recommendation ranking criteria based on the interactive feedback includes: Record the user's selection results in response to the furniture style recommendation information, and associate the selection results with the furniture style level tags corresponding to the output objects to form a feedback association record; When the feedback association record matches the furniture style hierarchy label in the user preference profile, the recommendation ranking criteria are updated according to the feedback association record. When the feedback association record does not match the furniture style hierarchy label in the user preference profile, the user preference profile is corrected according to the feedback association record, and updated furniture style recommendation information is generated based on the corrected user preference profile.

8. A method for intelligently recommending furniture styles, the method being used to implement the intelligent furniture style recommendation system as described in any one of claims 1-7, characterized in that, Includes the following steps: Collect user preference-related data and convert it into preference event data, wherein the preference event data includes behavioral attribute information; Read the furniture style hierarchy tags from the furniture style feature library, and map the preference event data to the furniture style hierarchy tags to obtain the mapping result; A user preference profile is generated based on the mapping result, current scenario requirement data is obtained, and the current scenario requirement data is converted into scenario constraint data. A set of candidate furniture styles is generated based on the user preference profile. When the candidate furniture styles in the set form a constraint matching relationship with the scene constraint data, furniture style recommendation information is output. When the candidate furniture styles in the set do not form a constraint matching relationship with the scene constraint data, the set of candidate furniture styles is filtered or rearranged. Collect interactive feedback on the furniture style recommendation information, and adjust the user preference profile or the recommendation ranking criteria formed by the recommendation matching process based on the interactive feedback to obtain updated furniture style recommendation information.

Images