House resource searching method and system based on big data

Abstract

The invention relates to the technical field of information management, in particular to a house resource searching method and system based on big data, and the method comprises the steps: determining the basic position of a user according to the login information of the user; dividing each basic position to form a corresponding basic area; determining a basic area of each user, and forming an input portrait corresponding to the basic area according to keyword division of each user in the basic position; positioning the keyword description in the search content in the basic area according to the input portrait; and outputting the keywords, and matching the indexes according to the keywords. The housing resource information and the search habit of the user group are included in the housing resource search engine by utilizing a mode of setting the basic area, and the housing resource information is more comprehensively displayed to the user by utilizing a mode of displaying the queue, so that the problem that the user does not know the housing resource information of the location of the housing resource is effectively avoided; therefore, the search habit of the user cannot effectively match the user demand, and the accuracy of house resource search is effectively improved.

Description

Technical Field

[0001] This invention relates to the field of information management technology, and in particular to a method and system for searching housing listings based on big data. Background Technology

[0002] Existing property search systems are mainly divided into consumer-facing property search platforms and internal management systems serving real estate agencies.

[0003] For housing search platforms, their technology tends to directly serve renters or homebuyers, focusing on improving search efficiency and user experience. For internal management systems, their technology tends to encompass the entire process of digital management, including property / customer management, contract and finance, and office automation (OA) functions.

[0004] Chinese Patent Publication No. CN117314689B discloses a cloud computing-based big data real estate information processing system, relating to the field of information processing technology. The system includes: acquiring a user's commuting information; identifying a ring-shaped target area based on the commuting information and confirming properties within that ring-shaped target area as initial recommended properties; identifying high-quality properties based on the user's historical browsing information within a first preset condition; identifying desired properties based on the user's historical browsing information within a second preset condition; confirming a candidate set of properties based on the initial recommended properties, high-quality properties, and desired properties; confirming the number of properties in the candidate set; when the number of properties exceeds a first preset number, acquiring the user's historical search records and confirming search keywords; confirming the similarity value of each property in the candidate set based on the search keywords; identifying a target property from the candidate set; and recommending the target property to the user. This patent improves the accuracy of real estate information platforms recommending properties to users.

[0005] Chinese Patent Publication No. CN116883010B discloses an information technology consultation management system based on big data, including: an information collection module, a demand analysis module, a preliminary screening analysis module, a secondary screening analysis module, a preferred salesperson analysis module, a display terminal, and a database. By collecting basic information about current consultants and analyzing their preference for certain types of properties, it provides reliable data support for subsequent matching analysis between consultants and salespersons. This also addresses the issue of insufficient authenticity and reliability of salespersons in current technologies, further providing a reliable platform for ensuring consultation security and efficiency for current consultants.

[0006] However, the above method has the following problems: 1. When identifying users' key concerns, searching based on their habits can lead to insufficient property listings and create information cocoons. 2. When matching properties, the system only uses the user's historical search terms, which makes it difficult to find the user's actual needs and recommend effective properties. Summary of the Invention

[0007] To address these issues, this invention provides a big data-based property search method to overcome the problems of existing technologies that, when acquiring users' focus, rely on user habits for searching, resulting in insufficient property display and information silos. Furthermore, when matching properties, they rely solely on users' historical search terms, making it difficult to retrieve users' actual needs and recommend effective properties, thus leading to a decrease in the accuracy of property search results.

[0008] To achieve the above objectives, on the one hand, the present invention provides a housing search method based on big data, comprising: Establish a housing database; Several indexes are created based on the property information; The index is matched based on the search content, and a display queue of the housing information is formed; When matching the search results, the process also includes: Determine the user's basic location based on the user's login information; The basic locations are divided into corresponding basic regions; Determine the basic region for each user, and form the input profile corresponding to the basic region based on the keywords of each user in the basic location; Based on the input profile, locate the keyword descriptions in the search content within this basic area; Output each keyword and match the index based on the keywords.

[0009] Furthermore, the steps for any user to search for the property information include: Determine the user's basic location; Based on the input profile of this basic location, several keywords for the search content are determined; Determine the corresponding index based on keywords; and Determine the corresponding search area and search core; The search area is the basic area corresponding to the property. The search core is any location within a preset distance from the search area, and this location is determined by the keyword.

[0010] Furthermore, when generating the index, the housing database divides housing information according to the basic area and forms a regional index set for the basic area; For a single base region, its corresponding region index includes several sub-indexes formed according to the keywords of each user in that base region; The sub-index corresponds to the keyword.

[0011] Furthermore, the step of generating the display queue for any user includes: Determine the base location and search area; Determine the corresponding set of region indexes based on the search region; Determine several keywords based on the input profile; The sub-indexes of the corresponding basic regions are used to generate corresponding display queues based on the keywords.

[0012] Furthermore, when the display queue is generated, it is sorted according to the word frequency of each keyword corresponding to the input profile, wherein, When the input profile is generated, the word frequency queue of each keyword is determined based on the keywords corresponding to the user search instructions in the basic region.

[0013] Furthermore, when the display queue is generated, it is sorted according to the word frequency of each keyword corresponding to the region index set, wherein, When the regional index set is generated, the frequency queue of each keyword is determined based on the keywords corresponding to the searched properties in the search region.

[0014] Furthermore, the word frequency queue is updated according to a preset period, wherein, For a single word frequency queue, it stores the word frequency ranking of keywords corresponding to two consecutive preset periods; When the preset period is reached, the word frequency ranking of the keywords corresponding to the new preset period is combined with the word frequency ranking of the keywords corresponding to the previous preset period to generate a new word frequency queue.

[0015] Furthermore, for any given user, the maximum number of searches is determined based on that user's search history, thus defining the user's search limit; When any user enters more than the maximum number of searches in a single search, it is determined that the user has reached the search limit, and the display queue is generated according to the input profile and the basic region to reset the display queue.

[0016] On the other hand, the present invention provides a housing search system based on big data, comprising: Several user terminals; A housing database used to store housing data; A user database used to store user address data; Also includes: The region identification module is used to identify and generate basic locations; The user server is used to generate an input profile corresponding to the user terminal based on the user database; A property listing server is used to form a regional index set based on the search area; The display server is used to generate display queues based on the user's search input, the input profile, and the regional index set.

[0017] Furthermore, the user terminal is also equipped with a language recognition module, which generates several corresponding keywords based on the search content.

[0018] Compared with the prior art, the beneficial effects of the present invention are that by setting a basic area, housing information and user search habits are incorporated into the housing search engine, and by using a display queue, housing information is presented to users in a more comprehensive way. This effectively avoids the situation where users' search habits cannot be effectively matched with their needs due to their lack of understanding of housing information in the location of the housing, and at the same time, it effectively improves the accuracy of housing search.

[0019] Furthermore, by setting up basic regions, sub-indexes for different keywords are generated for each property listing. A set of regional indexes is then used to store these sub-indexes, forming an index of indexes. This architecture not only effectively improves search efficiency but also enhances the storage efficiency of property information. By outputting the indexes, it effectively covers most properties listings with compound keywords, thereby significantly improving the accuracy of property searches.

[0020] Furthermore, by dividing users into regions, the search habits of users in those regions can be determined. This effectively avoids the problem of insufficient keywords due to user habits, while making the search results more adaptable to the actual needs of users, thereby effectively improving the accuracy of property search.

[0021] Furthermore, by setting a cycle and a maximum number of searches, the number of user searches can be limited, preventing users from constantly entering information cocoons caused by keywords in the property search results. At the same time, it ensures that the actual needs of users are not overlooked by the search results, making the search results more adaptable to the actual needs of users, thereby effectively improving the accuracy of property search.

[0022] The beneficial effects of this invention also lie in the fact that by setting up a housing database, a user database, an area identification module, a user server, a housing server, and a display server, the housing information is organized and sorted according to the search habits of users in a single area and the search status of housing information. The sorting results are then output. This effectively avoids the situation where users' search habits cannot be effectively matched with their needs due to their lack of understanding of housing information in the location of the housing, and at the same time, it effectively improves the accuracy of housing search. Attached Figure Description

[0023] Figure 1 This is a flowchart of the big data-based housing search method of the present invention; Figure 2 This is a schematic diagram of the housing search system of the present invention; Figure 3 This is a flowchart illustrating the process of searching for housing information according to an embodiment of the present invention; Figure 4 This is a flowchart illustrating the model training process in an embodiment of the present invention. Detailed Implementation

[0024] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0025] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0026] It should be noted that in the description of this invention, the terms "upper", "lower", "left", "right", "inner", "outer", etc., which indicate directions or positional relationships, are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and is not intended to indicate or imply that the device or element must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, it should not be construed as a limitation of this invention.

[0027] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0028] It is important to note that the user information and property information collected in this manual must be authorized by the user or property owner before collection. Unauthorized information will not be considered in this plan. In addition, the processed user information or property information and resources shall not be transmitted or displayed to any unauthorized third party.

[0029] Please see Figure 1 The flowchart shown is a big data-based housing search method of the present invention, including: Step S1: Establish a housing database; Step S2: Generate several indexes based on the property information; Step S3: Match the index according to the search content and form a queue of housing information for display; When matching search terms, this also includes: Step S301: Determine the user's basic location based on the user's login information; Step S302: Divide each basic location into corresponding basic areas; Step S303: Determine the basic region for each user, and form the input profile corresponding to the basic region based on the keywords of each user in the basic location; Step S304: Locate the keyword descriptions in the search content within the basic area according to the input profile; Step S305: Output each keyword and match the index based on the keywords.

[0030] By setting up basic regions, the property search engine incorporates property information and user search habits, and uses a display queue to present property information to users more comprehensively. This effectively avoids situations where users' search habits cannot be matched with their needs due to their lack of knowledge about the property's location, thus significantly improving the accuracy of property searches.

[0031] Example 1: Taking City A as an example: Establish a housing database: Collect housing information for the entire city of A, including property ID, detailed address (e.g., "No. 123, AB Road, XX Community"), administrative division (e.g., A1 District, A2 District), business district / sector (e.g., Aa Business District, Ab Business District), unit type, area, price, and tags (e.g., "near subway", "fully furnished").

[0032] Divide the basic area: City A is divided into basic areas such as "Aa Business District", "Ab Business District", "Ac Business District", "Ad Business District" and "Ae Business District".

[0033] Forming a regional index set: Create a regional index set for each base region (e.g., "Aa business district").

[0034] This index set contains multiple sub-indexes, each corresponding to a high-frequency keyword. For example: The “Aa Business District” regional index set may contain:

[0035] Sub-index Metro: Associates listings near Metro Line 1's "Museum Station" and "Stadium Station".

[0036] Sub-index Two-Bedroom: Associates all properties with a two-bedroom layout.

[0037] Sub-index "Furnished": Associates all properties with a "Furnished" status.

[0038] When conducting a search, the business district is used as the basis for sorting by the frequency of occurrence of sub-indexes.

[0039] Please see Figure 2 As shown, it is a schematic diagram of the structure of the housing search system of the present invention, including: Several user terminals are used to input users' search queries; The user database is connected to each user terminal and is used to store user address data; The region identification module is connected to user data to identify and generate basic locations; The housing database is connected to the area identification module and each user terminal to store housing data and the searched data of housing. The user server is connected to the region identification module and the user database to form an input profile corresponding to the user terminal based on the user database. The property server is connected to the area identification module and the property database to form an area index set based on the search area; The display server is connected to each user terminal, user server, and property server. It is used to generate display queues based on the user's search input, input profile, and regional index set.

[0040] By setting up a property database, a user database, a region identification module, a user server, a property server, and a display server, the system organizes and sorts property information based on users' search habits and the search status of property information in a single region, and outputs the sorting results. This effectively avoids situations where users' search habits cannot be effectively matched with their needs due to their lack of knowledge about property information in the property's location, and at the same time, it effectively improves the accuracy of property search.

[0041] Specifically, the user terminal is also equipped with a language recognition module, which generates several corresponding keywords based on the search content.

[0042] Please see Figure 3 The diagram shows a flowchart of a property search process according to an embodiment of the present invention. The steps for any user to search for property information include: Step S3021: Determine the user's basic location; Step S3022: Determine several keywords for the search content based on the input profile of the basic location; Step S3023: Determine the corresponding index based on the keywords; and Determine the corresponding search area and search core; The search area is the basic area corresponding to the property; The search core is any location within a preset distance from the search area, and this location is determined by the keywords.

[0043] Example 2: Based on Example 1, the division of the basic region can also be as follows: The administrative districts under the jurisdiction of City A (such as District A1, District A2, District A3, District A4, etc.) are used as the basic area.

[0044] It is understandable that this division can be based on the actual situation and is not restricted here. However, for the division results, the information of each property in the divided area should include at least 5 identical keywords.

[0045] Specifically, when generating an index, the housing database divides housing information according to a basic region and forms a regional index set for that basic region; For a single base region, its corresponding region index includes several sub-indexes formed according to the keywords of each user in that base region; The sub-indexes correspond to the keywords.

[0046] By setting up basic regions, sub-indexes for different keywords are generated for properties, and a set of generated regional indexes is used to store the sub-indexes, forming an index of indexes. This architecture not only effectively improves search efficiency, but also enhances the storage efficiency of property information. Furthermore, by outputting the indexes, it effectively covers most properties with compound keywords, thereby significantly improving the accuracy of property search.

[0047] Specifically, the steps for any user to generate a display queue include: Determine the base location and search area; Determine the corresponding set of region indexes based on the search region; Determine several keywords based on the input profile; Generate corresponding display queues for the sub-indexes of the corresponding basic regions based on keywords.

[0048] By segmenting users into their own regions, the search habits of users in those regions can be determined. This effectively avoids the problem of insufficient keywords due to user habits, while making the search results more adaptable to the actual needs of users, thereby effectively improving the accuracy of property search.

[0049] Example 3: Based on Example 1, the difference is that the user's input profile belongs to City H, which is north of City A: At this time, because users are unfamiliar with the traffic conditions in City A, they cannot effectively predict their local travel needs, so keywords such as "near the subway" and "short time" cannot be added to the search query. At this point, the keywords such as "near the subway" and "short time" are stored in the regional index set of City A. Without the user performing a search, this method displays the sub-index corresponding to the hidden keywords.

[0050] Example 4: Based on Example 1, the difference is that the user's input profile belongs to City S, which is south of City A: At this time, because users are unfamiliar with the climate of City A, they cannot effectively predict local essential needs such as heating. Therefore, relevant search keywords should be added. The keyword is stored in the regional index set of City A. Without the user performing a search, this method displays the sub-index corresponding to the hidden keyword.

[0051] Specifically, when the display queue is generated, it is sorted according to the word frequency of each keyword corresponding to the input profile. When the input profile is generated, the frequency queue of each keyword is determined based on the keywords corresponding to the user's search instructions in the basic area.

[0052] Specifically, when the display queue is generated, it is sorted according to the word frequency of each keyword corresponding to the region index set. When the regional index set is generated, the frequency queue of each keyword is determined based on the keywords corresponding to the searched properties in the search area.

[0053] Specifically, the word frequency queue is updated at a preset period, whereby... For a single word frequency queue, it stores the word frequency ranking of keywords corresponding to two consecutive preset periods; When a preset period is reached, the word frequency ranking of the keywords corresponding to the new preset period is combined with the word frequency ranking of the keywords corresponding to the previous preset period to generate a new word frequency queue.

[0054] In practice, the preset period can be one month; At this time, if a search is performed in March, the system collects information for March and displays the word frequency rankings for January and February. Upon entering April, the system displays the word frequency rankings for February and March, while storing the information for January.

[0055] Understandably, in actual use, this preset cycle should be adaptively adjusted according to local climate, population flow and other conditions, which will not be elaborated here.

[0056] Specifically, for any user, the maximum number of searches is determined based on the user's search history, thus determining the user's search limit; When any user enters more than the maximum number of searches in a single search, the user is deemed to have reached the search limit, and the display queue is generated according to the input profile and basic region to reset the display queue.

[0057] In implementation, the maximum number of searches should be set specifically according to user habits. A default setting, such as 5 searches, can be used first. If a user searches for a particular query less than 5 times, this setting will remain unchanged. If a user searches for a certain search term less than 5 times and continues to search, the number of searches will be increased by 1.

[0058] This approach allows for adaptation adjustments during implementation to prevent users from creating information cocoons. In particular, for search commands with longer statements, the maximum number of searches should be increased accordingly.

[0059] By setting a period and a maximum number of searches, the number of user searches is limited, preventing users from constantly entering information cocoons caused by keywords in the property search results. At the same time, it ensures that the actual needs of users are not overlooked by the search results, making the search results more relevant to the actual needs of users, thereby effectively improving the accuracy of property search.

[0060] Please see Figure 4 As shown, it is a flowchart of the model training process in an embodiment of the present invention; As one embodiment, the language recognition module can be provided by a big data model. This application provides a training method for a large language model, which is for reference only and should be verified and adjusted according to the actual situation. Step 1: Process user search history data to provide training samples for building user input profiles: Step 2, initialize the dataset; Step 3: Convert the original text into a numerical representation that the model can process, so as to provide standardized input for subsequent keyword extraction and index matching; Step 4: Convert the text segmentation results into an index sequence to facilitate subsequent vectorization processing and to unify the sequence length to ensure consistency of model input; Step 5: Tagging, labeling each word to indicate whether it is a keyword; Step 6: Use Chinese word segmentation technology to decompose the user input; Step 7: Remove stop words and invalid characters, and extract meaningful lexical units; Step 8: Construct a vocabulary list within the basic region and collect word frequency information; Step 9: Calculate the frequency of each word in user searches within the region; Step 10: Simulate different weights for users' attention to search results; Step 11: Calculate the attention weight of each word to reflect its importance to the search intent.

[0061] The steps described above are merely examples; adjustments may be made for specific regions. Specifically, in practice, if the sample size is too small, users or properties in that area will be merged into adjacent areas for processing.

[0062] Specifically, this application provides an example of the above training process, which can be divided into the following steps: 1. Data Collection and Preprocessing Module Regional division: It will be divided into several main basic regions; User search data collection: Collecting search behavior data from users in various regions; Housing database construction: Establish a complete housing information database; 2. Model Training Phase Phase 1: Region Input Profile Generation; Extract keywords specific to each region; Construct a word frequency queue to reflect regional search hotspots; Phase Two: Keyword Recognition Model Training; The architecture used is CNN+BiLSTM+Attention; Configure training parameters to optimize model performance, including: Learning rate, training batch, and number of training epochs The learning rate can be determined by the sample size of the region and the search size, such as: The sample size of region α is 600, the number of searches per week is 30, the judgment sample size and search volume are relatively low, the learning rate is set to 1.3, the training batch size is set to 24, and the number of training rounds is set to 30. The sample size of region β is 800, the number of searches per week is 50, the sample size and search volume are relatively balanced, the standard learning rate is set to 1, the training batch size is set to 36, and the number of training rounds is set to 15. The sample size of region γ is 1000, the number of searches per week is 50, the judgment sample size is relatively high, the learning rate is set to 0.9, the training batch size is set to 36, and the number of training rounds is set to 15. The sample size for region δ is 800, the number of searches per week is 30, the sample size is relatively high, but the number of searches is relatively low. The learning rate is set to 1.2, the training batch size is set to 24, and the number of training rounds is set to 30. 3. System Deployment and Updates Deployment strategy: Use canary releases to ensure stability; Regular updates: Establish a periodic update mechanism; Extreme handling: Preventing system overload; During implementation, a conservative strategy will be adopted when a new feature is first launched: Deployment strategy - Canary release parameters: Gray-scale release period: 14 days Initial user ratio: 0.5% (approximately 500 seed users); The order of region rolling is: Region α → Region β → Region γ → Region δ; Daily expansion rate: 50% (exponential growth); Failure rollback threshold: Error rate > 1% or response time > 2 seconds; A / B test load balancing ratio: 50% for new features and 50% for existing features; Regular updates - Periodic parameter updates: Data update: Property data: Updated incrementally every hour; User behavior data: Updated in batches every 15 minutes; Regional hot word frequency: Full calculation at 2 AM daily; Model update: Incremental training (20 rounds) every Monday at 4 AM. Full training (50 rounds) begins at 4 AM on the first Monday of each month. Deep training every quarter (100 rounds); Index Reconstruction: Incremental builds are performed daily at 3 AM. Full rebuild every Sunday at 11 PM; Reconstruction immediately after special events (such as housing fairs); Limit handling - Overload prevention parameters: Search limit: Regular users: 30 times per minute, 300 times per day; VIP users: 100 times per minute, 1000 times per day; Agent users: 200 times per minute, 5000 times per day; Search limit trigger: After 10 consecutive searches yielding no results, a recommendation list will be provided. When a single user makes more than 5 concurrent requests, the request is processed through a queue. When the system load is greater than 80%, automatically limit the rate of new requests; Emergency Response: If the database response time is >3 seconds, enable caching first. When the keyword recognition model fails, the basic matching mode is activated. When a region index becomes invalid, switch to an adjacent region index; When performing large-scale version upgrades, implement a smooth transition strategy: Deployment strategy - Canary release parameters: Gray-scale release period: 21 days Phased upgrade strategy: Days 1-7: Internal staff and test users (1%) Days 8-14: Core broker users (5%) Days 15-21: Open to regular users in batches; Regional batches: First batch: Region β and Region γ; Second batch: Region α; Third batch: Region δ; Performance monitoring thresholds: Error rate tolerance: 0.5%; Performance degradation tolerance: Response time increase <20%; Functional degradation trigger: CPU usage > 85% for 10 minutes; Regular updates - Periodic parameter updates: Emergency Fix: Critical bug: Hotfix within 4 hours of discovery; Data error: Stop the erroneous data source immediately upon detection; Security vulnerability: Emergency update within 2 hours of discovery; Version rollback mechanism: Retain the three most recent stable versions; Rollback detection time: 24 hours after upgrade; Rollback trigger condition: User complaint rate > 2%; Backup strategy: Model backup: Automatic backup after each training session; Data backup: Hourly incremental backup, daily full backup; Configuration backup: Manually back up the configuration before each change; Limit handling - Overload prevention parameters: System capacity planning: Daily capacity: Supports 100,000 concurrent users; Peak capacity: Supports 500,000 concurrent users (such as during a housing fair); Extended contingency plan: Automatically expand to 200% capacity; Overload protection mechanism: Search service degradation: Complex queries are automatically converted to simple queries; Cache penetration protection: Popular listings are cached locally; Database protection: Limit write operations and prioritize read operations; Failover strategy: Primary service failure: Switch to backup service within 30 seconds; Database failure: Switch to read-only replica within 60 seconds; CDN failure: Switch to backup CDN within 5 minutes; During routine iterations and optimizations, implement a fast iteration strategy: Deployment strategy - Canary release parameters: Release frequency: Every Tuesday and Thursday at 4 PM; Automated processes: Code submission → Automated testing → Pre-release environment → Automated deployment; Test coverage requirement: >80%; Automated test pass rate: 100%; Small-scale verification: Each batch of updates is first pushed to 0.1% of users; Verification time: 2 hours; Validation metrics: Click-through rate, conversion rate, error rate; Quick rollback: Automatic rollback trigger: Error rate > 0.3%; Rollback time: <3 minutes; Post-rollback verification: Automatically run core test cases; Regular updates - Periodic parameter updates: Data update strategy: Real-time data stream: Real-time processing of user behavior; Nearline data: Batch processing every 5 minutes; Offline data: Daily batch calculations; Model hot update: Keyword recognition model: Updated online daily; Ranking model: Updated online weekly; Input profile: Updated incrementally every hour; Index optimization: Hotspot analysis: Analyzed hourly; Index defragmentation: Defragment once a week; Query schedule update: Statistics are updated daily; Limit handling - Overload prevention parameters: Intelligent rate limiting strategy: Dynamic rate limiting based on user behavior; Flexible quotas based on time periods; Differentiated traffic restrictions based on regional hotspots; Resource optimization mechanism: Query cache time: 5 minutes for regular queries, 30 minutes for popular queries; Connection pool management: minimum 50 connections, maximum 500 connections; Memory management: JVM heap memory 8GB, cache memory 4GB; Monitoring and alarm system: Real-time monitoring: response time, error rate, throughput; Warning thresholds: Response time > 500ms, error rate > 0.1%; Alarm levels: Level 3 (Minor, Warning, Serious); The above parameters are determined based on the proportion of users and the time period of the gray-scale release. Among them, based on the amount of resources and user activity, region β is the key region, and the risk range can be controlled by promoting the release in different regions. When performing regular updates, the data update frequency is matched with business needs. At the same time, the effect and training cost are balanced according to the model training cycle, and a backup strategy is implemented to ensure data security. In extreme situations, prevention should be the primary focus, and reasonable access restrictions should be set. At the same time, elastic scaling should be implemented: dynamically adjust according to the activity characteristics of each region, and prioritize ensuring basic services in the event of a failure.

[0063] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

[0064] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A housing search method based on big data, comprising: Establish a housing database; Several indexes are created based on the property information; The index is matched based on the search content, and a display queue of the housing information is formed; The feature is that, when matching the search content, it further includes: Determine the user's basic location based on the user's login information; The basic locations are divided into corresponding basic regions; Determine the basic region for each user, and form the input profile corresponding to the basic region based on the keywords of each user in the basic location; Based on the input profile, locate the keyword descriptions in the search content within this basic area; Output each keyword and match the index based on the keywords.

2. The housing search method based on big data according to claim 1, characterized in that, The steps for any user to search for the property information include: Determine the user's basic location; Based on the input profile of this basic location, several keywords for the search content are determined; Determine the corresponding index based on keywords; and Determine the corresponding search area and search core; The search area is the basic area corresponding to the property. The search core is any location within a preset distance from the search area, and this location is determined by the keyword.

3. The housing search method based on big data according to claim 1, characterized in that, When generating the index, the housing database divides housing information according to the basic area and forms a regional index set for the basic area; For a single base region, its corresponding region index includes several sub-indexes formed according to the keywords of each user in that base region; The sub-index corresponds to the keyword.

4. The housing search method based on big data according to claim 2 or 3, characterized in that, The step of generating the display queue for any user includes: Determine the base location and search area; Determine the corresponding set of region indexes based on the search region; Determine several keywords based on the input profile; The sub-indexes of the corresponding basic regions are used to generate corresponding display queues based on the keywords.

5. The housing search method based on big data according to claim 4, characterized in that, When the display queue is generated, it is sorted according to the word frequency of each keyword corresponding to the input profile, wherein, When the input profile is generated, the word frequency queue of each keyword is determined based on the keywords corresponding to the user search instructions in the basic region.

6. The housing search method based on big data according to claim 4, characterized in that, When the display queue is generated, it is sorted according to the word frequency of each keyword corresponding to the region index set, wherein... When the regional index set is generated, the frequency queue of each keyword is determined based on the keywords corresponding to the searched properties in the search region.

7. The housing search method based on big data according to claim 5 or 6, characterized in that, The word frequency queue is updated at a preset period, whereby... For a single word frequency queue, it stores the word frequency ranking of keywords corresponding to two consecutive preset periods; When the preset period is reached, the word frequency ranking of the keywords corresponding to the new preset period is combined with the word frequency ranking of the keywords corresponding to the previous preset period to generate a new word frequency queue.

8. The housing search method based on big data according to claim 7, characterized in that, For any user, determine the maximum number of searches based on that user's search history, and thus determine that user's search limit. When any user enters more than the maximum number of searches in a single search, it is determined that the user has reached the search limit, and the display queue is generated according to the input profile and the basic region to reset the display queue.

9. A housing search system based on big data, used to perform the method described in claims 1-8, comprising: Several user terminals; A housing database used to store housing data; A user database used to store user address data; Its characteristic is that it further includes: The region identification module is used to identify and generate basic locations; The user server is used to generate an input profile corresponding to the user terminal based on the user database; The property listing server is used to create a regional index set based on the search area; The display server is used to generate display queues based on the user's search input, the input profile, and the regional index set.

10. The housing search system based on big data according to claim 9, characterized in that, The user terminal is also equipped with a language recognition module, which generates several corresponding keywords based on the search content.

Images