Real estate registration cross-provincial communication method and system based on artificial intelligence

Abstract

The invention discloses a real estate registration cross-provincial communication method and system based on artificial intelligence, and relates to the technical field of data processing. The method comprises the following steps: collecting a registration material and a verification material, extracting registration information by using an OCR model, obtaining a confidence score in combination with the verification material, constructing a federal intelligent collaborative model according to local databases of different provinces, obtaining material demand sets of the different provinces, and judging whether a user needs to supplement materials or not. Generating text conversion rules of different provinces according to the federal intelligent collaboration model, converting registration information of different provinces to obtain conversion information, judging whether the user meets a registration condition according to the conversion information, and generating a corresponding evidence storage record for the conversion information meeting the registration condition; the time consumed by offline business handling of the user can be reduced, and the cross-provincial business handling efficiency can be obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to an artificial intelligence-based method and system for cross-provincial real estate registration. Background Art

[0002] In the traditional real estate registration model, applicants must go to the location of the real estate in person to handle relevant business, which causes great inconvenience to applicants whose real estate is located outside the city. With the increasing maturity of technologies such as artificial intelligence, image recognition, and natural language processing, these technologies can be applied to the field of real estate registration to achieve rapid recognition and digital processing of paper materials submitted by applicants; In the prior art, data between different registration agencies often cannot be shared while ensuring their privacy, and different provinces often have different material requirements and differences in text descriptions for the same business, which directly leads to inefficiency in cross-provincial business and even makes it difficult to carry out. In view of the shortcomings of the prior art, the present invention provides a cross-provincial real estate registration method and system based on artificial intelligence. Summary of the invention

[0003] The purpose of the present invention is to provide a method and system for cross-provincial real estate registration based on artificial intelligence.

[0004] The purpose of the present invention can be achieved through the following technical solutions: A real estate registration inter-provincial system based on artificial intelligence, comprising the following modules: The data verification module is used to collect the user's registration materials and verification materials, pre-process them respectively, extract the corresponding registration information based on the registration materials using the OCR model, and obtain the user's confidence score in combination with the verification materials; The data collaboration module is used to obtain the geographic information of registration agencies in different provinces and set up corresponding local databases, and build a federal intelligent collaboration model for all provinces based on the local databases of different provinces; The data conversion module is used to obtain the material requirement sets of different provinces, and use the material requirement sets to determine whether the user needs to supplement the materials, generate text conversion rules for different provinces according to the federal intelligent collaborative model, and use the text conversion rules to convert the registration information of different provinces to obtain conversion information; The data evidence storage module is used to determine whether the user meets the registration conditions based on the acquired conversion information, and generate corresponding evidence records for the conversion information that meets the registration conditions.

[0005] Furthermore, the process of collecting the user's registration materials and verification materials and pre-processing the two materials includes: Setting an image acquisition unit, through which registration materials of the user are collected, wherein the registration materials refer to relevant images used for real estate registration; Setting a voice collection unit to collect verification materials of the user through the voice collection unit, wherein the verification materials refer to relevant voices used to verify the identity of the user; For registration materials, an adaptive threshold segmentation algorithm is used to remove background noise from registration materials, and a distortion correction algorithm is used to correct the tilt problem of the shooting angle. For verification materials, voiceprint separation technology is used to separate environmental noise from user voiceprints, and the voice features of user voiceprints are extracted.

[0006] Furthermore, the process of extracting corresponding registration information based on the registration materials using the OCR model and obtaining the user's confidence score in combination with the verification materials includes: Input the user's registration materials into the OCR model, use the OCR model to recognize the text content related to real estate registration in the registration materials, and fill the recognized text content into the preset registration form to generate corresponding registration information; Based on the pre-entered user's voice information, voice recognition technology is used to obtain the confidence score between the user's verification materials and his voice information, a corresponding confidence threshold is set, and the user's confidence score is compared with the confidence threshold. If the confidence score is greater than or equal to the confidence threshold, the user is marked as having passed the voice verification.

[0007] Furthermore, the geographical information of registration agencies in different provinces is obtained, and corresponding local databases are set up. The process of building a federal intelligent collaborative model for all provinces based on the local databases of different provinces includes: Collecting geographic information of registration agencies in different provinces, the geographic information refers to the geographical location of registration agencies in each province, and using GIS technology to construct GIS distribution maps of registration agencies in different provinces based on the collected geographic information; A corresponding local database is set up at each registration agency, and the local database is used to store the registration materials, verification materials, and registration information of users who have passed the voice verification at the corresponding registration agency; A lightweight model is deployed at each registration agency to extract features from all registration materials in its local database. The FedAvg algorithm is used to aggregate the model parameters of the registration agencies in each province to form a unified federal intelligent collaborative model.

[0008] Furthermore, the process of obtaining material requirement sets of different provinces and using the material requirement sets to determine whether the user needs to supplement materials includes: Obtain the business rules texts of registration agencies in different provinces, focus on the key clauses in the business rules texts of different provinces through the self-attention mechanism, and incorporate the necessary materials in the key clauses into the material requirement set of the registration agencies in the corresponding provinces; The province where the user expects to conduct real estate registration is taken as the target province, and the material requirement set of the user's target province is obtained. The various registration materials currently provided by the user are compared with the material requirement set to determine whether any registration materials are missing. If missing, the user is prompted to supplement the missing registration materials.

[0009] Furthermore, the text conversion rules of different provinces are generated according to the federal intelligent collaborative model, and the process of converting the registration information of different provinces using the text conversion rules to obtain conversion information includes: The text conversion rules are used to reflect the differences in the description texts used for the same data between different provinces. The target decoder is deployed in the lightweight model of the registration agency in the user's target province, and the shared encoder is deployed in the federated intelligent collaborative model, which is initialized and sent to the lightweight model of the user's target province. First, keep the parameters of the shared encoder unchanged, use the local database of the target province to train its target decoder to generate the target text, where the target text is the description text used by the target province, then keep the parameters of the target decoder unchanged, adjust the parameters of the shared encoder to adapt it to the target text, and upload the shared encoder to the federated intelligent collaborative model; The registration information corresponding to the registration materials of the user after the supplementary materials is obtained, and it is encoded through a shared encoder, and then decoded through a target decoder of the target province to obtain conversion information.

[0010] Furthermore, judging whether the user meets the registration conditions according to the acquired conversion information and generating corresponding evidence records for the conversion information that meets the registration conditions includes: According to the location relationship between the registration agencies in different provinces in the GIS distribution map, the corresponding blockchain distribution nodes are constructed to obtain the registration conditions of the user's target province, and the user's conversion information is compared with the registration conditions of the user's target province; If the registration conditions are met, real estate registration will be carried out for it, and a proof record will be generated, and the generated proof record will be uploaded to each blockchain distribution node for storage. If it is not met, real estate registration will not be carried out for it, and information that the conditions are not met will be fed back to the user.

[0011] A method for cross-provincial real estate registration based on artificial intelligence, comprising the following steps: Step S1: Collect the user's registration materials and verification materials, pre-process them respectively, extract the corresponding registration information based on the registration materials using the OCR model, and obtain the user's confidence score in combination with the verification materials; Step S2: Obtain the geographic information of registration agencies in different provinces, set up corresponding local databases, and build a federal intelligent collaboration model for all provinces based on the local databases of different provinces; Step S3: Obtain material requirement sets of different provinces, and use the material requirement sets to determine whether the user needs to supplement materials, generate text conversion rules of different provinces according to the federal intelligent collaborative model, and use the text conversion rules to convert the registration information of different provinces to obtain conversion information; Step S4: Determine whether the user meets the registration conditions based on the acquired conversion information, and generate a corresponding evidence record for the conversion information that meets the registration conditions.

[0012] Compared with the prior art, the present invention has the following beneficial effects: The present invention uses a federation algorithm to construct a federated intelligent collaborative model of all registration agencies based on the local database of each registration agency. On the one hand, it can integrate the data features of all registration agencies to form a unified large model, and on the other hand, it can protect the data privacy of each registration agency. By obtaining the material requirement sets of different provinces and comparing them with the registration materials provided by users during the business process, it is possible to promptly find the registration materials that users are missing and remind users to supplement them, which is conducive to reducing the time spent by users on offline business processing; By obtaining the text conversion rules between the user's province and the target province, the user's registration information can be converted into the descriptive text used by the target province, saving the cost of self-conversion for users and registration agencies, and significantly improving the efficiency of cross-provincial business handling. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 It is a schematic diagram of the present invention. DETAILED DESCRIPTION

[0014] like Figure 1 As shown in the figure, an AI-based inter-provincial real estate registration system includes the following modules: The data verification module is used to collect the user's registration materials and verification materials, pre-process them respectively, extract the corresponding registration information based on the registration materials using the OCR model, and obtain the user's confidence score in combination with the verification materials; The data collaboration module is used to obtain the geographic information of registration agencies in different provinces and set up corresponding local databases, and build a federal intelligent collaboration model for all provinces based on the local databases of different provinces; The data conversion module is used to obtain the material requirement sets of different provinces, and use the material requirement sets to determine whether the user needs to supplement the materials, generate text conversion rules for different provinces according to the federal intelligent collaborative model, and use the text conversion rules to convert the registration information of different provinces to obtain conversion information; The data evidence storage module is used to determine whether the user meets the registration conditions based on the acquired conversion information, and generate corresponding evidence records for the conversion information that meets the registration conditions.

[0015] It should be further explained that, in the specific implementation process, the process of collecting the user's registration materials and verification materials and pre-processing the two materials includes: An image acquisition unit is provided to collect the registration materials of the user through the image acquisition unit. The registration materials refer to the relevant images provided by the user for real estate registration, including applicant's ID, agent materials, ownership source documents, tax and fee certificates, and surveying and mapping data files; A voice collection unit is provided to collect verification materials of the user through the voice collection unit. The verification materials refer to relevant voices provided by the user for verifying his / her identity. In the embodiment of the present invention, the registration materials for real estate registration are in the form of images, and the verification materials for verifying the identity of the user are in the form of voices; The collected registration materials and verification materials are preprocessed respectively. For the registration materials, the background noise of the registration materials is removed by an adaptive threshold segmentation algorithm, and the distortion correction algorithm is used to correct the tilt problem of the shooting angle. For the verification materials, the voiceprint separation technology is used to separate the environmental noise from the user's voiceprint, and the speech features of the user's voiceprint are extracted and standardized into a 128-dimensional vector.

[0016] It should be further explained that, in the specific implementation process, the process of using the OCR model to extract the corresponding registration information based on the registration materials and obtaining the user's confidence score in combination with the verification materials includes: The full name of the OCR model is Optical Character Recognition, which is an optical character recognition model that converts text in an image into editable text through computer vision technology. It can recognize the text area in the image, convert the content of the text area into machine code, and optimize the recognition result based on semantic rules; Input the user's registration materials into the OCR model, use the OCR model to identify the text content related to real estate registration in the registration materials, such as the property certificate number, right holder, building area and other fields, and fill the identified text content into the preset registration form to generate the corresponding registration information; Based on the pre-recorded user's voice information, the user's voice information is also standardized into a 128-dimensional vector, and the similarity between the user's verification materials and his voice information is obtained by using voice recognition technology, recorded as a confidence score, and a corresponding confidence threshold is set, and the user's confidence score is compared with the confidence threshold; If the confidence score is greater than or equal to the confidence threshold, the user is marked as having passed the voice verification and is subsequently processed. Otherwise, if the confidence score is less than the confidence threshold, the user is marked as having failed the voice verification and is not subsequently processed.

[0017] It should be further explained that, in the specific implementation process, the geographical information of the registration agencies in different provinces is obtained, and the corresponding local databases are set up. The process of building a federal intelligent collaborative model for all provinces based on the local databases of different provinces includes: Collecting geographic information of registration agencies in different provinces, the geographic information refers to the geographical location of registration agencies in each province, expressed in longitude and latitude, and using GIS technology to construct GIS distribution maps of registration agencies in different provinces based on the collected geographic information; A corresponding local database is set up at the geographical location of each registration agency, and the local database is used to store the registration materials and verification materials of users who have passed the voice verification at the corresponding registration agency, as well as the corresponding registration information; A corresponding lightweight model is deployed at each registration agency to extract features from all registration materials in its local database. An update cycle is set. Every time an update cycle is reached, the FedAvg algorithm is used to aggregate the model parameters of the registration agencies in each province, such as gradients and weights, to form a unified feature model, which is recorded as a federated intelligent collaborative model. The FedAvg algorithm is short for Federated Averaging, which is a commonly used algorithm in federated learning technology. It is mainly used to solve model training problems under distributed data. It can obtain a global optimal model by iteratively exchanging model parameters between local databases and central servers multiple times without centralizing the data in each local database to a central server.

[0018] It should be further explained that, in the specific implementation process, the process of obtaining the material requirement sets of different provinces and using the material requirement sets to determine whether the user needs to supplement the materials includes: Obtain business rules texts of registration agencies in different provinces, such as rules and regulations documents, legal and regulatory documents, etc. The business rules texts are used to reflect the various material requirements required for the same registration business in different provinces; Through the self-attention mechanism, we focus on the key clauses in the business rules texts of different provinces, such as the list of required materials, and incorporate the required materials in the key clauses into the material requirement set of the registration agency of the corresponding province. The material requirement sets of different provinces are different. The province where the user expects to conduct real estate registration is taken as the target province, and the material requirement set of the user's target province is obtained. The various registration materials currently provided by the user are compared with the material requirement set of his target province to determine whether there are any missing registration materials. If so, the user is prompted to supplement the missing registration materials. If not, subsequent processing is carried out.

[0019] It should be further explained that, in the specific implementation process, the text conversion rules of different provinces are generated according to the federal intelligent collaborative model, and the process of converting the registration information of different provinces using the text conversion rules to obtain conversion information includes: Since the registration agencies in different provinces use different description texts for the same data, in order to realize cross-provincial real estate registration, it is necessary to obtain text conversion rules between different provinces, which are used to reflect the differences in the description texts used by different provinces for the same data; Deploy a target decoder in the lightweight model of the registration agency in the user's target province, deploy a shared encoder in the federated intelligent collaborative model, and send it to the lightweight model of the user's target province after initialization. First, keep the parameters of the shared encoder unchanged, and use the local database of the target province to train its target decoder to generate the target text, which is the description text used by the target province. Then keep the parameters of the target decoder unchanged, adjust the parameters of the shared encoder to adapt it to the target text, upload the shared encoder at this time to the federal intelligent collaborative model, obtain the registration information corresponding to the user's registration materials after the supplementary materials, encode it through the shared encoder, and then decode it through the target decoder of the target province to obtain the corresponding conversion information.

[0020] It should be further explained that, in the specific implementation process, judging whether the user meets the registration conditions based on the acquired conversion information, and generating corresponding evidence records for the conversion information that meets the registration conditions include: According to the location relationship between the registration agencies in different provinces in the GIS distribution map, the corresponding blockchain distribution nodes are constructed to obtain the registration conditions of the user's target province, which are used to reflect the various data requirements necessary for real estate registration in the target province; The user's conversion information is compared with the registration conditions of the user's target province. If the registration conditions are met, real estate registration will be performed for the user, and a corresponding evidence record will be generated. The generated evidence record will be uploaded to each blockchain distribution node for storage. If the conditions are not met, real estate registration will not be performed for the user, and information that the conditions are not met will be fed back to the user.

[0021] The embodiment of the present invention also includes a method for cross-provincial real estate registration based on artificial intelligence, comprising the following steps: Step S1: Collect the user's registration materials and verification materials, pre-process them respectively, extract the corresponding registration information based on the registration materials using the OCR model, and obtain the user's confidence score in combination with the verification materials; Step S2: Obtain the geographic information of registration agencies in different provinces, set up corresponding local databases, and build a federal intelligent collaboration model for all provinces based on the local databases of different provinces; Step S3: Obtain material requirement sets of different provinces, and use the material requirement sets to determine whether the user needs to supplement materials, generate text conversion rules of different provinces according to the federal intelligent collaborative model, and use the text conversion rules to convert the registration information of different provinces to obtain conversion information; Step S4: Determine whether the user meets the registration conditions based on the acquired conversion information, and generate a corresponding evidence record for the conversion information that meets the registration conditions.

[0022] The above embodiments are only used to illustrate the technical method of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical method of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical method of the present invention.

Claims

1. An artificial intelligence-based real estate registration inter-provincial system, characterized by: Includes the following modules: The data verification module is used to collect the user's registration materials and verification materials, pre-process them respectively, extract the corresponding registration information based on the registration materials using the OCR model, and obtain the user's confidence score in combination with the verification materials; The data collaboration module is used to obtain the geographic information of registration agencies in different provinces and set up corresponding local databases, and build a federal intelligent collaboration model for all provinces based on the local databases of different provinces; The data conversion module is used to obtain the material requirement sets of different provinces, and use the material requirement sets to determine whether the user needs to supplement the materials, generate text conversion rules for different provinces according to the federal intelligent collaborative model, and use the text conversion rules to convert the registration information of different provinces to obtain conversion information; The data evidence storage module is used to determine whether the user meets the registration conditions based on the acquired conversion information, and generate corresponding evidence records for the conversion information that meets the registration conditions.

2. According to claim 1, the inter-provincial real estate registration system based on artificial intelligence is characterized in that: The process of collecting registration materials and verification materials and pre-processing them separately includes: Setting an image acquisition unit, through which registration materials of the user are collected, wherein the registration materials refer to relevant images used for real estate registration; Setting a voice collection unit to collect verification materials of the user through the voice collection unit, wherein the verification materials refer to relevant voices used to verify the identity of the user; For registration materials, an adaptive threshold segmentation algorithm is used to remove background noise from registration materials, and a distortion correction algorithm is used to correct the tilt problem of the shooting angle. For verification materials, voiceprint separation technology is used to separate environmental noise from user voiceprints, and the voice features of user voiceprints are extracted.

3. According to claim 2, the inter-provincial real estate registration system based on artificial intelligence is characterized in that: The process of extracting registration information based on registration materials and obtaining confidence scores includes: Input the user's registration materials into the OCR model, use the OCR model to recognize the text content related to real estate registration in the registration materials, and fill the recognized text content into the preset registration form to generate corresponding registration information; Based on the pre-entered user's voice information, voice recognition technology is used to obtain the confidence score between the user's verification materials and his voice information, a corresponding confidence threshold is set, and the user's confidence score is compared with the confidence threshold. If the confidence score is greater than or equal to the confidence threshold, the user is marked as having passed the voice verification.

4. According to claim 3, the inter-provincial real estate registration system based on artificial intelligence is characterized in that: The process of setting up a local database and building a federated intelligent collaboration model includes: Collecting geographic information of registration agencies in different provinces, the geographic information refers to the geographical location of registration agencies in each province, and using GIS technology to construct GIS distribution maps of registration agencies in different provinces based on the collected geographic information; A corresponding local database is set up at each registration agency, and the local database is used to store the registration materials, verification materials, and registration information of users who have passed the voice verification at the corresponding registration agency; A lightweight model is deployed at each registration agency to extract features from all registration materials in its local database. The FedAvg algorithm is used to aggregate the model parameters of the registration agencies in each province to form a unified federal intelligent collaborative model.

5. According to claim 4, the inter-provincial real estate registration system based on artificial intelligence is characterized in that: The process of obtaining material requirement sets from different provinces and determining whether users need additional materials includes: Obtain the business rules texts of registration agencies in different provinces, focus on the key clauses in the business rules texts of different provinces through the self-attention mechanism, and incorporate the necessary materials in the key clauses into the material requirement set of the registration agencies in the corresponding provinces; The province where the user expects to conduct real estate registration is taken as the target province, and the material requirement set of the user's target province is obtained. The various registration materials currently provided by the user are compared with the material requirement set to determine whether any registration materials are missing. If missing, the user is prompted to supplement the missing registration materials.

6. According to claim 5, the inter-provincial real estate registration system based on artificial intelligence is characterized in that: The process of generating text conversion rules for different provinces and obtaining conversion information includes: The text conversion rules are used to reflect the differences in the description texts used for the same data between different provinces. The target decoder is deployed in the lightweight model of the registration agency in the user's target province, and the shared encoder is deployed in the federated intelligent collaborative model, which is initialized and sent to the lightweight model of the user's target province. First, keep the parameters of the shared encoder unchanged, use the local database of the target province to train its target decoder to generate the target text, where the target text is the description text used by the target province, then keep the parameters of the target decoder unchanged, adjust the parameters of the shared encoder to adapt it to the target text, and upload the shared encoder to the federated intelligent collaborative model; The registration information corresponding to the registration materials of the user after the supplementary materials is obtained, and it is encoded through a shared encoder, and then decoded through a target decoder of the target province to obtain conversion information.

7. According to claim 6, the inter-provincial real estate registration system based on artificial intelligence is characterized in that: The process of determining whether the user meets the registration conditions and generating a record of the conversion information that meets the registration conditions includes: According to the location relationship between the registration agencies in different provinces in the GIS distribution map, the corresponding blockchain distribution nodes are constructed to obtain the registration conditions of the user's target province, and the user's conversion information is compared with the registration conditions of the user's target province; If the registration conditions are met, real estate registration will be carried out for it, and a proof record will be generated, and the generated proof record will be uploaded to each blockchain distribution node for storage. If it is not met, real estate registration will not be carried out for it, and information that the conditions are not met will be fed back to the user.

8. A method for inter-provincial real estate registration based on artificial intelligence, which is implemented based on the inter-provincial real estate registration system described in any one of claims 1 to 7, characterized in that: The method comprises: Step S1: Collect the user's registration materials and verification materials, pre-process them respectively, extract the corresponding registration information based on the registration materials using the OCR model, and obtain the user's confidence score in combination with the verification materials; Step S2: Obtain the geographic information of registration agencies in different provinces, set up corresponding local databases, and build a federal intelligent collaboration model for all provinces based on the local databases of different provinces; Step S3: Obtain material requirement sets of different provinces, and use the material requirement sets to determine whether the user needs to supplement materials, generate text conversion rules of different provinces according to the federal intelligent collaborative model, and use the text conversion rules to convert the registration information of different provinces to obtain conversion information; Step S4: Determine whether the user meets the registration conditions based on the acquired conversion information, and generate a corresponding evidence record for the conversion information that meets the registration conditions.

Images