Segmented splitting similarity comparison method based on house addresses

By using a segmented similarity comparison method for property addresses, the accuracy and efficiency issues of property address matching in existing technologies are resolved, achieving automated and accurate address consistency judgment and reducing the cost of manual intervention.

CN118916711APending Publication Date: 2024-11-08BEIYIN FINANCIAL TECH CO LTD
View PDF 5 Cites 0 Cited by

Patent Information

Application Number
CN202411049234.5
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-01
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

Existing technologies for property address matching suffer from several problems: fuzzy index matching algorithms are sensitive to text differences, edit distance algorithms are greatly affected by address length, and manual judgment is costly and inefficient.

Method used

A method for segmenting and comparing the similarity of house addresses is adopted, including preprocessing, word segmentation and parsing, segment similarity calculation and address consistency judgment. The consistency of house addresses is calculated by similarity algorithms such as text edit distance and the weight of address element segmentation words.

Benefits of technology

It improves the accuracy of property address consistency judgment, realizes automated and efficient matching without human intervention, and reduces costs and time.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118916711A_ABST
    Figure CN118916711A_ABST
Patent Text Reader

Abstract

The invention discloses a similarity comparison method based on house address segmentation splitting, and the method comprises the steps: collecting house address information, carrying out the preprocessing of the house address information, and obtaining the preprocessed house address information; processing the preprocessed house address information by adopting a word segmentation analysis algorithm to obtain an analysis result; performing segmentation similarity calculation according to the analysis result to obtain similarity; and judging the consistency of the addresses according to the similarity. And the consistency judgment precision is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of house address standardization, and in particular to a similarity comparison method based on house address segmentation. Background Art

[0002] In the fields of housing management bureaus and provident funds, in order to smoothly handle real estate business, it is necessary to match the house address of historical real estate data with the current real estate data to determine whether they are the same house. As the business volume and house data volume increase, the cost of manual matching continues to increase, so matching needs to be done in an automated way. If the match is successful, the subsequent business process will be carried out, otherwise additional operations or manual review are required.

[0003] Existing technical solutions:

[0004] Currently, to determine whether house addresses are similar, the main method is to input the house address and use the fuzzy index matching algorithm to determine whether the historical house address is the same house as the current house; if no data with the same house address is found, the edit distance similarity algorithm is used to match and find the house address with the highest similarity match; then, manual judgment and other methods are used to determine whether the addresses are consistent.

[0005] The fuzzy index matching algorithm is suitable for the case where the text of the real estate address is completely consistent. Any slight difference in the text is considered inconsistent.

[0006] The edit distance similarity algorithm is sensitive to the length of the property address and is easily affected by the word segmentation order of the address identifier and the address segmentation words.

[0007] Manual judgment is costly, inefficient, and time-consuming. With the help of the edit distance similarity algorithm, the similarity of the house data is relatively high, and manual judgment is required to determine whether the house address data is consistent, but the implementation cost is high. Summary of the invention

[0008] In view of the above problems, the present invention is proposed to provide a similarity comparison method based on house address segmentation splitting to overcome the above problems or at least partially solve the above problems.

[0009] According to one aspect of the present invention, a similarity comparison method based on segmentation of house addresses is provided, the comparison method comprising:

[0010] Collecting house address information, and preprocessing the house address information to obtain preprocessed house address information;

[0011] Processing the pre-processed house address information using a word segmentation and parsing algorithm to obtain a parsing result;

[0012] Perform segment similarity calculation according to the analysis result to obtain similarity;

[0013] The consistency of the addresses is determined based on the similarity.

[0014] Optionally, the collecting house address information and preprocessing the house address information to obtain the preprocessed house address information specifically includes:

[0015] Collect house address information;

[0016] Unified filtering of special characters, simplified and traditional Chinese numbers, full-width and half-width input methods, administrative region name changes or mergers, and multiple synonyms for the same landmark of a building.

[0017] Optionally, the pre-processed house address information is processed using a word segmentation and parsing algorithm to obtain a parsing result, specifically comprising:

[0018] The house address is split according to the house address element segmentation words to obtain the house address element segmentation words corresponding to each level of administrative area, road level, building level and detail level and the identification words corresponding to the element segmentation words.

[0019] Optionally, the administrative regions include: provinces, cities and counties; the road levels include streets and numbers; the building levels include: buildings, communities and gardens; the detailed levels include: blocks, buildings, floors and households.

[0020] Optionally, performing segment similarity calculation according to the parsing result to obtain the similarity specifically includes:

[0021] For the address word segments obtained after splitting the two house addresses, the similarity of each address word segment is obtained through similarity algorithms such as text edit distance.

[0022] Optionally, judging the consistency of the addresses according to the similarity specifically includes:

[0023] Combine segment similarity calculation and address element segmentation word weight to determine whether the house addresses are consistent.

[0024] Optionally, the special character filtering includes: \r\n\t,,. ·..;;::、! ? @$%*^`~=+&'\"|\\ / 。

[0025] The present invention provides a similarity comparison method based on segmentation and splitting of house addresses, the comparison method comprising: collecting house address information, and preprocessing the house address information to obtain preprocessed house address information; processing the preprocessed house address information using a word segmentation and parsing algorithm to obtain a parsing result; performing segmentation similarity calculation based on the parsing result to obtain similarity; and judging the consistency of the address based on the similarity. The accuracy of consistency judgment is improved.

[0026] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use 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 ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0028] Figure 1 A flowchart of a similarity comparison method based on house address segmentation splitting provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0029] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0030] The terms "comprises" and "having" and any variations thereof in the description embodiments, claims and drawings of the present invention are intended to cover non-exclusive inclusions, for example, including a series of steps or units.

[0031] The technical solution of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments.

[0032] like Figure 1 As shown, a similarity comparison method based on segmented splitting of house addresses includes:

[0033] Preprocessing of house addresses: such as unified filtering of special characters, simplified and traditional Chinese numbers, full-width and half-width input methods, administrative region name change or merger, and multiple synonyms for the same landmark of a building.

[0034] Word segmentation and parsing algorithm: split the house address according to the house address element segmentation words, and obtain the house address element segmentation words corresponding to each level such as administrative area (province, city, county, etc.), road level (street, number, etc.), building level (building, community, garden, etc.), and detail level (block, building, floor, household, etc.) as well as the identification words corresponding to the element segmentation words.

[0035] Segment similarity calculation: For the address token fragments obtained after splitting the two house addresses, the similarity of each address token fragment is obtained through similarity algorithms such as text edit distance.

[0036] Address consistency judgment: Combine segment similarity calculation, address element segmentation word weights to judge whether the house addresses are consistent.

[0037] (1) House address preprocessing

[0038] Preprocess, clean and replace special characters, texts, etc. in the house address, and organize them into relatively standardized address texts.

[0039] Including:

[0040] Filter special characters: "\r\n\t,,。·..;;::、!?@$%*^`~=+&'\"|\\ / ";

[0041] Traditional Chinese to simplified Chinese replacement: Replace "壹、贰、叁到玖" with simplified numbers "一、二、三到九";

[0042] Special character replacement: Replace "~~---_——" with "-";

[0043] Simplified Chinese numeral replacement: Replace "一、二到九" with Arabic numerals "1、2, etc.";

[0044] (2) House address splitting

[0045] 1. House address text matching method:

[0046] For the two house address texts to be matched, use the address element segmentation words contained in the house address text to obtain their corresponding identifier words at the house address element level.

[0047] 2. House address element classification:

[0048] Determine the administrative region nouns, road nouns, building nouns, detailed nouns and detailed quantifiers in the house address string. The preset house address element segmentation word set includes administrative region, road level, road number level, building level, building area level, building number level, unit level, floor level and room level.

[0049] Use the house address element segmentation words as the starting or ending segmentation words of the clauses, and perform sentence segment division processing on the house address string to obtain one or more sentence segment fragments starting with the starting word and ending with the ending word;

[0050] 3. The designed house address element segmentation words are as follows

[0051] Administrative region element segmentation word set: province, city, county, district, township, town, street.

[0052] The segmentation words of house address elements at the road level include the following

[0053] The segmentation words of house address elements at the road level include: road, street, avenue, hutong, boulevard, main road, main street, street, community, block, lane;

[0054] The set of segmentation words for elements at the road number level: number;

[0055] The set of segmentation words for house address elements at the building level:

[0056] Mansion, building, community, garden, home, mansion, residence, villa, neighborhood, community, new village, garden, apartment, courtyard, center;

[0057] The set of segmentation words for house address elements at the detailed level includes:

[0058] At the building area level: phase, area, building

[0059] At the building level: building, building, building, building, building, building, building number;

[0060] At the unit level: unit;

[0061] At the level of floors: floor, building, staircase;

[0062] At the room level: household, room, house, room, number, door, -;

[0063] The set of quantifier segmentation character sets includes: Arabic numerals, Chinese numerals, Roman numerals, English capital letters, English lowercase letters, Chinese Heavenly Stems (such as Jia, Yi, Bing, Ding, etc.) characters, and directional words (such as east, west, south, north, etc.) characters

[0064] 4. Obtaining administrative region identification words:

[0065] Segment each house address text according to the pre-established administrative region segmentation words to determine the administrative region identification words at the administrative region level of each house address text. Use the administrative region element segmentation words as the starting or ending segmentation words of clauses to perform sentence segment division processing on the house address string, and obtain one or more segmented sentence segments of each house address text.

[0066] 5. Processing non-standard names of administrative regions:

[0067] For each house address text, search for non-standard names in the identification words at the administrative region level, and replace them with standard administrative region names using the pre-established synonym house address dictionary.

[0068] 6. Obtaining the initial selected text:

[0069] The identification words at the administrative district level in each house address text are removed to obtain a preliminary text corresponding to each house address text.

[0070] 7. Obtaining road-level identification words:

[0071] Each road-level house address element segmentation word set in each preliminary text is processed in a preset order. If the house address element keyword after the road-level house address element keyword is a number, it is determined that the house address element keyword corresponds to the road number level. If the house address element keyword after the road-level house address element keyword is not a number, and the house address element keyword is followed by at least one quantifier separator set, it is determined that the quantifier is an identification word of the road number level.

[0072] 8. Remove road level identifiers:

[0073] The road-level identification words in each house address text are removed to obtain a preliminary text corresponding to each house address text.

[0074] 9. Obtain building level identification words:

[0075] The house address element segmentation word set of each building level in each preliminary text is processed in a preset order to determine the building identification word in the house address character string.

[0076] If there is no building-level house address element separator, the detailed address element separator is used for word segmentation, and the segment before the first detailed address element separator is the building segment fragment; after the building segment fragment in the house address string is determined, the quantifier separator is used for word segmentation from back to front, and the segment fragment before the first continuous quantifier separator is the building-level identifier.

[0077] 10. Remove building level identifiers:

[0078] The building-level identifiers in each house address text are removed to obtain a preliminary text corresponding to each house address.

[0079] 11. Obtaining detailed level identifiers:

[0080] Utilize the detailed-level house address element segmentation words contained in the house address text to obtain the corresponding house address element level identification words. Process each house address element segmentation word in each preliminary text in turn: for the first house address element segmentation word in the preliminary text, determine the quantifier part of the preliminary text before the house address element segmentation word as the identification word of the house address element level corresponding to the house address element segmentation word; for any house address element segmentation word after the first house address element segmentation word in the preliminary text, determine the quantifier part of the preliminary text between the house address element segmentation word and the previous house address element segmentation word as the identification word of the house address element level corresponding to the house address element segmentation word.

[0081] 12. Special cases of address element segmentation words;

[0082] If the house address element keyword following the house address element keyword at the road number level or the landmark building level is a building, it is determined that the house address element keyword corresponds to the building level.

[0083] If the next house address element keyword of the building-level or unit-level house address element keyword is a floor, it is determined that the house address element keyword corresponds to a level.

[0084] If the next house address element segmentation word after the building-level, unit-level or floor-level house address element segmentation word is a number, it is determined that the house address element segmentation word corresponds to the household level.

[0085] If the building-level, unit-level or floor-level house address element separator is followed by at least one quantifier separator set and there is no detailed-level house address element separator, then the number is determined to be a household-level identification word.

[0086] If the building-level house address element separator does not exist, and the subsequent adjacent address of the building identifier is a hyphen "-" and a quantifier separator set, if there are two hyphen separators, there are no other characters.

[0087] If the building-level house address element separator exists, and the subsequent adjacent addresses at the building level are separated by a hyphen "-" and a quantifier separator, there is a total of one hyphen separator and no other characters.

[0088] (III) Segment address similarity calculation

[0089] According to the split house address element level, administrative region, road level, community level, and quantifier, the similarity is calculated based on the edit distance and cosine similarity algorithm of the two address identifiers.

[0090] (IV) House consistency judgment

[0091] A weight value is pre-configured for each house address element level; the similarity of the feature words of the two house address texts at each house address element level and the weighted average are used to calculate the consistency of the two house address texts. If the consistency exceeds a certain threshold, it is defined as the house address consistency.

[0092] Beneficial effect: No need for manual judgment, the accuracy of judgment is improved, and the method of automatic judgment is also realized.

[0093] The above specific implementation methods further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above are only specific implementation methods of the present invention and are not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A similarity comparison method based on house address segmentation. It is characterized in that The comparison method comprises: Collecting house address information, and preprocessing the house address information to obtain preprocessed house address information; Processing the pre-processed house address information using a word segmentation and parsing algorithm to obtain a parsing result; Perform segment similarity calculation according to the analysis result to obtain similarity; The consistency of the addresses is determined based on the similarity.

2. According to claim 1, a similarity comparison method based on house address segmentation splitting, It is characterized in that The collecting of house address information and preprocessing of the house address information to obtain the preprocessed house address information specifically includes: Collect house address information; Unified filtering of special characters, simplified and traditional Chinese numbers, full-width and half-width input methods, administrative region name changes or mergers, and multiple synonyms for the same landmark of a building.

3. According to claim 1, a similarity comparison method based on house address segmentation splitting, It is characterized in that The pre-processed house address information is processed using a word segmentation and parsing algorithm to obtain a parsing result, specifically comprising: The house address is split according to the house address element segmentation words to obtain the house address element segmentation words corresponding to each level of administrative area, road level, building level and detail level and the identification words corresponding to the element segmentation words.

4. According to claim 3, a similarity comparison method based on house address segmentation splitting, It is characterized in that The administrative regions include: provinces, cities and counties; the road levels include streets and numbers; the building levels include: buildings, communities and gardens; the detailed levels include: blocks, buildings, floors and households.

5. According to claim 1, a similarity comparison method based on house address segmentation splitting, It is characterized in that The segment similarity calculation according to the analysis result to obtain the similarity specifically includes: For the address word segments obtained after splitting the two house addresses, the similarity of each address word segment is obtained through similarity algorithms such as text edit distance.

6. According to claim 1, a similarity comparison method based on house address segmentation splitting, It is characterized in that The determining the consistency of the addresses according to the similarity specifically includes: Combine segment similarity calculation and address element segmentation word weight to determine whether the house addresses are consistent.

7. According to claim 2, a similarity comparison method based on house address segmentation splitting, It is characterized in that The special character filtering includes: \r\n\t,,. ·..;;::、! ? @$%*^`~=+&'\"|\\ / 。

Citation Information

Patent Citations

  • Evaluation method and device of address similarity

    CN106096024A

  • A text classification method based on similarity matching

    CN109033212A

  • House address similarity analysis method and device

    CN112347221A

  • Address matching method based on inverted index and neural network algorithm

    CN113326267A

  • Attention mechanism-based base data original address and standard address association method

    CN117520672A