A method for generating and matching a special address book for logistics

Abstract

The present invention provides a method for generating and matching a special address book for logistics, which relates to the technical field of logistics address services. Based on administrative divisions, combined with a list of special billing items for special addresses formulated, a special address list compiled, and historical order data, a special address book is generated by using a logistics address analysis model. After the shipper places an order, by matching the similarity between the delivery address of the shipper and the special addresses, suspected special address information can be determined, which can be used to remind the logistics order review staff, improving the order review efficiency of the logistics order review staff, enabling the logistics enterprise to process the distribution fee information at the early stage of the order, saving the communication time with the customer, and enhancing the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of logistics address services, and more specifically, to a method for generating and matching a logistics special address book. Background Art

[0002] With the booming development of the logistics industry, a large-scale and high-growth-potential logistics market system is gradually taking shape, and logistics services are becoming more and more specialized and refined; at the same time, logistics and express delivery enterprises are growing rapidly, their delivery scope is expanding day by day, and the address review work faced by logistics service personnel is increasing. Therefore, the rapid development of the logistics and express delivery business has also brought greater challenges to the review and confirmation work of delivery addresses.

[0003] Logistics address matching is a process of establishing a corresponding relationship between a textual description address and its spatial geographical location coordinates. Reliable and efficient logistics address matching can reduce the burden of the delivery review work of the logistics and express delivery business and is a guarantee for realizing refined logistics services. A method for matching addresses in the logistics business is disclosed in the prior art. This method performs word segmentation on the address text according to the place names in the address database to obtain word segmentation information, where the word segmentation information includes administrative division information and point-of-interest information. Then, it uses the administrative division information in the word segmentation information to match with the administrative divisions in the address database to obtain the administrative division to which the point of interest belongs; it uses the point-of-interest information in the word segmentation information to match with all the points of interest in the administrative division. If the same point of interest can be matched, the corresponding address coordinates are matched for the point of interest. Based on this method, the efficiency and accuracy of logistics address matching have both been improved.

[0004] However, in the actual logistics distribution process, in addition to the normal distribution after address matching, additional distribution costs may also be generated, such as warehousing fees, etc. Usually, enterprises in this part of the address area will regularly compile and add them to the review list, and the order review staff will check them when reviewing orders. However, in the actual implementation process, the compilation of special address areas will have a certain lag and one-sidedness, and the order review staff can only conduct manual comparison and cannot effectively compare the similarity of addresses, resulting in the collection of additional distribution costs only when they occur, increasing the additional communication work between the express delivery enterprise and the customer and affecting the user experience. Summary of the Invention

[0005] To solve the problems that in the logistics distribution process, special addresses cannot be matched in time, resulting in low work efficiency of the order review staff and poor customer experience, the present invention proposes a method for generating and matching a logistics special address book to realize the generation and matching of the logistics special address book.

[0006] In order to achieve the above technical effects, the technical solution of the present invention is as follows:

[0007] A method for generating and matching a special address book for logistics, comprising:

[0008] S1. Establish a basic address library for logistics administrative regions, determine special addresses in the basic address library for logistics administrative regions, and formulate a list of special billing items for special addresses;

[0009] S2. Obtain historical order data, match the historical order data with the list of special billing items for special addresses, obtain historical order data with special billing items, and use the delivery address in the historical order data as the special address of the historical order;

[0010] S3. Compile a list of special addresses, and perform data cleaning and parsing on the special addresses in the list of special addresses and the special addresses of historical orders;

[0011] S4. Establish a logistics address analysis model, input the special addresses in the list of special addresses after S3 and the special addresses of historical orders into the trained logistics address analysis model, output the address vector as the special address book, generate a special address area and activate it;

[0012] S5. Obtain the delivery address of the shipper, match it with the delivery address in the historical order data. If the match is successful and there is no corresponding special billing item for this delivery address, then execute step S6; otherwise, this delivery address is a special address, and push the special billing item of the special address;

[0013] S6. If the longitude and latitude of the delivery address fall within the range of the special address area, output the address list corresponding to the longitude and latitude to obtain the special billing item; otherwise, execute S7;

[0014] S7. Input the delivery address into the logistics address analysis model, and output the order address vector corresponding to the delivery address;

[0015] S8. Based on the logistics administrative region to which the delivery address belongs, obtain all special addresses at the same administrative division level in the logistics administrative region, calculate the cosine similarity between the order address vector and the special address vector corresponding to the special address, determine the suspected special address according to the calculation result, and output the corresponding order number and special billing item.

[0016] In this technical solution, based on the administrative region, combined with the formulated list of special billing items for special addresses, the compiled list of special addresses and historical order data, using the logistics address analysis model, a special address book is generated. After the shipper places an order, by matching the similarity between the shipper's delivery address and the special address, the suspected special address information can be determined, which can be used to remind the logistics order review staff, improving the order review efficiency of the logistics order review staff, enabling the logistics enterprise to process the distribution fee information at the early stage of the order, saving the communication time with the customer, and enhancing the user experience.

[0017] Preferably, the basic address database of logistics administrative divisions described in step S1 includes data of national five-level administrative divisions or above. The data source of national five-level administrative divisions is updated according to the division codes of the National Bureau of Statistics, and administrative divisions above national five-level can be obtained through web crawler technology. Based on historical orders, combined with the types of logistics services, special addresses in the basic address database of logistics administrative divisions are determined. The list of special billing items for special addresses includes customs declaration fees, warehousing fees, supermarket service fees, and remote area fees.

[0018] Preferably, the historical order data described in step S2 includes order numbers, order placement times, order delivery addresses, and billing items. The data cleaning and parsing of special addresses in the special address list and historical order special addresses in step S3 include: deleting non-address components in the address text, clearing invalid data with an empty "detailed address", and translating address symbols.

[0019] Preferably, the construction process of the logistics address analysis model described in step S4 is as follows:

[0020] S41. Divide addresses into 10 levels according to address elements, including: "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or sub-district", "village or community", "area", "road or street", "door number", "building number", and "POI".

[0021] S42. Set "village or community", "area", "road or street", "door number", "building number", and "POI" as the detailed address part, and set keywords of the detailed address part as the word segmentation library.

[0022] S43. Based on the basic address database of logistics administrative divisions and the word segmentation library, segment the detailed address part through the forward maximum matching algorithm and output, and label the segmented words according to the division levels of the word segmentation library.

[0023] S44. According to the address elements of "village or community", "area", "road or street", "door number", "building number", and "POI", vectorize the text of the detailed address part using TF-IDF, and save the vectorized results and the address parts of "province or municipality directly under the Central Government", "prefecture-level city", "district or county", and "township or sub-district" to the database to form a logistics address analysis model.

[0024] The special addresses in the special address list after S3 and the special address of the historical order are input into the logistics address analysis model, and the output address vector is used as the special address book, including vectors of "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", "village or community", "area section", "road or lane", "door number", "building number" and "POI", special address special fee charging items, geographical location longitude and latitude.

[0025] Preferably, in step S4, the process of generating and enabling the special address area is as follows:

[0026] S401. Obtain all special addresses in the special address book and clarify the longitude and latitude of all special addresses;

[0027] S402. Select one of the special addresses, and based on the longitude and latitude, calculate the distance d between this special address and each of the remaining special addresses, and set the distance division parameter as w2;

[0028] S403. If d <= w2, classify the special addresses that meet the distance into the special address area;

[0029] S404. Obtain other special addresses within the special address area. For each special address, calculate the distance d1 between it and each of the remaining special addresses in the special address book that have not been classified into the special address area;

[0030] S405. If d1 <= w2, classify the special addresses that meet the distance into the special address area;

[0031] S406. Repeat S404 - S405 until no new special addresses are classified into the special address area, and then execute step S407;

[0032] S407. Calculate the center point and radius of the special address area, and calculate the special address density a of the special address area based on the radius. The calculation formula for the special address density a is:

[0033] a = n / πr 2

[0034] Among them, n represents the number of special addresses in the special address area, and r represents the radius of the special address area;

[0035] S408. Set the special address area opening parameter as w3. When a is greater than w3, open the special address area.

[0036] Preferably, in step S5, the consignor's delivery address includes "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", and detailed address. The detailed address includes: "village or community", "area", "road or lane", "door number", "building number", and "POI". After obtaining the consignor's delivery address, before matching it with the delivery address in the historical order data, the delivery address information is cleaned, including deleting non-address components in the delivery address information, clearing invalid data with an empty "detailed address", and translating address symbols.

[0037] Preferably, in addition to matching the consignor's delivery address with the delivery address in the historical order data, special address keyword features can also be configured to assist in the matching. The special address keywords include: island, islet, hill, mine, customs, border, airport, wharf, port, prison, international, bonded area, tax port area, wind farm, expo, zoo, logistics park, convention center, drug rehabilitation center.

[0038] Preferably, the special address area range in step S6 is the area on the edge and inside of a circle with the center point of the special address area as the center and a radius of r.

[0039] Preferably, when determining a suspected special address based on the calculation result, a suspected special address determination parameter w4 is set. When the cosine similarity calculation result is not less than the suspected special address determination parameter w4, the order address vector corresponding to the delivery address is determined as the order address vector corresponding to the suspected special address.

[0040] Preferably, after step S8, it further includes: prompting the logistics order review staff with the suspected special address information. The logistics order review staff receives the prompt of the suspected special address on the order review interface, views the consignor's order information, then contacts the last-mile transportation staff to confirm the special charging items for the address, and adjusts the special charging items of the order according to the feedback of the last-mile transportation staff, and transmits them to the consignor. After the consignor confirms the charging items, the logistics order review staff adds this special address to the special address book.

[0041] Compared with the prior art, the beneficial effects of the technical solution of the present invention are:

[0042] The present invention proposes a method for generating and matching a logistics special address book. Based on the administrative division, combined with the special charging item list of the formulated special addresses, the compilation of the special address list, and the historical order data, a special address book is generated by using the logistics address analysis model. After the consignor places an order, by matching the similarity between the consignor's delivery address and the special address, the suspected special address information is determined, which can be used to remind the logistics order review staff, improving the order review efficiency of the logistics order review staff, enabling the logistics enterprise to process the distribution fee information at an early stage of the order, saving the communication time with the customer, and enhancing the user experience. Brief Description of the Drawings

[0043] Figure 1 It is a schematic flow chart showing the method for generating and matching a special address book for logistics proposed in Embodiment 1 of the present invention;

[0044] Figure 2 It is a schematic flow chart showing the construction of a logistics address analysis model proposed in Embodiment 1 of the present invention;

[0045] Figure 3 It is a schematic flow chart showing the process of generating and enabling a special address area proposed in Embodiment 1 of the present invention. Detailed Description of the Invention

[0046] The drawings are only for illustrative purposes and should not be construed as a limitation of this patent;

[0047] For better illustration of this embodiment, some parts of the drawings are omitted, enlarged or reduced, and do not represent the actual size;

[0048] For those skilled in the art, it is understandable that some well-known content descriptions in the drawings may be omitted.

[0049] The technical solutions of the present invention will be further described below with reference to the drawings and embodiments.

[0050] The description of the positional relationship in the drawings is only for illustrative purposes and should not be construed as a limitation of this patent;

[0051] Embodiment 1

[0052] This embodiment proposes a method for generating and matching a special address book for logistics. The schematic flow chart of this method is as Figure 1 shown, and the specific process includes:

[0053] S1. Establish a basic address library for logistics administrative regions, determine the special addresses in the basic address library for logistics administrative regions, and formulate a list of special charging items for special addresses;

[0054] The logistics administrative division basic address database includes data of the five-level administrative divisions of the country or administrative divisions above the five-level of the country. The data source of the five-level administrative divisions of the country can be updated according to the division codes of the National Bureau of Statistics. The five-level administrative divisions of the country include "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township, sub-district"; Administrative divisions above the five-level of the country can be obtained through web crawling technology, including: "village or community", "regional area", "road and lane", "doorplate", "building number plate" and "POI". This part serves as the detailed address part, which can improve the accuracy of grading addresses according to address elements. When making a special address determination, based on historical orders and combined with the logistics business type, the special addresses in the logistics administrative division basic address database are determined. In fact, it needs to be determined according to the core characteristics of special address orders and the special charging items of historical orders. The special charging item list of special addresses includes customs declaration fees, warehousing fees, supermarket service fees and remote area fees. The actual special charging items also need to be maintained according to the specific business model and synchronized with the charging items in the logistics system.

[0055] S2. Obtain historical order data, match the historical order data with the special charging item list of special addresses, obtain the historical order data with special charging items, and use the delivery address in the historical order data as the special address of the historical order;

[0056] In fact, the historical order data includes information such as order number, order placement time, order delivery address and charging items; match the historical order data according to the special charging item list of special addresses formulated in step S1, and the matched order delivery address is the special address of the historical order. In actual implementation, the receivables of logistics involve the actual charging and income of the order and need to be agreed upon with the customer. It is relatively accurate data during the order flow process and can be used as an effective data determination basis.

[0057] S3. Compile a special address list, and perform data cleaning and parsing on the special addresses in the special address list and the special addresses of historical orders;

[0058] In actual logistics services, logistics and express enterprises will regularly compile a special address list and adjust the special charging items of orders for the corresponding addresses. However, in the actual implementation process, the last-mile capacity of large-ticket logistics includes self-operated, outsourced routes, subcontracting and other complex situations. In fact, the final distribution sites cannot be effectively determined. Special addresses are actually difficult to manage manually. However, in the actual implementation process, the last-mile capacity of large-ticket logistics includes: self-operated, outsourced routes and subcontracting and other complex situations. In fact, the final distribution sites cannot be effectively determined. Special addresses are actually difficult to manage manually. When compiling the special address list, the data of historical orders will be combined for supplementation and continuous improvement.

[0059] Special address cleaning cleans and escapes invalid characters in the address text to make the address text more effective. Specifically, the data cleaning and parsing of special addresses in the special address list and special addresses of historical orders include: deleting non-address components in the address text, such as space and comma components, clearing invalid data with an empty "detailed address", and translating address symbols, such as translating "17L" to 17th floor. After data cleaning, the cleaned address is parsed through the map interface to obtain the longitude and latitude corresponding to the address and save them to the database.

[0060] S4. Establish a logistics address analysis model, input the special addresses in the special address list after S3 and the special addresses of historical orders into the trained logistics address analysis model, output the address vector as the special address book, generate the special address area and activate it;

[0061] See Figure 2 , the construction process of the logistics address analysis model is as follows:

[0062] S41. Divide the address into 10 levels according to address elements, including: "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", "village or community", "area", "road or lane", "house number", "building number" and "POI"; in the division process of this step, among them, levels 1-4, namely "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", are manually selected or obtained when parsing the longitude and latitude of the address through the map interface. Levels 5-10, namely "village or community", "area", "road or lane", "house number", "building number" and "POI", contain the detailed address part of the address. Therefore, levels 5-10 are the parts that really need to be parsed, and step S42 is executed;

[0063] S42. Set "village or community", "area", "road or lane", "house number", "building number" and "POI" as the detailed address part, and set the keywords of the detailed address part as the word segmentation library. For example, for the 5th level of "village or community", the set keywords include: village, community, villagers' committee, village committee.

[0064] S43. Based on the logistics administrative division basic address library and the word segmentation library, segment the detailed address part through the forward maximum matching algorithm and output it, and label the segmentation according to the division level of the word segmentation library;

[0065] In this step, considering that the actual amount of address data is not enough to support the neural network algorithm, according to the location of the address text, the optimized FMM (Forward Maximum Matching method) algorithm is used to parse the detailed address.

[0066] S44. According to the address elements of "village or community", "regional area", "road and street", "house number", "building number" and "POI", use TF-IDF to vectorize the text of the detailed address part, and save the vectorized result and the address parts of "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township and street" to the database to form a logistics address analysis model. The data used is all address data in the address database as of the day before the implementation; actually, the logistics administrative division basic address database of S101 and POI addresses can be supplemented based on web crawlers. The standard for word segmentation needs to match the level of address elements. For example, for Baiyun Building, 10th Floor, No. 113 Baiyun Road, the word segmentation is: level 7: Baiyun Road, level 8: No. 113, level 9: 10th Floor, POI: Baiyun Building.

[0067] The special addresses in the special address list after S3 and the special addresses of historical orders are input into the logistics address analysis model, and the output address vectors are used as the special address book, including the vectors of "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township and street" administrative divisions, "village or community", "regional area", "road and street", "house number", "building number" and "POI", address special fee charging items, and geographical location longitude and latitude.

[0068] In this embodiment, further, refer to Figure 3 , the process of generating and enabling the special address area is as follows:

[0069] S401. Obtain all special addresses in the special address book and clarify the longitude and latitude of all special addresses;

[0070] S402. Select one of the special addresses, and based on the longitude and latitude, calculate the distance d between this special address and each of the remaining special addresses. Set the distance division parameter as w2, which can be set to 300 meters;

[0071] S403. If d <= w2, classify the special addresses that meet the distance into the special address area;

[0072] S404. Obtain other special addresses within the special address area. For each special address, calculate the distance d1 between it and each of the remaining special addresses in the special address book that have not been classified into the special address area;

[0073] That is, this step is the processing done for the special addresses other than the selected special address in step 402 after calculating its distances with all other addresses and classifying them into the special address area.

[0074] S405. If d1 <= w2, classify the special addresses that meet the distance into the special address area;

[0075] Repeat S404 - S405 until no new special addresses are classified into the special address area, and then execute step S407;

[0076] S407. Calculate the center point and radius of the special address area, and calculate the special address density a of the special address area based on the radius. The calculation formula for the special address density a is:

[0077] a = n / πr 2

[0078] where n represents the number of special addresses in the special address area, and r represents the radius of the special address area;

[0079] S408. Set the special address area activation parameter as w3. When a is greater than w3, activate the special address area for the processing after step S5 and when the actual shipper places an order.

[0080] S5. Obtain the delivery address of the shipper, match it with the delivery addresses in the historical order data. If the match is successful and there is no corresponding special billing item for this delivery address, then execute step S6; otherwise, this delivery address is a special address, and push the special billing item of the special address.

[0081] S6. If the longitude and latitude of the delivery address fall within the range of the special address area, output the address list corresponding to the longitude and latitude to obtain the special billing item; otherwise, execute S7; the range of the special address area is the area of the edge and inside of the circle with the center point of the special address area as the center and the radius as r.

[0082] S7. Input the delivery address into the logistics address analysis model and output the order address vector corresponding to the delivery address;

[0083] S8. Based on the logistics administrative division to which the delivery address belongs, obtain the special address vector at the same division level in the logistics administrative division, calculate the cosine similarity between the order address vector and the special address vector, determine the suspected special address according to the calculation result, and output the corresponding order number and special billing item.

[0084] Embodiment 2

[0085] Based on the generation of the special address book in Embodiment 1, this embodiment gives an explanation for steps S5 and subsequent steps when the actual shipper places an order.

[0086] In step S5, the shipper places an order through the shipper's client. The shipper can select from the existing address book or create a new address book. The address includes administrative divisions at levels 1 to 4 and the detailed address part after the fourth level. In this embodiment, obtaining the shipper's delivery address includes "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", and the detailed address. The detailed address includes: "village or community", "area", "road or street", "house number", "building number", and "POI". After obtaining the shipper's delivery address, before matching it with the delivery address in the historical order data, the delivery address information is cleaned, including deleting non-address components in the delivery address information, clearing invalid data with an empty "detailed address", and translating address symbols. Then, the order address is matched with the historical delivery address. For orders corresponding to such addresses, if there is no special fee list, proceed to the next step. If the delivery address matches the historical order and there is a special address charge item, directly prompt the corresponding order information and the special address charge item.

[0087] In addition to matching the shipper's delivery address with the delivery address in the historical order data, special address keyword features can be configured to assist in the matching. The special address keywords include: island, islet, hill, mine, customs, border, airport, wharf, port, prison, international, bonded area, tax port area, wind farm, expo, zoo, logistics park, convention center, drug rehabilitation center.

[0088] Obtain the enabled special address areas, and use the longitude and latitude of the address to determine with the special address areas. If the longitude and latitude fall within the special area range, output the address list of the corresponding area, the special order number of each address, and the special billing item of each order. The special area range is determined based on a circle formed by the center point and radius of the special area. If the longitude and latitude of the current address fall within this circle, it is determined that this address falls within the special area range.

[0089] Here, if there are many special addresses and orders falling within the special area range, output the top 3 orders of the top 5 addresses with the most orders.

[0090] If all are matched, the shipper's delivery address is input into the logistics address analysis model to output the corresponding address vector. According to the 1-4 level administrative divisions of the address, all special addresses under the same fourth-level address are obtained. The order address vector and the special address vector are calculated through cosine similarity. The calculation process of cosine similarity will not be elaborated here. When determining a suspected special address based on the calculation result, a suspected special address determination parameter w4 is set. When the cosine similarity calculation result is not less than the suspected special address determination parameter w4, it is determined that the order address vector corresponding to the delivery address is the order address vector corresponding to the suspected special address. Among them, the configured value of w4 is adjusted according to the integrity of the actual address data, and the reference value is set to 80%. For the matching step in the special address area range, when there are many matched order addresses, the first 3 orders of the first 5 addresses with the most orders are output.

[0091] Embodiment 3

[0092] After the generation of the special address book based on Embodiment 1 and the matching process of Embodiment 2, after step S8, it further includes: prompting the logistics order review staff with suspected special address information. The logistics order review staff receives the prompt of the suspected special address on the order review interface, views the shipper's order information, then contacts the last-mile transportation staff to confirm the special charge item of the address, and adjusts the special charge item of the order according to the feedback of the last-mile transportation staff, and transmits it to the shipper. The shipper receives the change of the order charge item through the shipper client, and after confirming the charge item on the shipper client, the logistics order review staff adds this special address to the special address book, or after the shipper pays all the amounts of the order, this address will be automatically added to the special address book, and then the special address area is generated and activated.

[0093] Obviously, the above embodiments of the present invention are only examples for clearly explaining the present invention, and are not limitations on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to enumerate all implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A method for generating and matching a special address book for logistics, characterized in that, Including: S1. Establish a basic address database for logistics administrative divisions, determine special addresses in the basic address database for logistics administrative divisions, and formulate a list of special billing items for special addresses; S2. Obtain historical order data, match the historical order data with the list of special billing items for special addresses, obtain historical order data with special billing items, and use the delivery address in this historical order data as the historical order special address; S3. Compile a list of special addresses, and perform data cleaning and parsing on the special addresses in the list of special addresses and the historical order special addresses; S4. Establish a logistics address analysis model, input the special addresses in the list of special addresses after S3 and the historical order special addresses into the trained logistics address analysis model, output the address vector as the special address book, generate a special address area and activate it; S5. Obtain the delivery address of the shipper, match it with the delivery address in the historical order data. If the match is successful and there is no corresponding special billing item for this delivery address, then execute step S6; otherwise, this delivery address is a special address, and push the special billing item of the special address; S6. If the longitude and latitude of the delivery address fall within the range of the special address area, output the list of addresses corresponding to the longitude and latitude to obtain the special billing item; Otherwise, execute S7; S7. Input the delivery address into the logistics address analysis model and output the order address vector corresponding to the delivery address; S8. Based on the logistics administrative division to which the delivery address belongs, obtain the special address vectors at the same division level in the logistics administrative division, calculate the cosine similarity between the order address vector and the special address vector, determine the suspected special address according to the calculation result, and output the corresponding order number and special billing item.

2. The generation and matching method of the special address book for logistics according to claim 1, characterized in that The basic address database for logistics administrative divisions described in step S1 includes data of national five-level administrative divisions or above national five-level administrative divisions. The data source of national five-level administrative divisions is updated according to the division codes of the National Bureau of Statistics, and administrative divisions above national five-level can be obtained through web crawling technology; based on historical orders, combined with the types of logistics services, determine the special addresses in the basic address database for logistics administrative divisions. The list of special billing items for special addresses includes customs declaration fees, warehousing fees, supermarket service fees, and remote area fees.

3. The generation and matching method of the special address book for logistics according to claim 1, characterized in that The historical order data described in step S2 includes order number, order placement time, order delivery address, and billing items; the data cleaning and parsing of the special addresses in the list of special addresses and the historical order special addresses described in step S3 include: deleting non-address components in the address text, clearing invalid data with an empty "detailed address", and translating address symbols.

4. The method for generating and matching a special address book for logistics according to claim 3, characterized in that The construction process of the logistics address analysis model described in step S4 is as follows: S41. Divide addresses into 10 levels according to address elements, including: "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", "village or community", "area", "road or lane", "door number", "building number", and "POI"; S42. Set "village or community", "area section", "road and street", "house number", "building number" and "POI" as the detailed address part, and set the keywords of the detailed address part as the word segmentation library; S43. Based on the logistics administrative division basic address library and the word segmentation library, segment the detailed address part through the forward maximum matching algorithm and output it, and label the segments according to the administrative division level of the word segmentation library; S44. According to the address elements of "village or community", "area section", "road and street", "house number", "building number" and "POI", use TF-IDF to vectorize the text of the detailed address part, and save the vectorized result and the address parts of "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township and street" to the database to form a logistics address analysis model; The special addresses in the special address list after S3 and the special addresses of historical orders are input into the logistics address analysis model, and the output address vectors are used as the special address book, including the vectors of "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township and street" administrative divisions, "village or community", "area section", "road and street", "house number", "building number" and "POI", the special address fee charging items, and the geographical longitude and latitude.

5. The generation and matching method of the special address book for logistics according to claim 4, characterized in that, In step S4, the process of generating and enabling the special address area is as follows: S401. Obtain all the special addresses in the special address book and clarify the longitude and latitude of all the special addresses; S402. Select one of the special addresses, and based on the longitude and latitude, calculate the distance d between this special address and each of the remaining special addresses, and set the distance division parameter as w2; S403. If d <= w2, classify the special addresses that meet the distance into the special address area; S404. Obtain other special addresses within the special address area. For each special address, calculate the distance d1 between it and each of the remaining special addresses in the special address book that have not been classified into the special address area; S405. If d1 <= w2, classify the special addresses that meet the distance into the special address area; S406. Repeat S404 - S405 until no new special addresses are classified into the special address area, and execute step S407; S407. Calculate the center point and radius of the special address area, and calculate the special address density a of the special address area based on the radius. The calculation formula for the special address density a is: a = n / πr 2 where n represents the number of special addresses in the special address area, and r represents the radius of the special address area; S408. Set the special address area enabling parameter as w3. When a is greater than w3, enable the special address area.

6. The generation and matching method of the special address book for logistics according to claim 5, characterized in that, In step S5, the consignor's delivery address includes "province or municipality directly under the Central Government", "prefecture-level city", "district or county", "township or street", and detailed address. The detailed address includes: "village or community", "area section", "road or lane", "house number", "building number", and "POI". After obtaining the consignor's delivery address, before matching it with the delivery address in the historical order data, the delivery address information is cleaned, including deleting non-address components in the delivery address information, clearing invalid data with an empty "detailed address", and translating address symbols.

7. The generation and matching method of the special address book for logistics according to claim 5, characterized in that In addition to matching the consignor's delivery address with the delivery address in the historical order data, special address keyword features can also be configured to assist in the matching. The special address keywords include: island, islet, hill, mine, customs, border, airport, wharf, port, prison, international, bonded area, tax port area, wind farm, expo, zoo, logistics park, convention center, drug rehabilitation center.

8. The method for generating and matching a special address book for logistics according to claim 6 or 7, characterized in that, The special address area range described in step S6 is the area at the edge and inside of a circle with the center point of the special address area as the center and a radius of r.

9. The generation and matching method of the special address book for logistics according to claim 8, wherein, In step S8, when determining a suspected special address based on the calculation result, a suspected special address determination parameter w4 is set. When the cosine similarity calculation result is not less than the suspected special address determination parameter w4, it is determined that the order address vector corresponding to the delivery address is the order address vector corresponding to the suspected special address.

10. The method for generating and matching the special address book for logistics according to claim 9, characterized in that, After step S8, it also includes: prompting the logistics order review staff with the suspected special address information. The logistics order review staff receives the prompt of the suspected special address on the order review interface, views the consignor's order information, then contacts the last-mile transportation staff to confirm the special charging items for the address, and adjusts the special charging items of the order according to the feedback of the last-mile transportation staff, and transmits them to the consignor. After the consignor confirms the charging items, the logistics order review staff adds this special address to the special address book.

Images