Series connection method, device and equipment of building beacon information and readable storage medium

Abstract

The invention discloses a series connection method, device and equipment for building beacon information and a readable storage medium, and the method comprises the steps: obtaining batch building beacon information which comprises bid invitation beacon information and bid winning beacon information; the building beacon information is processed based on a preset coding model, beacon vectors are generated, the beacon vectors are compressed and stored in a database, and the beacon vectors comprise bid invitation data vectors and bid winning data vectors; in response to an association instruction of the bid information input by the front end, analyzing the association instruction to obtain to-be-associated bid winning information; matching the to-be-associated bid-winning information with the bid vector to obtain a target bid-winning data vector and a candidate bid invitation data vector; and determining a target bid invitation data vector according to the similarity between the target bid-winning data vector and the candidate bid invitation data vector, and storing a matching relationship between the target bid-winning data vector and the target bid invitation data vector into a database. According to the invention, the efficiency and accuracy of building beacon information series connection are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method, device, equipment and readable storage medium for serial connection of building signal information. Background Art

[0002] In the construction industry, bidding is a competitive market transaction method that aims to allocate construction projects to suitable units through selection. Among them, bidding information is an important part of bidding activities, which covers information on various links such as bidding announcements, bidding invitations, bidding documents, and bidding results. This information is usually released through announcements, websites, newspapers and other channels so that potential bidders can obtain necessary bidding information and participate in bidding competition.

[0003] In the actual production process of bidding information, it may be necessary to conduct correlation analysis on the bidding and winning bid announcements of the same project. Different bidding information announcements contain different information. After correlation, more project information in the bidding and winning process can be used in subsequent data analysis and machine learning tasks. Since the bidding information of the same project will be released separately at different bidding stages, the bidding and winning bid information of the same project released by the platform or captured by the data vendor is usually separate; although the public resource trading platforms of some provinces and cities will associate bidding information by project within the platform, most local platforms still publish them separately, and the same bidding or winning bid announcement may be published on multiple platform websites, which ultimately makes it difficult to associate information through the platform.

[0004] There is currently no effective solution to the technical problem that the bidding information of the same project is released separately in the existing technology, which makes it difficult to associate the platform information. Summary of the invention

[0005] The purpose of the present invention is to provide a method, device, equipment and readable storage medium for serializing building signal information, which can solve the technical problem of low accuracy in the prior art of recognizing tables in images based on traditional computer vision algorithms or deep neural network models.

[0006] One aspect of the present invention provides a method for serializing building bid information, the method comprising: obtaining batches of building bid information, wherein the building bid information comprises bidding bid information and winning bid information; processing the building bid information based on a preset coding model to generate a bid information vector, and compressing and storing the bid information vector in a database, wherein the bid information vector comprises a bidding data vector and a winning bid data vector; in response to an association instruction of the bid information input by a front end, parsing the association instruction to obtain the winning bid information to be associated; matching the winning bid information to be associated with the bid information vector to obtain a target winning bid data vector and a candidate bidding data vector; determining the target bidding data vector by the similarity between the target winning bid data vector and the candidate bidding data vector, and storing the matching relationship between the target winning bid data vector and the target bidding data vector in the database.

[0007] Optionally, the building bid information is processed based on a preset coding model to generate a bid vector, and the bid vector is compressed and stored in a database, including: classifying the building bid information to determine the category to which the sub-data belongs, wherein the category includes title, symbol, bid section, and tender notice type; determining a processing method for the sub-data according to the category to which the sub-data belongs, and pre-processing the sub-data according to the processing method, wherein the category and the processing method uniquely correspond; loading the preset coding model, vectorizing the pre-processed sub-data based on the preset coding model to generate a bid vector; converting the bid vector into a string, encoding and compressing the converted string through a string compression algorithm, and storing the compressed data in a database.

[0008] Optionally, the sub-data is preprocessed by a processing method, including: if the category of the sub-data is title, then the key fields of the sub-data are extracted for structuring, wherein the key fields of the sub-data include the bidding information title, release time, province, city, project type, bidding unit, bidding amount, and winning bid amount; if the category of the sub-data is symbol, the sub-data is standardized using a unified format; if the category of the sub-data is section, the section number is determined by keyword search and word segmentation, and the section number is extracted as a separate section field; if the category of the sub-data is bidding announcement type, preset invalid words are matched with the sub-data, and the successfully matched sub-data is deleted.

[0009] Optionally, the winning bid information to be associated includes the winning bid data to be matched and the winning bid conditions, and the winning bid information to be associated is matched with the bid signal vector to obtain a target winning bid data vector and a candidate bidding data vector, including: matching the winning bid data to be matched with the winning bid string vector of the bid signal vector, and decompressing the successfully matched winning bid string vector to obtain the target winning bid data vector; serializing the winning bid conditions, and filtering and merging the bidding string vector of the bid signal vector through the serialized winning bid conditions to obtain a bidding string vector that meets the conditions; and decompressing the bidding string vector that meets the conditions to obtain a candidate bidding data vector.

[0010] Optionally, the target bidding data vector is determined by the similarity between the target winning bid data vector and the candidate bidding data vector, including: extracting the target winning bid data vector and the candidate bidding data vector into matrices respectively to obtain a winning bid matrix and a bidding matrix; multiplying the winning bid matrix and the bidding matrix to obtain a similarity matrix; judging whether each element of the similarity matrix is ​​greater than a preset threshold; screening the row numbers of elements greater than the preset threshold, and obtaining a candidate high-similarity bidding data subset through the row numbers of the elements; performing similarity scoring on the candidate high-similarity bidding data subsets through the discriminant elements, and taking the candidate high-similarity bidding data with the highest score as the target bidding data vector.

[0011] Optionally, similarity scores are performed on candidate high-similarity bidding data subsets through discriminant elements, including: sorting the candidate high-similarity bidding data subsets in order of vector similarity from high to low and release time from recent to distant to obtain a bidding subset to be scored; extracting each data of the bidding subset to be scored and the target winning bid data vector in turn to calculate the score of the discriminant element, wherein the discriminant elements include vector similarity value, release time, city, project type, bidding and winning bid amount, bidding unit, and section set; using a preset multi-scheme comparison scoring algorithm to comprehensively score the scores of all judgment elements and select the highest score.

[0012] Another aspect of the present invention provides a serial connection device for building bid information, the device comprising: an acquisition module, used to acquire batch building bid information, wherein the building bid information includes bidding bid information and winning bid information; an encoding module, used to process the building bid information based on a preset encoding model, generate a bid information vector, and compress and store the bid information vector in a database, wherein the bid information vector includes a bidding data vector and a winning bid data vector; a parsing module, used to respond to the association instruction of the bid information input by the front end, parse the association instruction, and obtain the winning bid information to be associated; a matching module, used to match the winning bid information to be associated with the bid information vector, and obtain a target winning bid data vector and a candidate bidding data vector; an association module, used to determine the target bidding data vector through the similarity between the target winning bid data vector and the candidate bidding data vector, and store the matching relationship between the target winning bid data vector and the target bidding data vector in the database.

[0013] Optionally, the encoding module is also used to: classify the building bid information and determine the category of the sub-data, wherein the categories include title, symbol, section, and tender notice type; determine the processing method of the sub-data according to the category of the sub-data, and pre-process the sub-data according to the processing method, wherein the category and the processing method uniquely correspond; load a preset encoding model, vectorize the pre-processed sub-data based on the preset encoding model, and generate a bid vector; convert the bid vector into a string, encode and compress the converted string through a string compression algorithm, and store the compressed data in a database.

[0014] Another aspect of the present invention provides a computer device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method for serializing building signal information of any of the above-mentioned embodiments when executing the computer program.

[0015] Another aspect of the present invention provides a computer storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for serializing building bid information of any of the above embodiments is implemented. Further, the computer-readable storage medium may mainly include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function, etc.; the data storage area may store data created according to the use of blockchain nodes, etc.

[0016] The present invention vectorizes the bidding and winning bid information through a preset coding model to obtain a high-dimensional and high-precision floating-point numerical vector, and then compresses the vector data and stores it in a database, saving storage space. Finally, the vector similarity of the bidding and winning bid data is calculated for semantic matching, and accurate matching can be achieved by relying on only a small number of related fields. The serial connection accuracy and efficiency of building bidding information are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present invention. Moreover, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings: Figure 1 An optional flow chart of the method for serially connecting building bid information provided by the first embodiment of the present invention is shown; Figure 2 It shows a structural block diagram of a device for serially connecting building signal information provided by the second embodiment of the present invention; and Figure 3 A block diagram of a computer device suitable for implementing a serial connection method of building signal information provided by a third embodiment of the present invention is shown. DETAILED DESCRIPTION

[0018] In order to make the purpose, technical scheme and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0019] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a ..." does not exclude the existence of other identical elements in the process, method, article or device including the element.

[0020] Embodiment 1 This embodiment provides a method for serializing building signal information. Figure 1 A flow chart showing the serial connection method of the building bid information is shown in FIG. Figure 1As shown, the building signal information serial connection method may include steps S101 to S105, wherein: Step S101, obtaining batches of construction bidding information, wherein the construction bidding information includes bidding information and winning bidding information; Batch construction bidding information can be obtained from the documents published on the public resource platform website.

[0021] Step S102, processing the building bidding information based on a preset coding model to generate a bidding information vector, and compressing the bidding information vector and storing it in a database, wherein the bidding information vector includes a bidding data vector and a winning bid data vector; The preset encoding model can be a BGE series model, namely the small-zh-v1.5 model and the large-zh-v1.5 model. The small-zh-v1.5 model has fewer parameters and higher precision, and has better support for Chinese. The large-zh-v1.5 model has a vector dimension of 1024. A higher dimension can store richer semantic information during the text encoding process, and can more accurately match the bid and the winning bid in the matching stage, but at the same time the encoding performance will be reduced and more storage space will be occupied.

[0022] Since direct compression of float64 high-precision floating-point values ​​will result in precision loss, it is necessary to convert the high-dimensional vector into a string type first. Convert each value in the vector into a string first, and use "," to connect each string value to convert the string vector into a vector string. Encode and compress using a string compression algorithm, and it must be lossless compression. Use the zlib module that comes with Python for compression. After compression, store the original bidder ID, compressed string vector, and bid section field in the database.

[0023] Step S103, in response to the association instruction of the bidding information input by the front end, the association instruction is parsed to obtain the winning bid information to be associated; Specifically, the winning bid information to be associated includes the winning bid data to be matched and the winning bid conditions; the winning bid data to be matched can be the construction unit ID, and the winning bid conditions can be any one or a combination of the following: bidding unit, release time, province, city, and winning bid amount.

[0024] Step S104, matching the bid winning information to be associated with the bid information vector to obtain a target bid winning data vector and a candidate bidding data vector; The database stores a large amount of bidding data in vector form. Only by traversing the database and matching the bidding information to be associated with the bidding information vector can the target vector data vector and the candidate bidding data vector be obtained.

[0025] Step S105 , determining the target bidding data vector by the similarity between the target bid winning data vector and the candidate bidding data vector, and storing the matching relationship between the target bid winning data vector and the target bidding data vector in a database.

[0026] The similarity can be vector cosine similarity. By calculating vector cosine similarity for semantic matching, the goal of accurate matching can be achieved by relying on only a small number of related fields, and the coverage and accuracy of fields such as project type, amount, city, etc. are not required to be high. At the same time, no additional labeled data is required, which improves the accuracy of bidding data matching.

[0027] This embodiment uses a small model for vectorization of large quantities of bidding and winning bid data, which improves efficiency while ensuring the semantic compression accuracy of the vector. In the matching stage, the search space of the matching process is greatly reduced by constraining the matching conditions of bidding and winning bids. At the same time, matrix multiplication calculation is used to greatly improve the correlation efficiency of bid information.

[0028] Preferably, step S102 may include steps S1021 to S1024, wherein: Step S1021, classifying the construction bidding information to determine the category to which the sub-data belongs, wherein the category includes title, symbol, bid section, and bidding notice type; Step S1022, determining a processing method for the sub-data according to the category to which the sub-data belongs, and pre-processing the sub-data according to the processing method, wherein the category and the processing method are uniquely corresponding; This embodiment can improve the accuracy of processing the bidding information by setting corresponding preprocessing methods for different contents of the bidding information. Specifically, the preprocessing methods include title preprocessing, symbol preprocessing, bid section preprocessing and bidding announcement type preprocessing.

[0029] Preferably, step S1022 may include steps A1 to A4, wherein: Step A1, if the category of the sub-data is title, extract the key fields of the sub-data for structuring, wherein the key fields of the sub-data include the title of the bidding information, the release time, the province, the city, the project type, the bidding unit, the bidding amount, and the winning bid amount; Since the actual rules formulated by each website platform are different, the published bidding information titles are non-standardized, there is no unified format template, and the text formats and grammars used are diverse. Therefore, it is necessary to preprocess the titles into a relatively standard format, so that the matching accuracy of the bidding information after vectorization is higher. In this method, the obtained bidding and winning data has been pre-stored in the database, and the bidding and winning bidding information has been partially structured, that is, the bidding information title, release time, province, city, project type, bidding unit, bidding amount, winning amount and other key fields are extracted, but the quality of extraction such as coverage and accuracy is not too high.

[0030] Step A2, if the category of the sub-data is symbol, standardize the sub-data using a unified format; The unified format can include English half-angle brackets, Arabic numerals, English lowercase letters, and no spaces. Specifically, the same symbol may be used in two forms to indicate the bidding and winning bids. It is necessary to unify the large, medium, and small brackets in the bidding and winning bid titles into English half-angle brackets, unify the Chinese numerals, circle symbol numbers, and Roman numerals into Arabic numerals, unify English uppercase letters into lowercase letters, and remove spaces in the text.

[0031] Step A3: If the category of the sub-data is a bid section, determine the bid section number through keyword search and word segmentation, and extract the bid section number as a separate bid section field; Sometimes in the construction industry, bidding information for multiple bidding sections in a project bidding stage is released in the same bidding information, but when the bid is successful, it will be published separately in multiple bidding announcements according to the multiple bidding sections. The title is usually reflected as "Bidding Section 1" or "Bidding Section 1". The bidding section number can be extracted as a separate bidding section field through keyword search and word segmentation, which is used to assist in matching and determining whether it is the same project and the same bidding section.

[0032] Step A4: if the category of the sub-data is the tender notice type, the preset invalid words are matched with the sub-data, and the sub-data that is successfully matched is deleted.

[0033] The preset invalid words may be tender notice, competitive consultation notice and tender plan.

[0034] Through experiments, it is found that the above words will greatly affect the accuracy of vector similarity. Therefore, invalid words such as "tender announcement", "competitive consultation announcement", "tender plan" that may appear at the end of the tender announcement are eliminated to improve the accuracy of the bidding information concatenation.

[0035] Step S1023, loading a preset coding model, vectorizing the preprocessed sub-data based on the preset coding model, and generating a signal vector; The fastembed library in Python is used for model loading and subsequent vectorization. The advantage is that a special acceleration algorithm is used internally to support efficient vectorized calculations under CPU conditions.

[0036] The preprocessed title is vectorized by the loaded model. The principle of vectorization is to use the Transformer model architecture at the core of the NLP large model to map and encode Chinese text into high-precision numerical vectors in high-dimensional space to compress the semantic information of the text. In this embodiment, the BGE small-zh-v1.5 model will map the title text into a 512-dimensional float64 floating-point vector. Python fastembed supports parallel calculation of multiple texts at the same time, and can also support the use of CPU efficient encoding when GPU resources are insufficient.

[0037] Step S1024, converting the signal vector into a character string, encoding and compressing the converted character string using a character string compression algorithm, and storing the compressed data in a database.

[0038] Since the values ​​in the vector are of float64 type, occupying 8 bytes, each 512-dimensional vector of data needs to occupy 512*8 / 1024=4K bytes. For tens of millions of bidding data, the storage of a single vector will take up hundreds of GB of space, so it needs to be compressed to avoid excessive occupation of system resources by bidding information.

[0039] Preferably, step S104 may include steps S1041 to S1043, wherein: Step S1041, matching the winning bid data to be matched with the winning bid character string vector of the bid information vector, and decompressing the successfully matched winning bid character string vector to obtain the target winning bid data vector; In the bidding process, the announcement of the winning bid must have a corresponding bidding announcement in most cases. Therefore, the bidding and winning bid matching process starts with the winning bid announcement to find the matching bidding announcement. In the actual operation process, the winning bid data to be matched can be extracted in batches according to the release time from early to late according to business needs, and matched with the winning bid string vector stored in the database. The python zlib library is used to decompress the winning bid string vector and restore it to the original vector string. At the same time, the string is split using "," to form a string vector. Each element is a string value, and all elements are traversed to restore them to float64 precision values. At this point, the restoration of high-dimensional and high-precision numerical vectors is completed.

[0040] Step S1042, serializing the bid winning conditions, and screening and merging the bidding character string vectors of the bid information vectors according to the serialized bid winning conditions to obtain the bidding character string vectors that meet the conditions; Serialize the winning conditions. The winning conditions are mainly based on the two dimensions of province and release time as screening. It is necessary to first make a condition set for the winning data to be matched in step S103, such as all the provinces existing in the winning data to be matched and the years and months corresponding to the provinces. For example, if the province range of the winning data to be matched is Hebei Province, Shandong Province, and Beijing City, the corresponding release time is concentrated in Hebei Province: ["2023-10", "2024-03", "2024-09"], Shandong Province: ["2024-01", "2024-03"], Beijing City: ["2023-11", "2024-04"].

[0041] The list of winning provinces and corresponding time serialized in the previous step is traversed separately to filter out the bidding data that meets the time and region. Experience shows that the winning announcement should be published within three months after the bidding announcement is released. In some provinces with many projects, the time can be appropriately extended. Therefore, the bidding data selects the bidding data of the same province three months before the winning year and month. The bidding data needs to extract the id, bidding unit, bidding amount, project type, province, city, release time, and the bidding compression quantized data and bid section data completed in step S102 through id association from the database.

[0042] After the traversal is completed, the bidding data is merged and deduplicated to form a bidding string vector that meets the conditions.

[0043] Step S1043, decompress the bidding character string vector that meets the conditions to obtain a candidate bidding data vector.

[0044] The compressed vectorized data of the bidding data extracted in the previous step is decompressed to restore a high-dimensional and high-precision numerical vector, namely, the candidate bidding data vector.

[0045] Preferably, step S105 may include steps S1051 to S1055, wherein: Step S1051, extracting the target winning bid data vector and the candidate bidding data vector into matrices respectively, to obtain a winning bid matrix and a bidding matrix; Extract the vectors corresponding to the winning bid and bidding data into matrices respectively, and use matrix operations to accelerate performance. For example, the winning bid subset data extraction matrix is ​​Mat1(N,512), and the bidding subset data extraction matrix is ​​Mat2(M,512). The matrix row numbers are the same as the index row numbers of each subset data.

[0046] Step S1052, multiplying the winning bid matrix and the bidding matrix to obtain a similarity matrix; Use the matrix multiplication operation Mat1⊙Mat2^T to obtain the Mat3(N,M) matrix, where the M elements in each row are the similarity values ​​calculated between the winning bid data in that row and the bidding subset data.

[0047] Step S1053, determining whether each element of the similarity matrix is ​​greater than a preset threshold; Step S1054, screening element row numbers greater than a preset threshold, and obtaining a candidate high-similarity bidding data subset through the element row numbers; Determine whether each element of the matrix Mat3 is greater than the similarity threshold, and obtain the discriminant matrix Mat4 through Mat3>=sharehold. The rows and columns of Mat4 are consistent with Mat3, and each element is True or False, indicating whether it meets the threshold condition. Use the python data processing package numpy and the where method to obtain the row and column numbers of all True elements of Mat4. The column number is the row number corresponding to the bidding subset data, and the candidate high-similarity bidding data subset can be screened out. Similarly, using matrix operations can greatly improve the calculation efficiency compared to multi-row traversal.

[0048] Step S1055, similarity scoring is performed on the candidate high-similarity bidding data subsets through the discriminant elements, and the candidate high-similarity bidding data with the highest score is used as the target bidding data vector.

[0049] The candidate high-similarity bidding data subset obtained in the previous step is subjected to relevant rule judgment to obtain the most likely matching bidding data.

[0050] Preferably, step S1055 may include steps B1 to B3, wherein: Step B1, sorting the candidate high-similarity bidding data subsets in the order of vector similarity from high to low and release time from recent to distant, to obtain a bidding subset to be scored; The candidate high-similarity bidding data subsets are sorted in reverse order from high to low according to vector similarity, and the release time is sorted from near to far compared to the release time of the winning bid data. Subsequently, each data in the bidding subset is traversed in turn to perform rule judgment.

[0051] Step B2, sequentially extracting each data of the tender subset to be scored and the target winning bid data vector to calculate the score of the discriminant element, wherein the discriminant element includes vector similarity value, release time, city, project type, tender and winning bid amount, tender unit, and bid section set; The scores of the discriminant elements are: the original value of vector similarity f1, the number of days between the time of the winning bid announcement and the time of the bidding announcement f2, whether the city is consistent f3, 100 points if consistent otherwise 0 point, whether the project type is consistent f4, 100 points if consistent otherwise 0 point, the difference between the bidding and winning bid amounts f5, if the difference ratio is too high, 0 point, whether the bidding units are consistent f6, 100 points if consistent otherwise 0 point, whether the bidding section meets the requirements f7, if the winning bid section is in the bidding information set, it is 100 points, otherwise 0 point.

[0052] Step B3: Use a preset multi-scheme comparison scoring algorithm to comprehensively score the scores of all judgment elements and select the highest score.

[0053] The preset multi-scheme comparison and scoring algorithm can be the Topsis optimal and worst solution. Using the multi-scheme comparison and scoring algorithm, the element values ​​f1-f7 under the candidate bidding subset data are comprehensively scored, and the bidding data with the highest score is the one that is most similar to the winning bid data. The bidding data matched by all the winning bid data are stored in the table.

[0054] This embodiment uses a preset coding model to vectorize the bidding and winning bid information to obtain a high-dimensional and high-precision floating-point numerical vector, and then compresses the vector data and stores it in the database, saving storage space. Finally, the vector similarity of the bidding and winning bid data is calculated for semantic matching, and accurate matching can be achieved by relying on only a small number of related fields. This improves the serialization accuracy and efficiency of building bid information.

[0055] Embodiment 2 The second embodiment of the present invention further provides a device for serializing building signal information, which corresponds to the method for serializing building signal information provided in the first embodiment. The corresponding technical features and technical effects are not described in detail in this embodiment, and the relevant parts can refer to the first embodiment. Specifically, Figure 2 The structural block diagram of the serial connection device of the building signal information is shown in FIG. Figure 2 As shown, the serial device 200 for serializing building signal information includes an acquisition module 201, an encoding module 202, a parsing module 203, a matching module 204 and an association module 205, wherein: An acquisition module 201 is used to acquire batches of building bidding information, wherein the building bidding information includes bidding information and winning bidding information; The encoding module 202 is connected to the acquisition module 201, and is used to process the building bidding information based on a preset encoding model, generate a bidding information vector, and compress and store the bidding information vector in a database, wherein the bidding information vector includes a bidding data vector and a winning bid data vector; The parsing module 203 is connected to the encoding module 202 and is used to respond to the association instruction of the bidding information input by the front end, parse the association instruction, and obtain the winning bid information to be associated; The matching module 204 is connected to the analysis module 203 and is used to match the bid winning information to be associated with the bid information vector to obtain the target bid winning data vector and the candidate bidding data vector; The association module 205 is connected to the matching module 204 and is used to determine the target bidding data vector according to the similarity between the target winning bid data vector and the candidate bidding data vector, and store the matching relationship between the target winning bid data vector and the target bidding data vector in the database.

[0056] Optionally, the encoding module is also used to: classify the building bid information and determine the category of the sub-data, wherein the categories include title, symbol, section, and tender notice type; determine the processing method of the sub-data according to the category of the sub-data, and pre-process the sub-data according to the processing method, wherein the category and the processing method uniquely correspond; load a preset encoding model, vectorize the pre-processed sub-data based on the preset encoding model, and generate a bid vector; convert the bid vector into a string, encode and compress the converted string through a string compression algorithm, and store the compressed data in a database.

[0057] Optionally, the encoding module is also used for: if the category of the sub-data is title, then extracting the key fields of the sub-data for structuring, wherein the key fields of the sub-data include the bidding information title, release time, province, city, project type, bidding unit, bidding amount, and winning bid amount; if the category of the sub-data is symbol, standardizing the sub-data using a unified format; if the category of the sub-data is section, determining the section number through keyword search and word segmentation, and extracting the section number as a separate section field; if the category of the sub-data is bidding announcement type, matching the preset invalid words with the sub-data, and deleting the successfully matched sub-data.

[0058] Optionally, the matching module is also used to: match the winning bid data to be matched with the winning bid string vector of the bid information vector, and decompress the successfully matched winning bid string vector to obtain a target winning bid data vector; serialize the winning bid conditions, and filter and merge the bidding string vector of the bid information vector through the serialized winning bid conditions to obtain a bidding string vector that meets the conditions; decompress the bidding string vector that meets the conditions to obtain a candidate bidding data vector.

[0059] Optionally, the association module is also used to: extract the target winning bid data vector and the candidate bidding data vector into matrices respectively to obtain a winning bid matrix and a bidding matrix; multiply the winning bid matrix and the bidding matrix to obtain a similarity matrix; determine whether each element of the similarity matrix is ​​greater than a preset threshold; filter out the element row numbers greater than the preset threshold, and obtain a candidate high-similarity bidding data subset through the element row numbers; perform similarity scoring on the candidate high-similarity bidding data subsets through the discriminant elements, and use the candidate high-similarity bidding data with the highest score as the target bidding data vector.

[0060] Optionally, the association module is also used to: sort the candidate high-similarity bidding data subsets in order of vector similarity from high to low and release time from near to far to obtain the bidding subset to be scored; extract each data of the bidding subset to be scored and the target winning bid data vector in turn to calculate the score of the discriminant element, wherein the discriminant elements include vector similarity value, release time, city, project type, bidding and winning bid amount, bidding unit, and bid section set; use a preset multi-scheme comparison scoring algorithm to comprehensively score the scores of all judgment elements and select the highest score.

[0061] Embodiment 3 Figure 3 The block diagram of a computer device suitable for implementing the serial connection method of building bid information provided by the third embodiment of the present invention is shown. In this embodiment, the computer device 300 can be a smart phone, tablet computer, laptop computer, desktop computer, rack server, blade server, tower server or cabinet server (including independent servers, or server clusters composed of multiple servers) that executes a program. Figure 3 As shown, the computer device 300 of this embodiment includes at least but is not limited to: a memory 301, a processor 302, and a network interface 303 that can be interconnected through a system bus. It should be noted that Figure 3 Only computer device 300 having components 301 - 303 is shown, but it should be understood that implementing all of the components shown is not a requirement, and more or fewer components may alternatively be implemented.

[0062] In this embodiment, the memory 303 includes at least one type of computer-readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 301 can be an internal storage unit of the computer device 300, such as a hard disk or memory of the computer device 300. In other embodiments, the memory 301 can also be an external storage device of the computer device 300, such as a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. equipped on the computer device 300. Of course, the memory 301 can also include both the internal storage unit of the computer device 300 and its external storage device. In this embodiment, the memory 301 is generally used to store the operating system and various application software installed on the computer device 300, such as the program code of the serial connection method of the building bid information, etc.

[0063] In some embodiments, the processor 302 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chips. The processor 302 is generally used to control the overall operation of the computer device 300. For example, the processor 302 executes control and processing related to data interaction or communication with the computer device 300. In this embodiment, the processor 302 is used to run the program code of the steps of the serial connection method of the building bid information stored in the memory 301.

[0064] In this embodiment, the serial connection method of the building signal information stored in the memory 301 can also be divided into one or more program modules and executed by one or more processors (processor 302 in this embodiment) to complete the present invention.

[0065] The network interface 303 may include a wireless network interface or a wired network interface, and the network interface 303 is generally used to establish a communication link between the computer device 300 and other computer devices. For example, the network interface 303 is used to connect the computer device 300 to an external terminal through a network, and to establish a data transmission channel and a communication link between the computer device 300 and the external terminal. The network can be a wireless or wired network such as an intranet, the Internet, the Global System of Mobile communication (GSM), Wideband Code Division Multiple Access (WCDMA), 4G network, 5G network, Bluetooth, Wi-Fi, etc.

[0066] Embodiment 4 The present embodiment also provides a computer-readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, disk, optical disk, server, App application store, etc., on which a computer program is stored, and when the computer program is executed by a processor, the steps of the method for serializing building bid information are implemented.

[0067] Obviously, those skilled in the art should understand that the modules or steps of the above-mentioned embodiments of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, and optionally, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order from that here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. In this way, the embodiments of the present invention are not limited to any specific combination of hardware and software.

[0068] It should be noted that the serial numbers of the embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0069] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method.

[0070] The above are only preferred embodiments of the present invention, and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for serial connection of building bid information, characterized in that: The method comprises: Acquire batches of construction bidding information, wherein the construction bidding information includes bidding information and winning bidding information; Processing the building bidding information based on a preset coding model to generate a bidding information vector, and compressing and storing the bidding information vector in a database, wherein the bidding information vector includes a bidding data vector and a winning bid data vector; In response to the association instruction of the bidding information input by the front end, the association instruction is parsed to obtain the winning bid information to be associated; Matching the bid winning information to be associated with the bid information vector to obtain a target bid winning data vector and a candidate bidding data vector; The target bidding data vector is determined by the similarity between the target bid winning data vector and the candidate bidding data vector, and the matching relationship between the target bid winning data vector and the target bidding data vector is stored in a database.

2. The method according to claim 1, characterized in that The processing of the building signal information based on a preset coding model to generate a signal vector, and compressing and storing the signal vector in a database includes: Classifying the construction bidding information to determine the category to which the sub-data belongs, wherein the category includes title, symbol, bid section, and bidding notice type; Determine a processing method for the sub-data according to the category to which the sub-data belongs, and pre-process the sub-data according to the processing method, wherein the category to which the sub-data belongs is uniquely corresponding to the processing method; Loading a preset coding model, vectorizing the preprocessed sub-data based on the preset coding model, and generating a signal vector; The signal vector is converted into a character string, the converted character string is encoded and compressed by a character string compression algorithm, and the compressed data is stored in a database.

3. The method according to claim 2, characterized in that The preprocessing of the sub-data by the processing method includes: If the category of the sub-data is title, extract the key fields of the sub-data for structuring, wherein the key fields of the sub-data include the title of the bidding information, the release time, the province, the city, the project type, the bidding unit, the bidding amount, and the winning bid amount; If the category of the sub-data is symbol, standardizing the sub-data using a unified format; If the category of the sub-data is a bid section, the bid section number is determined by keyword search and word segmentation, and the bid section number is extracted as a separate bid section field; If the category to which the sub-data belongs is a tender announcement type, a preset invalid word is matched with the sub-data, and the sub-data with successful matching is deleted.

4. The method according to claim 1, characterized in that: The to-be-associated winning bid information includes the to-be-matched winning bid data and the winning bid condition, and the matching of the to-be-associated winning bid information with the bid information vector to obtain a target winning bid data vector and a candidate bidding data vector includes: Matching the winning bid data to be matched with the winning bid character string vector of the bid information vector, and decompressing the successfully matched winning bid character string vector to obtain a target winning bid data vector; Serializing the bid winning conditions, screening and merging the bidding character string vectors of the bid information vectors according to the serialized bid winning conditions, and obtaining the bidding character string vectors that meet the conditions; The bidding character string vector that meets the conditions is decompressed to obtain a candidate bidding data vector.

5. The method according to claim 1, characterized in that The step of determining the target bidding data vector by the similarity between the target winning bid data vector and the candidate bidding data vector comprises: Extracting the target winning bid data vector and the candidate bidding data vector into matrices respectively, to obtain a winning bid matrix and a bidding matrix; Multiplying the winning bid matrix and the bidding matrix to obtain a similarity matrix; Determine whether each element of the similarity matrix is ​​greater than a preset threshold; Filtering element row numbers greater than a preset threshold, and obtaining a candidate high-similarity bidding data subset through the element row numbers; The candidate high-similarity bidding data subset is scored for similarity by using discriminant elements, and the candidate high-similarity bidding data with the highest score is used as the target bidding data vector.

6. The method according to claim 5, characterized in that The similarity scoring of the candidate high-similarity bidding data subsets by using the discriminant elements includes: Sorting the candidate high-similarity bidding data subsets in the order of vector similarity from high to low and release time from recent to distant, to obtain a bidding subset to be scored; Sequentially extract each data of the tender subset to be scored and the target winning bid data vector to calculate the score of the discriminant element, wherein the discriminant element includes vector similarity value, release time, city, project type, tender and winning bid amount, tender unit, and bid section set; The preset multi-scheme comparison scoring algorithm is used to comprehensively score the scores of all judgment elements and select the highest score.

7. A device for serial connection of building bid information, characterized in that: The device comprises: An acquisition module, used for acquiring batches of building bidding information, wherein the building bidding information includes bidding information and winning bidding information; An encoding module, used for processing the building bidding information based on a preset encoding model, generating a bidding information vector, and compressing and storing the bidding information vector in a database, wherein the bidding information vector includes a bidding data vector and a winning bid data vector; The parsing module is used to respond to the association instruction of the bidding information input by the front end, parse the association instruction, and obtain the winning bid information to be associated; A matching module, used for matching the to-be-associated winning bid information with the bid information vector to obtain a target winning bid data vector and a candidate bidding data vector; The association module is used to determine the target bidding data vector according to the similarity between the target winning bid data vector and the candidate bidding data vector, and store the matching relationship between the target winning bid data vector and the target bidding data vector in a database.

8. The device according to claim 7, characterized in that The encoding module is also used for: Classifying the construction bidding information to determine the category to which the sub-data belongs, wherein the category includes title, symbol, bid section, and bidding notice type; Determine a processing method for the sub-data according to the category to which the sub-data belongs, and pre-process the sub-data according to the processing method, wherein the category to which the sub-data belongs is uniquely corresponding to the processing method; Loading a preset coding model, vectorizing the preprocessed sub-data based on the preset coding model, and generating a signal vector; The signal vector is converted into a character string, the converted character string is encoded and compressed by a character string compression algorithm, and the compressed data is stored in a database.

9. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the method according to any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images