Item identification method and related equipment

Abstract

The embodiment of the invention provides an item identification method and related equipment, and relates to the technical field of computers. The method comprises the following steps: determining a project level of corresponding text content in a procurement project bidding document file, wherein the project level indicates the importance degree of the corresponding text content in the procurement project bidding document file; extracting keywords and key semantic word vectors thereof in the procurement project bidding document file; determining the level weight of the keyword according to the item level of the keyword; determining a key information vector of the procurement project bidding document file according to a key semantic word vector of the keyword and a hierarchical weight of the key semantic word vector; and according to the key information vector of the procurement project bidding document file, identifying the similarity between the procurement project bidding document file and other procurement project bidding document files.

Description

Background Art

[0002] We are currently in an era of comprehensive development of the digital industry. With the vigorous development of emerging technologies and industries, the scale of production will usher in new growth points. Information technology companies such as operators and communications companies are actively deploying industry digitalization and striving to seek new opportunities and business opportunities. Enterprises often need to find business opportunity information that meets their requirements from a large number of bidding projects. The transaction projects in the public bidding market are becoming more and more complex and diverse. The traditional reliance on simple keyword matching cannot effectively identify similar semantic content of actual projects. Summary of the invention

[0003] An embodiment of the present disclosure provides a project identification method, which includes: determining a project level at which corresponding text content in a procurement project bid document is located, wherein the project level indicates the importance of the corresponding text content in the procurement project bid document; extracting keywords and their key semantic word vectors in the procurement project bid document; determining the level weight of the keywords according to the project level at which the keywords are located; determining the key information vector of the procurement project bid document according to the key semantic word vector of the keywords and their level weight; and identifying the similarity between the procurement project bid document and other procurement project bid documents according to the key information vector of the procurement project bid document.

[0004] An embodiment of the present disclosure provides a project identification device, which includes: a determination unit, used to determine the project level at which corresponding text content in a procurement project bid document is located, and the project level indicates the importance of the corresponding text content in the procurement project bid document; an extraction unit, used to extract keywords and their key semantic word vectors in the procurement project bid document; the determination unit is also used to determine the level weight of the keyword according to the project level at which the keyword is located; the determination unit is also used to determine the key information vector of the procurement project bid document according to the key semantic word vector of the keyword and its level weight; an identification unit is used to identify the similarity between the procurement project bid document and other procurement project bid documents according to the key information vector of the procurement project bid document.

[0005] An embodiment of the present disclosure provides an electronic device, the electronic device comprising: a processor; and a memory for storing executable instructions of the processor, wherein the processor is configured to execute the method in any embodiment of the present disclosure by executing the executable instructions.

[0006] An embodiment of the present disclosure provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method in any embodiment of the present disclosure is implemented.

[0007] An embodiment of the present disclosure provides a computer program product, including a computer program. When the computer program is executed, the method in any embodiment of the present disclosure is executed. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 A schematic diagram of the structure of a system in an embodiment of the present disclosure is shown.

[0009] Figure 2 A flow chart of a project identification method in an embodiment of the present disclosure is shown.

[0010] Figure 3 A flow chart showing another project identification method in an embodiment of the present disclosure.

[0011] Figure 4 A structural block diagram of an item identification device in an embodiment of the present disclosure is shown.

[0012] Figure 5 A structural block diagram of an electronic device in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0013] Figure 1 A schematic diagram of the structure of a system in an embodiment of the present disclosure is shown. Figure 1 In the system shown, a user accesses services provided by a server 20 via a network through a client (eg, a browser or an application (APP)) installed on a terminal 10 .

[0014] The terminal 10 may also be referred to as a terminal device, a user device, a mobile station, a mobile terminal, etc. The terminal can be widely used in various scenarios, for example, device-to-device (D2D), vehicle to everything (V2X) communication, machine-type communication (MTC), Internet of Things (IOT), virtual reality, augmented reality, industrial control, automatic driving, telemedicine, smart grid, smart furniture, smart office, smart wear, smart transportation, smart city, etc. The terminal may be a mobile phone, a tablet computer, a computer with wireless transceiver function, a wearable device, a vehicle, a drone, a helicopter, an airplane, a ship, a robot, a mechanical arm, a smart home device, etc. The embodiments of the present disclosure do not limit the specific technology and specific device form adopted by the terminal.

[0015] Those skilled in the art will know that Figure 1 The number of terminals and servers in the example is only illustrative, and any number of terminals and servers may be provided according to actual needs. This disclosure embodiment does not limit this.

[0016] Under the above system architecture, an item identification method is provided in the embodiments of the present disclosure, and the method can be executed by any electronic device with information processing capabilities. In some embodiments, the item identification method provided in the embodiments of the present disclosure can be executed by a terminal of the above system architecture; in other embodiments, the item identification method provided in the embodiments of the present disclosure can be implemented by a server in the above system architecture. In some other embodiments, the item identification method provided in the embodiments of the present disclosure can be implemented by the terminal and server in the above system architecture through interaction.

[0017] Figure 2 The method provided in the embodiment may be executed by a terminal, but the present disclosure is not limited thereto. In some steps, the terminal and the server may be interactively executed and finally presented on the terminal. Figure 2 As shown, the method provided by the embodiment of the present disclosure may include the following steps.

[0018] In S210, the project level at which the corresponding text content in the procurement project bid document is located is determined, and the project level indicates the importance of the corresponding text content in the procurement project bid document.

[0019] The procurement project bid document is a document provided by the tenderer to the bidder and is necessary for bidding. Its function is to explain the nature of the goods or works to be purchased, inform the rules and procedures based on the bidding, and inform the conditions for signing the contract.

[0020] In the embodiment of the present disclosure, various project levels are pre-set, and different project levels indicate the importance of different text contents in the procurement project bid document. After obtaining the original procurement project bid document, it is parsed and processed, and the text contents therein are placed in the corresponding project level to determine that the text contents in different project levels have different importance for extracting the key information vector of the procurement project bid document.

[0021] In an exemplary embodiment, determining the project level at which the corresponding text content in the procurement project bid document is located includes: obtaining text content at five project levels, namely, the title, project overview, procurement content, bidder qualifications, and matters needing attention from the procurement project bid document to form a project level content database. The project levels of the title, project overview, procurement content, bidder qualifications, and matters needing attention are arranged in descending order.

[0022] In the following embodiments, the text content in the procurement project bidding document is respectively placed into the five project levels of the project title, project overview, procurement content, bidder qualifications, and matters needing attention for illustration, and the project level of the title is preset to be higher than the project overview, the project level of the project overview is higher than the procurement content, and the project level of the bidder qualifications is higher than the matters needing attention. That is, the text content corresponding to the title in the procurement project bidding document is placed into the project level of the title, the text content related to the project overview in the procurement project bidding document is placed into the project level of the project overview, the text content related to the procurement content in the procurement project bidding document is placed into the project level of the procurement content, the text content related to the bidder qualifications in the procurement project bidding document is placed into the project level of the bidder qualifications, and the text content related to matters needing attention in the procurement project bidding document is placed into the project level of the matters needing attention.

[0023] Project overview refers to the first brief and comprehensive introduction of the basic situation of a project when introducing a project. For example, for an investment promotion project, it is more comprehensive than the general project overview, including project construction content, construction scale, total investment, market prospects, economic benefits, social benefits, geographical location, transportation conditions, climate environment, human environment, preferential policies, etc.

[0024] The procurement content in the tender document refers to the specific information of the goods, works or services to be purchased that is clearly listed during the bidding process. The procurement content includes a detailed description of the required products, services or works, specifications, quantities, technical requirements and other relevant requirements. In addition, the procurement content will also specify the scope, standards, technical requirements and qualifications that suppliers need to meet.

[0025] Bidder qualifications refer to the conditions that a bidder must meet to bid for a specific project. Precautions are written instructions to the tenderer to remind the bidder to answer the specific precautions in a comprehensive and correct manner in the bid.

[0026] In S220, keywords and key semantic word vectors in the procurement project bid document are extracted.

[0027] In an exemplary embodiment, extracting keywords and their key semantic word vectors in the procurement project bid document includes: performing word segmentation on the procurement project bid document to form a word segmentation result of the procurement project bid document; inputting the word segmentation result into a natural language pre-training model to obtain a semantic word vector of the word segmentation result; performing stop word removal on the word segmentation result of the project to obtain noun-based vocabulary in the procurement project bid document; processing the vocabulary using a keyword extraction graphical model to extract n keywords from the procurement project bid document, where n is a positive integer greater than or equal to 1, and the word segmentation result includes n keywords; and extracting the key semantic word vector of the keyword from the semantic word vector of the word segmentation result.

[0028] In the embodiments of the present disclosure, the natural language pre-training model refers to a machine learning model for natural language processing (NLP) pre-training. In the following embodiments, the BERT (Bidirectional Encoder Representations from Transformers, a machine learning technology based on transformers) model is used as an example, but the present disclosure is not limited to this.

[0029] BERT uses MLM (Machine Learning Modeling) to pre-train bidirectional Transformers to generate deep bidirectional language representations. After pre-training, only an additional output layer needs to be added for fine-tuning to achieve state-of-the-art performance in a variety of downstream tasks. In this process, no task-specific structural modifications are required to BERT.

[0030] In the embodiments of the present disclosure, the keyword extraction graphical model refers to a graphical model that can extract keywords from a text. In the following embodiments, TextRank is used as an example, but the present disclosure is not limited thereto.

[0031] TextRank is a graphical model for text summarization and keyword extraction. It can sort the words and sentences in a document according to their importance, and thus solve the tasks of keyword extraction and text summarization.

[0032] In S230, the hierarchical weight of the keyword is determined according to the project level at which the keyword is located.

[0033] In an exemplary embodiment, the hierarchical weight of the keyword is determined according to the project level where the keyword is located, including: obtaining the full text length of the procurement project bid document and the text length of the project level where the keyword is located; and obtaining the hierarchical weight of the keyword according to the full text length and the text length.

[0034] In some embodiments, it is assumed that n keywords are obtained from the procurement project bid document. The full text length |D| of the procurement project bid document can be obtained, and the text length L of the project level where each keyword is located can be obtained. Then, the hierarchical weight of the project level where each keyword is located can be calculated according to the number of keywords n, the full text length |D|, and the text length L of the project level where each keyword is located. Among them, the smaller the number of keywords n, the greater the ratio of the full text length to the text length L of the project level where the keyword is located, the greater the hierarchical weight of the keyword at the project level; conversely, the smaller. For different keywords, the shorter the text length of the project level where they are located, the greater their hierarchical weight. For the same keyword, the higher the project level where it is located, the greater its hierarchical weight. Or, for the same keyword, the shorter the text length of the project level where it is located, the greater its hierarchical weight.

[0035] For example, there are keywords A and B, keyword A is at the item level of title, and keyword B is at the item level of project overview, then the level weight of keyword A is greater than the level weight of keyword B. For another example, if keyword A is at both the item level of title and the item level of project overview, then the level weight of keyword A at the item level of title is greater than the level weight of keyword A at the item level of project overview.

[0036] In some embodiments, the hierarchical weight K of each keyword may be calculated according to the following formula:

[0037]

[0038] It is understandable that the formula for calculating the hierarchical weight K of each keyword is not limited to the above example. For example, it can also be calculated by the following formula:

[0039]

[0040] In an exemplary embodiment, determining the hierarchical weight of the keyword according to the project level where the keyword is located includes: determining the project level where the keyword is located; determining the highest project level where the keyword is located according to the level of the project level; obtaining the full text length of the procurement project bid document and the text length of the highest project level where the keyword is located; and obtaining the hierarchical weight of the keyword according to the full text length and the text length. The shorter the text of the project level where the keyword is located, the greater the hierarchical weight of the keyword.

[0041] In other embodiments, after determining the project level of each keyword, it is further determined whether the same keyword is in multiple (two or more) project levels at the same time. If the same keyword is in multiple project levels at the same time, the level weight of the keyword is calculated based on the highest project level that the keyword is in. For example, if keyword A is in both the title and project overview project levels, the level weight of keyword A is calculated based on the title project level.

[0042] In an exemplary embodiment, the hierarchical weight K(i) of the i-th keyword is obtained according to the following formula:

[0043]

[0044] Wherein, i is a positive integer greater than or equal to 1 and less than or equal to n; n is the number of keywords extracted from the procurement project bid document, and n is a positive integer greater than or equal to 1; |D| represents the full text length; |L(i)| represents the text length of the highest project level where the keyword is located.

[0045] It is understandable that the formula for calculating the hierarchical weight K of each keyword is not limited to the above example. For example, it can also be calculated by the following formula:

[0046]

[0047] In S240, the key information vector of the procurement project bid document is determined according to the key semantic word vector of the keyword and its hierarchical weight.

[0048] In some embodiments, if the hierarchical weight of the keywords at each project level is calculated based on the full text length of the procurement project bid document and the text length of the project level where the keywords are located, the key information vectors of each project level are calculated respectively, and then the key information vectors of each project level are accumulated to obtain the key information vector of the procurement project bid document. When calculating the key information vectors of each project level, the key semantic word vectors of the keywords at the project level and the hierarchical weight of the keywords at the project level are weighted and summed.

[0049] In other embodiments, if the hierarchical weight of the keywords at each project level is calculated based on the full text length of the procurement project bid document and the text length of the highest project level where the keywords are located, the key semantic word vector of the keywords in the procurement project bid document and the hierarchical weight of the keywords in the highest project level are weightedly summed to obtain the key information vector of the procurement project bid document.

[0050] In S250, the similarity between the procurement project bid document and other procurement project bid documents is identified based on the key information vector of the procurement project bid document.

[0051] In the embodiment of the present disclosure, the key information vectors of other procurement project bid documents can be calculated in a similar manner to the key information vectors of the procurement project bid document, and then the similarities between different procurement project bid documents are calculated based on the key information vectors of different procurement project bid documents. The embodiment of the present disclosure does not limit the calculation method of the similarity. In the following embodiment, the cosine distance is taken as an example, but the present disclosure is not limited thereto.

[0052] The project identification method provided by the embodiments of the present disclosure, on the one hand, identifies the similarity between different procurement project bid documents by extracting the key semantic word vectors of the keywords in the procurement project bid document, instead of using the full text of the procurement project bid document for identification, thereby reducing the amount of data calculation without affecting the accuracy of the identification result; on the other hand, not only the key semantic word vectors of the keywords are considered, but also the hierarchical weight of the keywords in the procurement project bid document is comprehensively considered, that is, the importance of different parts of the text content in the procurement project bid document is distinguished, thereby more accurately identifying the accuracy of the procurement project bid document.

[0053] Combine the following Figure 3 The method provided in the embodiment of the present disclosure is illustrated by way of example. Figure 3 The method provided in the embodiment integrates both key semantics (such as the key semantic word vector of the keyword) and hierarchical information (such as the hierarchical weight of the keyword). Therefore, the method provided in the embodiment of the present disclosure can also be considered as a project identification method integrating key semantics and hierarchical information. Figure 3 As shown, the method provided by the embodiment of the present disclosure may include the following steps:

[0054] In S301, the original bidding project documents are obtained through the bidding website.

[0055] In S302, a full-text hierarchical database is created.

[0056] The method provided by the embodiment of the present disclosure is mainly aimed at identifying similar projects in bidding and procurement content. The original bidding project documents (i.e., the original procurement project bid documents) are obtained through the bidding website, and five levels of text content, including project title, project overview, procurement content, bidder qualifications, and precautions, are sorted out from the full text of the original bidding project documents (i.e., different text contents are divided into different project levels), among which the title level is the highest and the precautions level is the lowest. The higher the level, the more refined the project content is, thereby forming a project level content database (i.e., a project full-text level database).

[0057] In S303, the full text is preprocessed: Chinese word segmentation, stop word removal, and noun vocabulary extraction.

[0058] In S304, the project keywords are extracted using the TextRank algorithm.

[0059] In S305, a keyword list of the project content is formed.

[0060] In the disclosed embodiment, Chinese word segmentation and stop word removal are performed on the full text in the database formed in S302, noun-like words in the project are retained, a co-occurrence matrix of words before and after the text is constructed, and n keywords are extracted using the TextRank algorithm to form a project keyword list.

[0061] In S306, the project word vector weight matrix is ​​trained and updated using the Chinese pre-trained BERT.

[0062] Among them, S304 and S306 can be executed independently and in parallel.

[0063] In S307, a spatial representation V(i) of the word vector corresponding to keyword i is formed.

[0064] In the disclosed embodiment, the word vector weight matrix is ​​trained and updated by pre-training the BERT word vector model, taking the word segmentation result of the full text of the project as input, and obtaining the spatial representation V(i) of the word vector corresponding to the i-th keyword among the n keywords. That is, the word segmentation result of the full text is first input into the BERT model, and the semantic word vector of each word segmentation result is obtained by output. The extracted n keywords are included in the word segmentation result of the full text. Therefore, the semantic word vector of each keyword can be extracted from the word segmentation result of the full text according to the project keyword list. Since the semantic word vector of each keyword represents the key information of the project, its semantic word vector is called the key semantic word vector.

[0065] Step 3: Use the pre-trained BERT word vector model to take the full-text word segmentation results as input, train and update the word vector weight matrix, and obtain the spatial representation of the word vector corresponding to the i-th keyword. or V(i).

[0066] Among them, the more similar the semantic environments of two words are, the closer their spatial vectors (i.e., the spatial representation of word vectors) are.

[0067] In S308, the highest item level position where the keyword is located is located in the database.

[0068] In S309, the hierarchical weight corresponding to the keyword i is calculated.

[0069] In the disclosed embodiment, the hierarchical weight of the keyword is calculated using the above formula (3) or (4) according to the hierarchical situation of the keyword in the project hierarchical content database, combined with the full text length |D| and the text length |L(i)| of the highest level where the keyword is located. The shorter the text of the level where the keyword is located, the more important the representative word information is.

[0070] In S310, the keyword vector (ie, the key semantic word vector) and the hierarchical weight information (ie, the hierarchical weight of the keyword) are fused to obtain a key information vector representation of the project content (ie, the key information vector).

[0071] In the embodiment of the present disclosure, the keyword vector and the level weight information are integrated, and the vectorized representation of the key information of the project content is obtained by multiplying the vector matrix / word vector matrix composed of the level weight and the keyword vector. For example, it can be calculated by the following formula:

[0072]

[0073] In the disclosed embodiment, n may be initially set to 5, and may be adjusted later according to the actual matching effect. For example, if the actual matching result of similar items does not meet the accuracy requirement, the value of n may be further increased.

[0074] In S311, the similarity between the contents of any two procurement items is calculated by using the cosine distance, and the closer to 1, the more similar they are.

[0075] In the embodiment of the present disclosure, the similarity between any two procurement project contents is calculated by cosine distance. For example, the similarity F(A, B) between procurement project bid document A and procurement project bid document B can be calculated by the following formula:

[0076]

[0077] In the above formula, The key information vector representation of the procurement project bidding document A is shown in Figure 1. The key information vector representation of procurement project bid document B is shown. The closer the F value is to 1, the more similar the two procurement project bid documents are.

[0078] Based on word vectors, the disclosed embodiment integrates the hierarchical position information (i.e., project level) in the text where the keywords are located, which is more in line with the characteristics that the importance of information in different paragraphs in the project bid text (i.e., procurement project bid file) is different. The calculation of the hierarchical weight of the keywords provides a reliable reference for measuring the importance of information in the text.

[0079] The disclosed embodiment utilizes the product of the hierarchical weight and the word vector matrix, and by adjusting the number of keywords, effectively integrates the key semantics and hierarchical information in the project, which is better than directly using keyword vectors for similarity matching and improves the accuracy of project similarity matching.

[0080] The disclosed embodiment provides a method for constructing a full-text hierarchical database of procurement projects (i.e., a project-level content database) and a project keyword list in the case of massive bidding text data, and effectively vectorizing similar procurement contents in massive projects by training the word vector space of the projects and calculating the hierarchical weight information of the keywords, thereby facilitating further text mining.

[0081] In some embodiments, the pre-trained word vector model (such as the BERT model) can be optimized and adjusted based on a large amount of project information for project information of different industries, such as government affairs and communications industries, to establish a word library and word vector space for industry projects as the key semantic environment of the industry, and then to mine similar project matches in special industries. For example, BERT models are trained for different industries respectively, and when matching similar projects for a certain industry, the BERT model corresponding to the industry is called to output semantic word vectors.

[0082] The disclosed embodiment is based on the full-text information in the bidding document (i.e., the procurement project bid document), uses the TextRank algorithm to extract text keywords, and obtains the spatial representation of each keyword through the BERT word vector model, which is better than directly training the word vector model for the original full text, and improves the accuracy of the key semantic representation.

[0083] The method provided in the embodiment of the present disclosure calculates the hierarchical weight of the keyword based on the hierarchical situation of the keyword in the project hierarchical content database, combined with the full text length and the length of the highest level text where the keyword is located, obtains the key information vector representation of the project content, and calculates the similarity of any two procurement project contents through cosine distance, which can quickly, conveniently and accurately identify data sets with similar procurement project contents. The present disclosure relates to matching and identifying similar projects in bidding and procurement content, and more specifically, to a project identification method based on key semantics and hierarchical information, which belongs to the field of emerging information technology.

[0084] There are few methods for matching similar content in bidding projects in the related art. Text similarity mining is mainly based on methods such as keywords and text vectors, lacking the extraction and matching of semantic information of key project content. The disclosed embodiment proposes a method that combines bidding text keyword extraction, semantic vector space and text hierarchical information to filter and extract project content in bidding documents, and vectorizes the key information of the proposed project content. It also uses a similarity matching calculation method to quickly, conveniently and accurately identify data sets with similar project content.

[0085] Figure 4 The item identification device 400 provided in the embodiment includes a determination unit 410 , an extraction unit 420 and an identification unit 430 .

[0086] The determination unit 410 is used to determine the project level at which the corresponding text content in the procurement project bid document is located, and the project level indicates the importance of the corresponding text content in the procurement project bid document.

[0087] The extraction unit 420 is used to extract keywords and their key semantic word vectors in the procurement project bid document.

[0088] The determination unit 410 is further configured to determine the level weight of the keyword according to the project level at which the keyword is located.

[0089] The determination unit 410 is further configured to determine the key information vector of the procurement project bid document according to the key semantic word vector of the keyword and its hierarchical weight.

[0090] The identification unit 430 is used to identify the similarity between the procurement project bid document and other procurement project bid documents according to the key information vector of the procurement project bid document.

[0091] In an exemplary embodiment, the determination unit 410 is further used to: obtain the full text length of the procurement project bid document and the text length of the project level where the keyword is located; and obtain the level weight of the keyword based on the full text length and the text length.

[0092] In an exemplary embodiment, the determination unit 410 is also used to: determine the project level at which the keyword is located; determine the highest project level at which the keyword is located based on the level of the project level; obtain the full text length of the procurement project bid document and the text length of the highest project level at which the keyword is located; obtain the level weight of the keyword based on the full text length and the text length; wherein, the shorter the text of the project level at which the keyword is located, the greater the level weight of the keyword.

[0093] In an exemplary embodiment, the determination unit 410 is further configured to obtain the hierarchical weight K(i) of the i-th keyword according to the following formula:

[0094]

[0095] Wherein, i is a positive integer greater than or equal to 1 and less than or equal to n; n is the number of keywords extracted from the procurement project bid document, and n is a positive integer greater than or equal to 1; |D| represents the full text length; |L(i)| represents the text length of the highest project level where the keyword is located.

[0096] In an exemplary embodiment, the extraction unit 420 is also used to: perform word segmentation on the procurement project bid document to form a word segmentation result of the procurement project bid document; input the word segmentation result into a natural language pre-training model to obtain a semantic word vector of the word segmentation result; remove stop words from the word segmentation result of the project to obtain noun-based vocabulary in the procurement project bid document; process the vocabulary using a keyword extraction graphical model to extract n keywords from the procurement project bid document, where n is a positive integer greater than or equal to 1, and the word segmentation result includes n keywords; and extract the key semantic word vector of the keyword from the semantic word vector of the word segmentation result.

[0097] In an exemplary embodiment, the determination unit 410 is also used to obtain text content of five project levels, namely, the project title, project overview, procurement content, bidder qualifications, and matters needing attention from the procurement project bid document to form a project level content database; wherein the project levels of title, project overview, procurement content, bidder qualifications, and matters needing attention are arranged in descending order.

[0098] Figure 4 For other contents of the embodiment, reference can be made to the above embodiment and will not be repeated here.

[0099] Furthermore, an embodiment of the present disclosure further provides an electronic device, comprising: a processor; and a memory for storing executable instructions of the processor, wherein the processor is configured to execute the method described in any embodiment of the present disclosure by executing the executable instructions.

[0100] Furthermore, an embodiment of the present disclosure also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the method described in any embodiment of the present disclosure is implemented.

[0101] Furthermore, an embodiment of the present disclosure also provides a computer program product, which includes a computer program. When the computer program is executed, the method described in any embodiment of the present disclosure is executed.

[0102] It should be noted that the above modules / units as part of the device can be executed in a computer system such as a set of computer executable instructions.

[0103] Those skilled in the art will appreciate that various aspects of the present disclosure may be implemented as systems, methods or program products. Therefore, various aspects of the present disclosure may be specifically implemented in the following forms, namely: complete hardware implementation, complete software implementation (including firmware, microcode, etc.), or a combination of hardware and software, which may be collectively referred to herein as "circuits", "modules" or "systems".

[0104] Refer to the following Figure 5 The electronic device 200 according to this embodiment of the present disclosure is described. Figure 5 The electronic device 200 shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0105] like Figure 5 As shown, the electronic device 200 is in the form of a general computing device. The components of the electronic device 200 may include but are not limited to: at least one processing unit 210, at least one storage unit 220, and a bus 230 connecting different system components (including the storage unit 220 and the processing unit 210).

[0106] The storage unit 220 stores program codes, which can be executed by the processing unit 210, so that the processing unit 210 executes the steps described in the above “Exemplary Method” section of this specification according to various exemplary embodiments of the present disclosure.

[0107] The storage unit 220 may include a readable medium in the form of a volatile storage unit, such as a random access memory unit (RAM) 2201 and / or a cache memory unit 2202 , and may further include a read-only memory unit (ROM) 2203 .

[0108] The storage unit 220 may also include a program / utility 2204 having a set (at least one) of program modules 2205, such program modules 2205 including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which or some combination may include an implementation of a network environment.

[0109] Bus 230 may represent one or more of several types of bus structures, including a memory unit bus or memory unit controller, a peripheral bus, an accelerated graphics port, a processing unit, or a local bus using any of a variety of bus architectures.

[0110] The electronic device 200 may also communicate with one or more external devices 240 (e.g., keyboards, pointing devices, Bluetooth devices, etc.), may also communicate with one or more devices that enable a user to interact with the electronic device 200, and / or communicate with any device that enables the electronic device 200 to communicate with one or more other computing devices (e.g., routers, modems, etc.). Such communication may be performed via an input / output (I / O) interface 250. Furthermore, the electronic device 200 may also communicate with one or more networks (e.g., local area networks (LANs), wide area networks (WANs), and / or public networks, such as the Internet) via a network adapter 260. As shown, the network adapter 260 communicates with other modules of the electronic device 200 via a bus 230. It should be understood that, although not shown in the figure, other hardware and / or software modules may be used in conjunction with the electronic device 200, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

Claims

1. A project identification method, characterized in that: include: Determining a project level at which corresponding text content in a procurement project bid document is located, wherein the project level indicates the importance of the corresponding text content in the procurement project bid document; Extracting keywords and key semantic word vectors from the procurement project bid document; Determining the level weight of the keyword according to the project level at which the keyword is located; Determining the key information vector of the procurement project bid document according to the key semantic word vector of the keyword and its hierarchical weight; According to the key information vector of the procurement project bid document, the similarity between the procurement project bid document and other procurement project bid documents is identified.

2. The method according to claim 1, characterized in that Determining the level weight of the keyword according to the project level where the keyword is located includes: Obtaining the full text length of the procurement project bid document and the text length of the project level where the keyword is located; The hierarchical weight of the keyword is obtained according to the full text length and the text length.

3. The method according to claim 1, characterized in that Determining the level weight of the keyword according to the project level where the keyword is located includes: Determine the project level at which the keyword is located; According to the level of the project, determine the highest project level where the keyword is located; Obtaining the full text length of the procurement project bid document and the text length of the highest project level where the keyword is located; Obtaining the hierarchical weight of the keyword according to the full text length and the text length; The shorter the text of the item level where the keyword is located, the greater the level weight of the keyword.

4. The method according to claim 3, characterized in that The hierarchical weight K(i) of the i-th keyword is obtained according to the following formula: Wherein, i is a positive integer greater than or equal to 1 and less than or equal to n; n is the number of keywords extracted from the procurement project bid document, and n is a positive integer greater than or equal to 1; |D| represents the full text length; |L(i)| represents the text length of the highest project level where the keyword is located.

5. The method according to claim 1, characterized in that Extracting keywords and key semantic word vectors from the procurement project bidding document, including: Segmenting the procurement project bid document to form a segmentation result of the procurement project bid document; Inputting the word segmentation result into a natural language pre-training model to obtain a semantic word vector of the word segmentation result; Remove stop words from the word segmentation results of the project to obtain noun-like words in the bidding document of the procurement project; Processing the vocabulary using a keyword extraction graphical model to extract n keywords from the procurement project bid document, where n is a positive integer greater than or equal to 1, and the word segmentation result includes n keywords; The key semantic word vector of the keyword is extracted from the semantic word vector of the word segmentation result.

6. The method according to claim 1, characterized in that Determine the project level of the corresponding text content in the procurement project bid document, including: Obtaining textual contents of five project levels, namely, project title, project overview, procurement content, bidder qualifications, and matters needing attention, from the procurement project bid documents to form a project level content database; Among them, the project levels of title, project overview, procurement content, bidder qualifications, and matters needing attention are arranged in descending order.

7. An item identification device, characterized in that: include: A determination unit, configured to determine a project level at which a corresponding text content in a procurement project bid document is located, wherein the project level indicates a degree of importance of the corresponding text content in the procurement project bid document; An extraction unit, used to extract keywords and key semantic word vectors from the procurement project bid document; The determining unit is further used to determine the level weight of the keyword according to the project level where the keyword is located; The determining unit is further used to determine the key information vector of the procurement project bid document according to the key semantic word vector of the keyword and its hierarchical weight; The identification unit is used to identify the similarity between the procurement project bidding document and other procurement project bidding documents according to the key information vector of the procurement project bidding document.

8. An electronic device, characterized in that: include: processor; as well as A memory, configured to store executable instructions of the processor; The processor is configured to perform the method of any one of claims 1 to 6 by executing the executable instructions.

9. 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.

10. A computer program product, comprising a computer program, which executes the method according to any one of claims 1 to 6 when the computer program is executed.

Images