Method for generating bidding document, electronic equipment and computer program product

Abstract

The invention discloses a method for generating a bidding document, electronic equipment and a computer program product. The method comprises the following steps: acquiring business demand information, business process information and label clause information; based on the business demand information and the business process information, determining framework information of the bidding file; index information in the label clause information is extracted through the sequence labeling model, and the index information is used for determining historical label clause files corresponding to the label clause information; and generating the target bidding file based on the framework information and the index information. According to the invention, the technical problems of high time cost, low compiling efficiency and low quality of the to-be-labeled file in the generation method of the to-be-labeled file in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the field of bid document generation, and in particular to a bid document generation method, electronic equipment and computer program product. Background Art

[0002] Faced with fierce competition, rapid response to customer needs and other constraints, it is crucial to quickly prepare high-quality bidding documents. However, the existing bidding document preparation methods have the following shortcomings, which not only affect the competitive position of enterprises, but also reduce customer satisfaction.

[0003] (1) Insufficient customization: Existing methods for preparing bid documents often rely on preset templates that may be suitable for a variety of bid types but lack a deep understanding of specific project or client needs. This “one-size-fits-all” approach ignores the individual needs of clients, resulting in bid documents that are unable to fully demonstrate the company’s deep understanding of client needs, limiting the company’s opportunities to showcase its unique selling points and value-added services.

[0004] (2) Too much time: When faced with urgent bidding deadlines, the existing method of preparing bid documents involves a lot of manual work, from collecting information, writing documents to reviewing and revising, and each step takes time. This method is not only inefficient, but also often results in companies being unable to complete all preparations before the deadline, thus missing out on bidding opportunities.

[0005] (3) Insufficient consideration of technical scoring points: Technical scoring points are key indicators for evaluating the technical capabilities of bidders. However, the compilers may not fully understand and cover all technical scoring points. This puts the bid documents at a disadvantage when scoring, affecting the company's chances of winning the bid.

[0006] (4) Uneven quality: The quality of bidding documents often depends on the experience and ability of the person compiling them. Due to the lack of unified standards and processes, the quality of documents compiled by different people may vary greatly. This inconsistency not only affects the professional image of the company, but also leads to fluctuations in the success rate of bidding.

[0007] To address the above-mentioned problems, no effective solution has been proposed yet. Summary of the invention

[0008] The embodiments of the present invention provide a method for generating a bid document, an electronic device and a computer program product, so as to at least solve the technical problems in the related art that the method for generating a bid document has high time cost, low compilation efficiency and low quality of the bid document.

[0009] According to one embodiment of the present invention, a method for generating a bid document is provided, comprising: acquiring business requirement information, business process information and bid clause information; determining target framework information of the bid document based on the business requirement information and the business process information; extracting index information in the bid clause information using a sequence labeling model, wherein the index information is used to determine historical bid documents corresponding to the bid clause information; and generating a target bid document based on the target framework information and the index information.

[0010] Optionally, the sequence labeling model includes at least: a bidirectional long short-term memory network layer and a conditional random field network layer. Using the sequence labeling model to extract index information from the bid clause information includes: using the bidirectional long short-term memory network layer to extract the bid clause information to obtain context information; using the conditional random field network layer to label the context information to generate a label sequence; and identifying index information based on the label sequence.

[0011] Optionally, generating a target bid document based on the target framework information and the index information includes: using the index information to search in a bid document database to obtain search results, wherein the bid document database is constructed based on historical bid documents; compiling the target framework information according to the search results to obtain an initial bid document; and adjusting the initial bid document based on the business demand information to obtain a target bid document.

[0012] Optionally, determining the target framework information of the bid document based on the business requirement information and the business process information includes: preprocessing the business process information to obtain text feature information; performing semantic analysis on the text feature information using a natural language processing model to obtain scoring point information; generating initial framework information based on the business requirement information and the scoring point information; and adjusting the initial framework information to obtain target framework information.

[0013] Optionally, generating the initial framework information based on the business requirement information and the scoring point information includes: analyzing the business requirement information to determine the chapter information in the initial framework information; and allocating the scoring point information to the chapter information to obtain the initial framework information.

[0014] Optionally, the method for generating a bid document also includes: preprocessing the bid clause information to obtain text feature information; performing word segmentation on the text feature information to obtain word segmentation results; converting the word segmentation results using a word embedding model to obtain vector feature information; and inputting the vector feature information into a bidirectional long short-term memory network layer to extract context information.

[0015] Optionally, the conditional random field network layer determines a label sequence based on context information and transition probabilities between labels.

[0016] According to one embodiment of the present invention, there is also provided a device for generating a bid document, comprising: an acquisition module for acquiring business requirement information, business process information and bid clause information; a determination module for determining target framework information of the bid document based on the business requirement information and the business process information; an extraction module for extracting index information from the bid clause information using a sequence labeling model, wherein the index information is used to determine historical bid documents corresponding to the bid clause information; and a generation module for generating a target bid document based on the target framework information and the index information.

[0017] Optionally, the extraction module is also used to: extract the corresponding label information using a bidirectional long short-term memory network layer to obtain context information; annotate the context information using a conditional random field network layer to generate a label sequence; and identify index information based on the label sequence.

[0018] Optionally, the generation module is also used to: use index information to search in a bid document database to obtain search results, wherein the bid document database is constructed based on historical bid documents; compile target framework information based on the search results to obtain initial bid documents; adjust the initial bid documents based on business demand information to obtain target bid documents.

[0019] Optionally, the determination module is also used to: pre-process the business process information to obtain text feature information; use a natural language processing model to perform semantic analysis on the text feature information to obtain scoring point information; generate initial framework information based on business demand information and scoring point information; adjust the initial framework information to obtain target framework information.

[0020] Optionally, the generating module is further used to: analyze the business demand information to determine the chapter information in the initial framework information; and assign the scoring point information to the chapter information to obtain the initial framework information.

[0021] Optionally, the extraction module is also used to: pre-process the corresponding label clause information to obtain text feature information; perform word segmentation on the text feature information to obtain word segmentation results; use a word embedding model to convert the word segmentation results to obtain vector feature information; input the vector feature information into a bidirectional long short-term memory network layer to extract context information.

[0022] Optionally, the conditional random field network layer determines a label sequence based on context information and transition probabilities between labels.

[0023] According to one embodiment of the present invention, there is also provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein the above-mentioned method for generating a bid document is executed when the program is running.

[0024] According to one embodiment of the present invention, a non-volatile storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for generating a bid document when it is run.

[0025] According to one embodiment of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the above-mentioned method for generating a bid document is implemented.

[0026] In an embodiment of the present invention, a method of obtaining business requirement information, business process information and bid clause information is adopted, target framework information of the bid document is determined based on the business requirement information and business process information, and index information in the bid clause information is extracted using a sequence labeling model. Finally, a target bid document is generated based on the target framework information and the index information, thereby achieving the purpose of quickly generating a bid document and comprehensively considering the technical scoring points of the bid document, thereby achieving the technical effect of shortening the time period of the bid document and generating high-quality bid documents, thereby solving the technical problems of high time cost, low compilation efficiency and low quality of the bid document in the related art. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0028] Figure 1 is a flow chart of a method for generating a bidding document according to one embodiment of the present invention;

[0029] Figure 2 is a flow chart of another method for generating a bidding document according to one embodiment of the present invention;

[0030] Figure 3 The present invention is a structural block diagram of a device for generating a bidding document according to one embodiment of the present invention. DETAILED DESCRIPTION

[0031] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0032] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0033] According to an embodiment of the present invention, an embodiment of a method for generating a bid document is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0034] The method embodiment can be executed in an electronic device or a similar computing device including a memory and a processor. Taking running on a vehicle terminal as an example, the vehicle terminal may include one or more processors (the processor may include but is not limited to a central processing unit (CPU), a graphics processing unit (GPU), a digital signal processing (DSP) chip, a microcontroller unit (MCU), a programmable logic device (Field Programmable Gate Array, FPGA), a neural network processor (Neural-network Processor Unit, NPU), a tensor processor (Tensor Processing Unit, TPU), an artificial intelligence (Artificial Intelligence, AI) type processor, etc.) and a memory for storing data. Optionally, the above-mentioned vehicle terminal may also include a transmission device, an input and output device, and a display device for communication functions. It can be understood by those of ordinary skill in the art that the above-mentioned structural description is only for illustration and does not limit the structure of the above-mentioned vehicle terminal. For example, the vehicle terminal may also include more or fewer components than the above-mentioned structural description, or have a configuration different from the above-mentioned structural description.

[0035] The memory can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the method for generating bid documents in the embodiment of the present invention. The processor executes various functional applications and data processing by running the computer program stored in the memory, that is, the above-mentioned method for generating bid documents is realized. The memory may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory may further include a memory remotely arranged relative to the processor, and these remote memories may be connected to the mobile terminal via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0036] The transmission device is used to receive or send data via a network. The specific example of the above network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device can be a radio frequency (Radio Frequency, referred to as RF) module, which is used to communicate with the Internet wirelessly.

[0037] The display device may be, for example, a touch screen type liquid crystal display (LCD) and a touch display (also referred to as a "touch screen" or "touch display screen"). The liquid crystal display may enable a user to interact with a user interface of a mobile terminal. In some embodiments, the mobile terminal has a graphical user interface (GUI), and a user may interact with the GUI by finger contacts and / or gestures on a touch-sensitive surface, wherein the human-computer interaction functions here may optionally include the following interactions: creating web pages, drawing, word processing, making electronic documents, games, video conferencing, instant messaging, sending and receiving emails, call interfaces, playing digital videos, playing digital music and / or web browsing, etc. The executable instructions for executing the above human-computer interaction functions are configured / stored in a computer program product or readable storage medium executable by one or more processors.

[0038] Figure 1 is a flow chart of a method for generating a bid document according to one embodiment of the present invention. Figure 1 As shown, the method comprises the following steps:

[0039] Step S10, obtaining business demand information, business process information and bidding terms information;

[0040] Step S12, determining the target framework information of the bidding document based on the business demand information and the business process information;

[0041] In step S12, the target framework information is used to represent the chapter outline information of the bidding document.

[0042] Step S14, extracting index information from the bidding clause information using a sequence labeling model, wherein the index information is used to determine the historical bidding documents corresponding to the bidding clause information;

[0043] In step S14, the index information is used to represent the scoring key words in the bidding clause information. The index information can be used to index relevant documents in the bidding database, where the bidding database stores historical bidding documents.

[0044] Step S16, generating a target bidding document based on the target framework information and index information.

[0045] Specifically, analyze business demand information and quickly sort out business processes to clarify link scenarios and quickly locate key scoring points. For example, if the customer's bidding documents require the provision of goods delivery management instructions, how to ensure delivery, and how to deal with bad weather, the key scoring point should be "bad" weather response measures;

[0046] Furthermore, the initial framework information of the bidding document is generated based on the business demand information and key scoring point information, and the staff adjusts it based on the initial framework information to obtain the target framework information. For example, the customer's technical requirements require an introduction to the company profile, management structure, delivery guarantee measures, etc., and a response outline is listed according to the customer's requirements.

[0047] Specifically, the sequence labeling model is used to extract the index information in the bidding terms information, that is, to extract the key words of the points, and use the index information to index the relevant documents in the bidding database. For example, if the customer requires the provision of 25 vehicles, then "25 vehicles" is the keyword. The relevant documents searched based on the index information are filled in the target framework to generate the initial bidding documents; the staff adjusts the initial bidding documents according to the business demand information to obtain the target bidding documents.

[0048] Based on the above steps S10 to S16, the business requirement information, business process information and bid clause information are obtained, the target framework information of the bid document is determined based on the business requirement information and the business process information, and the index information in the bid clause information is extracted using the sequence labeling model. Finally, the target bid document is generated based on the target framework information and the index information, thereby achieving the purpose of quickly generating the bid document and comprehensively considering the technical scoring points of the bid document, thereby achieving the technical effect of shortening the time cycle of the bid document and generating high-quality bid documents, thereby solving the technical problems of high time cost, low compilation efficiency and low quality of the bid document generation method in the related technology.

[0049] Optionally, the sequence labeling model at least includes: a bidirectional long short-term memory network layer and a conditional random field network layer. In step 14, extracting index information from the bidding terms information using the sequence labeling model includes:

[0050] Step S141, extracting the corresponding term information using a bidirectional long short-term memory network layer to obtain context information;

[0051] In step S141, the above-mentioned bidirectional long short-term memory network layer (Bidirectional Long Short-Term Memory, BiLSTM) is a special recurrent neural network (RNN) that processes the bidding terms information by considering both the start and end information. In the sequence labeling task, BiLSTM can capture the contextual information in the bidding terms information so as to better understand the contextual meaning of each word or phrase in the bidding terms information.

[0052] Step S142, using the conditional random field network layer to label the context information and generate a label sequence;

[0053] Step S143: identifying index information based on the tag sequence.

[0054] In step S142, the above-mentioned conditional random fields (CRF) network layer is used to predict the output label sequence according to the context information. In sequence labeling, the CRF network layer can consider the dependency between labels and optimize the label allocation of the entire sequence to improve the overall labeling performance.

[0055] Specifically, the bidding terms information is input into the BiLSTM network layer. The BiLSTM network layer processes each word in the bidding terms and considers the relationship between the word and its context to generate context information. Once the BiLSTM generates the context information, the CRF network layer will annotate it according to the context information and the label template in the bidding terms. For example, the label template can be "technical specifications", "safety standards", "budget requirements", etc. The CRF network layer generates an optimal label sequence by learning the dependencies between labels. The system identifies specific index information based on the label sequence generated by the CRF layer. This index information will be used to determine the historical bidding documents corresponding to the bidding terms information so that it can be referenced when compiling the bidding documents. For example, it is necessary to extract information about "budget requirements" from the bidding documents. The bidding documents contain the following clause: "The budget for this project shall not exceed 1 million yuan." The BiLSTM network layer is used to process this information to understand the contextual relationship between "budget" and "1 million yuan". The CRF network layer annotates "budget" as a "keyword" and "not more than 1 million yuan" as a "specific requirement". Generate a label sequence: [Other][Keyword][Specific requirements]. According to the tag sequence, "budget" and "not to exceed 1 million yuan" are identified as index information, so based on the index information: "budget requirement: 1 million yuan", the relevant documents are indexed in the bidding database using the index information.

[0056] Based on the above steps S141 to S143, the bidirectional long short-term memory network layer is used to extract the bidding clause information to obtain context information; the conditional random field network layer is used to annotate the context information to generate a label sequence; based on the label sequence, the index information is identified, and the key index information can be automatically and accurately extracted from a large amount of bidding clause information, thereby improving the efficiency and accuracy of preparing bidding documents.

[0057] Optionally, in step S16, generating a target bidding document based on the target framework information and the index information includes:

[0058] Step S161, searching in the bidding document database using the index information to obtain search results, wherein the bidding document database is constructed based on historical bidding documents;

[0059] Step S162, compiling the target framework information according to the search results to obtain an initial bidding document;

[0060] Step S163: Adjust the initial bidding document based on the business requirement information to obtain a target bidding document.

[0061] Specifically, use the index information to search in the database to find historical documents and information fragments related to the current project. Integrate the search results into the target framework to form a preliminary draft of the response document. Make detailed adjustments and improvements to the initial response document according to the latest business requirement information. The adjusted document meets the specific requirements of the customer and is ready to be submitted.

[0062] Based on steps S161 to S163, use the index information to search in the response document database to obtain search results, where the response document database is constructed based on historical response documents; compile the target framework information according to the search results to obtain an initial response document; adjust the initial response document based on the business requirement information to obtain the target response document. The automated search and filling process can significantly reduce the time for manual document compilation, and reusing historical documents and information fragments avoids repetitive work, thus reducing the time cost of the response document generation method and improving the compilation efficiency.

[0063] Optionally, in step S12, determining the target framework information of the response document based on the business requirement information and the business process information includes:

[0064] Step S121, preprocess the business process information to obtain text feature information;

[0065] In step S121, the above-mentioned business process information includes but is not limited to all aspects such as the loading, transportation, unloading, and tracking of goods.

[0066] Specifically, use preprocessing methods such as text cleaning (removing irrelevant characters, spaces, stop words, etc.), word segmentation, and part-of-speech tagging to preprocess the business process information to obtain the text feature information of the business process information.

[0067] Specifically, remove irrelevant characters: delete non-printable characters, special symbols, HTML tags, etc. in the text. Remove spaces: delete redundant spaces in the text, including spaces at the beginning and end of the string and redundant spaces in the middle of the string. Remove stop words: delete common words that do not affect semantic understanding in the text, such as "of", "and", "is", etc. Word segmentation: recombine the continuous character sequence into a sequence of words according to certain specifications. Part-of-speech tagging: assign a part-of-speech tag to each word after word segmentation, such as noun, verb, adjective, etc. Feature extraction: convert the preprocessed text into a feature vector for subsequent semantic analysis. Feature vectorization: methods such as bag-of-words model, TF-IDF, Word2Vec, etc. can be used for vectorization.

[0068] Step S122, use a natural language processing model to perform semantic analysis on the text feature information to obtain scoring point information;

[0069] In step S122, the above-mentioned natural language processing (NLP) model includes but is not limited to a pre-trained deep learning model such as a BERT model and a GPT model for semantic analysis.

[0070] Specifically, the natural language processing model is used to analyze the semantic content of text feature information to identify key concepts, intentions, and emotions. Based on the results of semantic analysis, the scoring points that need to be focused on in the bidding documents are extracted, such as customer needs, technical advantages, cost-effectiveness, etc.

[0071] Step S123, generating initial framework information based on the business requirement information and the scoring point information;

[0072] Specifically, according to the business demand information, the specific requirements of the customer are clarified. For example, transportation time, cost, safety, tracking system, etc. The scoring point information indicates the content that needs to be highlighted in the bid document. Based on the business demand information and scoring point information, the basic outline of the bid document, such as the introduction, technical solution, project management, quotation, etc., is preliminarily determined.

[0073] Step S124, adjusting the initial framework information to obtain target framework information.

[0074] Specifically, staff members optimize and adjust the initial framework according to the importance of business needs and scoring point information to obtain the bid document framework, ensuring that it can most effectively respond to customer needs and highlight the company's competitive advantages. For example, a logistics company is preparing to bid for a new vehicle transportation project: collect business process information, such as vehicle loading, transportation routes, tracking systems, etc. Preprocess this information and extract text features, such as transportation efficiency, cost control, and safety. Use the NLP model to perform semantic analysis on the preprocessed text. The NLP model identifies key scoring points, such as "real-time tracking", "cost-effectiveness", and "safe transportation". Analyze business needs, such as customers need a logistics solution that can safely transport vehicles within a specified time and provide real-time tracking. Combined with the scoring point information, generate initial framework information, including: Introduction: introduce the company and project overview. Technical solution: Detailed description of the real-time tracking system, cost-effectiveness strategy, and safe transportation measures. Project management: Show project management methods and milestone plans. Based on the customer's high attention to real-time tracking, adjust the framework and add a special chapter on the tracking system. The final target framework information may include: Introduction: introduce the company and project overview. Technical Solution: Details the real-time tracking system, cost-effectiveness strategy, and safe transportation measures. Project Management: Shows how the project will be delivered on time through a strict project management process. Tracking System: A new chapter is added to specifically explain the working principle and customer interface of the real-time tracking system.

[0075] Based on the above steps S121 to S124, the business process information is preprocessed to obtain text feature information; the text feature information is semantically analyzed using a natural language processing model to obtain scoring point information; initial framework information is generated based on the business demand information and the scoring point information; the initial framework information is adjusted to obtain target framework information. By analyzing the specific links of the business process, the bid document can more accurately match the actual needs of the customer and increase the probability of winning the bid. The semantic analysis of the text using the NLP model can more accurately capture and respond to the implicit requirements and professional terms in the bidding documents. The systematic preprocessing and semantic analysis steps improve the quality of the bid document, making it more standardized and professional, thereby improving the quality of the preparation of the bid document.

[0076] Optionally, in step S123, generating initial framework information based on the business requirement information and the scoring point information includes:

[0077] Step S1231, analyzing the business demand information to determine the chapter information in the initial framework information;

[0078] Step S1232, assigning the score point information to the chapter information to obtain initial framework information.

[0079] Specifically, the initial framework information for generating the bid document is to combine the business requirement information and the scoring point information to create a preliminary structured document framework. Collect all the requirement documents provided by the customer, organize these documents, and extract key information and specific business requirements. Based on the business requirement information, determine the topics or chapters that the bid document needs to include. Each chapter should target a specific requirement or business area. For each chapter, define the expected content and questions that need to be answered. Determine the logical order of the chapters to ensure the fluency and coherence of the document. Create a detailed outline for each chapter, including the main titles and subtitles, and ensure that the outline covers all business requirement points.

[0080] Going a step further, from the scoring information obtained from the semantic analysis of the business process information, determine which scoring points are the most important and how they affect the success of the bid. Based on the content of the scoring points, assign them to the most relevant sections. For example, if the scoring point is "Real-time Tracking", it should be assigned to the "Real-time Tracking" section under the "Technical Details" section. Based on the scoring point information, enhance the content of the corresponding sections. Make sure that the scoring points are highlighted in the document and provide detailed explanations and instructions. Integrate all the sections and scoring point information to form a complete initial framework.

[0081] Based on steps S1231 to S1232, analyze the business requirement information to determine the chapter information in the initial framework information; allocate the scoring point information to the chapter information to obtain the initial framework information. By analyzing specific business requirements, the generated tender response documents can more accurately address the customer's concerns, ensure the logic of the document structure and content, make the document easier to understand and evaluate by reviewers, and clarify the scoring points, which helps the tender documents stand out among numerous competitors. Finally, creating a clear initial framework can speed up the subsequent document writing and revision processes, thereby improving the generation efficiency of the tender response documents.

[0082] Optionally, the method for generating a tender response document further includes:

[0083] Step S171, preprocess the tender clause information to obtain text feature information;

[0084] Specifically, preprocess the tender clause information to convert the original tender clause text into a form convenient for machine processing. The preprocessing includes the following methods: Text cleaning: Remove special characters, redundant whitespace characters, line break characters, etc. from the text. Standardization: Convert all text to a unified case, such as all lowercase, to eliminate the impact of inconsistent case. Stop word removal: Delete common and meaningless words, such as "de", "he", "shi", etc., to reduce noisy data. Stemming / lemmatization: Restore the vocabulary to its basic form, for example, convert the past tense of a verb to its original form.

[0085] Step S172, perform word segmentation on the text feature information to obtain a word segmentation result;

[0086] Specifically, perform word segmentation on the text feature information to split a continuous sequence of characters into independent units according to semantic and grammatical rules. The word segmentation processing includes the following methods: Chinese word segmentation: For Chinese text, word segmentation processing is required because there are no obvious delimiters between Chinese words. English word segmentation: For English text, the processing includes identifying and splitting compound words, etc. Part-of-speech tagging: Mark the part of speech (noun, verb, etc.) of each word.

[0087] Step S173, use a word embedding model to transform the word segmentation result to obtain vector feature information;

[0088] Specifically, use a word embedding model to transform the word segmentation result to convert words into numerical vectors that can express their semantic information. Select a word embedding model: For example, the word embedding layer of Word2Vec, GloVe, or BERT. Transformation: Convert each word into a vector of a fixed dimension through the word embedding model. These vectors capture the semantic information and context relationships of the words. Out-of-vocabulary word processing: For words not included in the word embedding model, special word vector representations can be adopted or predefined vectors can be used.

[0089] Step S174, inputting the vector feature information into a bidirectional long short-term memory network layer to extract context information.

[0090] Specifically, the vector feature information is input into the BiLSTM network layer to use the BiLSTM model to understand the context information in the text fragment.

[0091] Specifically, construct a BiLSTM model: design a neural network model containing bidirectional long short-term memory units. Input vector features: take the generated word vector sequence as input. Forward propagation: The BiLSTM model processes the input word vector sequence from two directions (forward and backward). Obtain contextual information: The BiLSTM model can generate context-related representations of each word based on the context. For example, a bidding clause is as follows: "The project must be completed before June 1, 2024." After text preprocessing, "The project must be completed before June 1, 2024" is obtained. The word segmentation result is ["project", "must", "June 1, 2024", "before", "completed"]. Each word is converted into a vector through the Word2Vec model. These vectors are input into the BiLSTM model, and the model learns the contextual importance of the date "June 1, 2024" in the sentence, providing key time point information for further bidding document generation.

[0092] Based on the above steps S171 to S174, the bid clause information is preprocessed to obtain text feature information; the text feature information is segmented to obtain segmentation results; the segmentation results are converted using a word embedding model to obtain vector feature information; the vector feature information is input into a bidirectional long short-term memory network layer to extract context information. BiLSTM can capture long-distance dependencies in the text, which is crucial for understanding complex business terms. Moreover, through part-of-speech tagging and word embedding, the model can more accurately understand the semantics and context of each word. Finally, the automated text processing process improves efficiency and reduces the time and labor intensity of manually reading and understanding a large number of bid clauses.

[0093] Optionally, the conditional random field network layer determines a label sequence based on context information and transition probabilities between labels.

[0094] Specifically, CRF is an undirected graph model used to predict the output label sequence (e.g., part-of-speech tagging, named entity recognition labels) for a given input sequence (e.g., a sequence after word segmentation). The CRF network layer takes into account the transition probabilities between labels and the characteristics of the input sequence. The CRF network layer learns the dependency information between labels, that is, the transition probabilities between labels. These probabilities reflect the possibility of label transitions between adjacent elements given an input sequence. By learning the label transition patterns in the dataset, the CRF layer can correct the output of the BiLSTM layer to ensure that the output label sequence is statistically reasonable. By combining contextual information and the transition probabilities between labels, the CRF layer is able to correct the output of other models (such as LSTM) to improve the accuracy of labeling. For example, even if the LSTM layer incorrectly outputs a high "B-LOC" score for a word, the CRF layer can correct it to the correct "B-PER" label by considering the transition probability and contextual information.

[0095] Specifically, the transition probability is used to characterize the probability of transitioning from label yi-1 to label yi given the input sequence X. In the CRF network layer, the transition probability is usually learned from the training data. The CRF network layer counts the co-occurrence frequencies between adjacent labels in the training data and uses these frequencies to estimate the transition probability. In sequence labeling tasks, such as named entity recognition (NER) and part-of-speech tagging (POS Tagging), each element (such as a word) needs to be assigned a label. There are dependencies between these labels. The transition probability can capture the dependencies between labels, thereby ensuring that the output label sequence is statistically reasonable. For example, in the NER task, the CRF layer can ensure that "B-PER" is not mistakenly followed by "B-LOC" by considering the transition probability.

[0096] Figure 2 FIG. 1 is a flowchart of another method for generating a bid document according to one embodiment of the present invention. Figure 2 As shown in the figure, the generation of bidding documents mainly includes the following execution steps:

[0097] Step S201, obtaining business demand information, business process information and bidding terms information;

[0098] Step S202, pre-processing the business process information to obtain text feature information;

[0099] Step S203, using a natural language processing model to perform semantic analysis on the text feature information to obtain score point information;

[0100] Step S204, analyzing the business demand information to determine the chapter information in the initial framework information;

[0101] Step S205, assigning the score point information to the chapter information to obtain initial framework information;

[0102] Step S206, adjusting the initial framework information to obtain target framework information.

[0103] Step S207, extracting the corresponding term information using the bidirectional long short-term memory network layer to obtain context information;

[0104] Step S208, using the conditional random field network layer to label the context information and generate a label sequence;

[0105] Step S209, identifying index information based on the tag sequence;

[0106] Step S210, searching in the bidding document database using the index information to obtain search results;

[0107] Step S211, compile the target framework information according to the search results to obtain the initial bidding document;

[0108] Step S212: adjusting the initial bidding document based on the business requirement information to obtain a target bidding document.

[0109] Based on the above steps S201 to S212, the business requirement information, business process information and bid clause information are obtained, the target framework information of the bid document is determined based on the business requirement information and the business process information, and the index information in the bid clause information is extracted using the sequence labeling model. Finally, the target bid document is generated based on the target framework information and the index information, thereby achieving the purpose of quickly generating the bid document and comprehensively considering the technical scoring points of the bid document, thereby achieving the technical effect of shortening the time cycle of the bid document and generating high-quality bid documents, thereby solving the technical problems of high time cost, low compilation efficiency and low quality of the bid document generation method in the related technology.

[0110] Through the description of the above implementation methods, those skilled in the art can clearly understand that the method according to the above embodiment can be implemented by means of software plus a necessary general hardware platform, and of course can also be implemented by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present invention, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, a magnetic disk, or an optical disk), and includes a number of instructions for enabling a terminal device (which can be a mobile phone, a computer, a server, or a network device, etc.) to execute the methods described in each embodiment of the present invention.

[0111] In the embodiments of the present invention, a device for generating a bid document is also provided, which is used to implement the above embodiments and preferred implementation modes, and the descriptions that have been made will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceivable.

[0112] Figure 3 FIG. 1 is a structural block diagram of a device for generating a bidding document according to one embodiment of the present invention. Figure 3 As shown, the device comprises:

[0113] The acquisition module 301 is used to acquire business demand information, business process information and bidding terms information;

[0114] A determination module 302, for determining target framework information of a bidding document based on business requirement information and business process information;

[0115] An extraction module 303 is used to extract index information from the bidding clause information using a sequence labeling model, wherein the index information is used to determine the historical bidding documents corresponding to the bidding clause information;

[0116] The generating module 304 is used to generate a target bidding document based on the target framework information and the index information.

[0117] Optionally, the extraction module 303 is also used to: extract the corresponding label information using a bidirectional long short-term memory network layer to obtain context information; annotate the context information using a conditional random field network layer to generate a label sequence; and identify index information based on the label sequence.

[0118] Optionally, the generation module 304 is also used to: use the index information to search in the bid document database to obtain search results, wherein the bid document database is constructed based on historical bid documents; compile the target framework information according to the search results to obtain an initial bid document; and adjust the initial bid document based on the business demand information to obtain a target bid document.

[0119] Optionally, the determination module 302 is also used to: pre-process the business process information to obtain text feature information; use a natural language processing model to perform semantic analysis on the text feature information to obtain scoring point information; generate initial framework information based on the business demand information and the scoring point information; adjust the initial framework information to obtain target framework information.

[0120] Optionally, the generating module 304 is further used to: analyze the business demand information to determine the chapter information in the initial framework information; and assign the score point information to the chapter information to obtain the initial framework information.

[0121] Optionally, the extraction module 303 is also used to: pre-process the corresponding label information to obtain text feature information; perform word segmentation on the text feature information to obtain word segmentation results; use a word embedding model to convert the word segmentation results to obtain vector feature information; input the vector feature information into a bidirectional long short-term memory network layer to extract context information.

[0122] Optionally, the conditional random field network layer determines a label sequence based on context information and transition probabilities between labels.

[0123] It should be noted that the above modules can be implemented by software or hardware. For the latter, it can be implemented in the following ways, but not limited to: the above modules are all located in the same processor; or the above modules are located in different processors in any combination.

[0124] According to one embodiment of the present invention, there is also provided an electronic device, comprising: a memory storing an executable program; and a processor for running the program, wherein the above-mentioned method for generating a bid document is executed when the program is running.

[0125] Optionally, in this embodiment, the processor may be configured to perform the following steps through a computer program:

[0126] Step S1, obtaining business demand information, business process information and bidding terms information;

[0127] Step S2, determining the target framework information of the bidding document based on the business demand information and the business process information;

[0128] Step S3, extracting index information from the bidding clause information using a sequence labeling model, wherein the index information is used to determine the historical bidding documents corresponding to the bidding clause information;

[0129] Step S4, generating a target bidding document based on the target framework information and index information.

[0130] According to one embodiment of the present invention, a non-volatile storage medium is provided, in which a computer program is stored, wherein the computer program is configured to execute the above-mentioned method for generating a bid document when it is run.

[0131] Optionally, in this embodiment, the storage medium may be configured to store a computer program for performing the following steps:

[0132] Step S1, obtaining business demand information, business process information and bidding terms information;

[0133] Step S2, determining the target framework information of the bidding document based on the business demand information and the business process information;

[0134] Step S3, extracting index information from the bidding clause information using a sequence labeling model, wherein the index information is used to determine the historical bidding documents corresponding to the bidding clause information;

[0135] Step S4, generating a target bidding document based on the target framework information and index information.

[0136] Optionally, in this embodiment, the above-mentioned storage medium may include but is not limited to: a USB flash drive, a read-only memory (ROM), a random access memory (RAM), a mobile hard disk, a magnetic disk or an optical disk, and other media that can store computer programs.

[0137] According to one embodiment of the present invention, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the above-mentioned method for generating a bid document is implemented.

[0138] Optionally, in this embodiment, the computer program product may be configured as a computer program for executing the following steps:

[0139] Step S1, obtaining business demand information, business process information and bidding terms information;

[0140] Step S2, determining the target framework information of the bidding document based on the business demand information and the business process information;

[0141] Step S3, extracting index information from the bidding clause information using a sequence labeling model, wherein the index information is used to determine the historical bidding documents corresponding to the bidding clause information;

[0142] Step S4, generating a target bidding document based on the target framework information and index information.

[0143] Optionally, the specific examples in this embodiment may refer to the examples described in the above embodiments and optional implementation modes, and this embodiment will not be described in detail here.

[0144] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0145] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0146] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0147] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0148] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0149] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0150] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for generating a bidding document, characterized in that: include: Obtain business demand information, business process information and bidding terms information; Determine target framework information of the bidding document based on the business requirement information and the business process information; Extracting index information from the bidding clause information using a sequence labeling model, wherein the index information is used to determine a historical bidding document corresponding to the bidding clause information; A target bidding document is generated based on the target framework information and the index information.

2. The method for generating a bidding document according to claim 1, characterized in that: The sequence labeling model at least includes: a bidirectional long short-term memory network layer and a conditional random field network layer. The index information extracted from the bidding terms information by using the sequence labeling model includes: Extracting the bidding terms information using the bidirectional long short-term memory network layer to obtain context information; Using the conditional random field network layer to label the context information and generate a label sequence; The index information is identified based on the tag sequence.

3. The method for generating a bidding document according to claim 1, characterized in that: Generating a target bidding document based on the target framework information and the index information includes: Using the index information to search in a bidding document database to obtain search results, wherein the bidding document database is constructed based on the historical bidding documents; Compile the target framework information according to the search results to obtain an initial bidding document; The initial bidding document is adjusted based on the business requirement information to obtain the target bidding document.

4. The method for generating a bid document according to claim 1, characterized in that: Determining the target framework information of the bidding document based on the business requirement information and the business process information includes: Preprocessing the business process information to obtain text feature information; Using a natural language processing model to perform semantic analysis on the text feature information to obtain score point information; Generate initial framework information based on the business requirement information and the scoring point information; The initial framework information is adjusted to obtain the target framework information.

5. The method for generating a bid document according to claim 4, characterized in that: Generating initial framework information based on the business requirement information and the scoring point information includes: Analyze the business requirement information to determine chapter information in the initial framework information; The score point information is allocated to the chapter information to obtain the initial framework information.

6. The method for generating a bidding document according to claim 2, characterized in that: The method further comprises: Preprocessing the bidding clause information to obtain text feature information; Performing word segmentation processing on the text feature information to obtain a word segmentation result; The word segmentation result is converted by using a word embedding model to obtain vector feature information; The vector feature information is input into the bidirectional long short-term memory network layer to extract the context information.

7. The method for generating a bid document according to claim 2, characterized in that: The conditional random field network layer determines a label sequence according to the context information and the transition probability between labels.

8. An electronic device, characterized in that: include: A memory storing an executable program; A processor is used to run the program, wherein the program, when running, executes the method for generating a bid document as described in any one of claims 1 to 7.

9. A non-volatile storage medium, characterized in that: The storage medium stores a computer program, wherein the computer program is configured to execute the method for generating a bid document as described in any one of claims 1 to 7 when running.

10. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the method for generating a bid document according to any one of claims 1 to 7.

Images