Object recommendation method, device and storage medium for realizing IA by combining RPA and AI

Abstract

The application relates to the computer technical field, in particular to an object recommendation method, device and storage medium for realizing IA by combining RPA and AI. The object recommendation method comprises the following steps: obtaining a tender document and a bid document set through a robot process automation (RPA) system; obtaining first knowledge representation information corresponding to the tender document on a first knowledge graph by adopting a target knowledge representation model, and obtaining second knowledge representation information corresponding to at least one bid document in the bid document set, wherein the first knowledge graph corresponds to the tender document and the bid document set; obtaining a matching degree set between the first knowledge representation information and the at least one second knowledge representation information by adopting a matching network model, and determining a target bid document in the at least one bid document based on the matching degree set; and recommending a target bid object corresponding to the target bid document. The application can improve the accuracy of object recommendation.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an object recommendation method, apparatus and storage medium that combines RPA and AI to implement IA. Background Technology

[0002] Robotic Process Automation (RPA) is a process that uses specific "robot software" to simulate human operations on a computer and automatically execute process tasks according to rules.

[0003] Artificial intelligence (AI) is a technical science that studies and develops theories, methods, technologies, and application systems for simulating, extending, and expanding human intelligence.

[0004] Intelligent Automation (IA) is a collective term for a series of technologies ranging from Robotic Process Automation (RPA) to Artificial Intelligence (AI). It combines RPA with various AI technologies such as Optical Character Recognition (OCR), Intelligent Character Recognition (ICR), Process Mining, Deep Learning (DL), Machine Learning (ML), Natural Language Processing (NLP), Automatic Speech Recognition (ASR), Text-to-Speech (TTS), and Computer Vision (CV) to create end-to-end business processes that can think, learn, and adapt. It covers the entire process from process discovery and automation to automatically and continuously collecting data, understanding the meaning of data, and using data to manage and optimize business processes.

[0005] With the development of science and technology and the increasing maturity of terminal technology, the convenience of users' production and life has been improved. In terminal application scenarios, terminals can recommend information based on historical user preference information or product and user-product interaction information. However, in bidding and tendering application scenarios, relying solely on historical user preference information or interaction information may lead to low accuracy in object recommendations due to a lack of consideration for bidding and tendering background knowledge. Summary of the Invention

[0006] This application provides an object recommendation method, apparatus, and storage medium that combines RPA and AI to implement IA, in order to solve the problems existing in related technologies. The technical solution is as follows:

[0007] In a first aspect, embodiments of this application provide an object recommendation method combining RPA and AI, including:

[0008] Obtain the collection of bidding documents and tender documents through a Robotic Process Automation (RPA) system;

[0009] A target knowledge representation model is used to obtain the first knowledge representation information corresponding to the bidding document on the first knowledge graph, and to obtain the second knowledge representation information corresponding to at least one bidding document in the bidding document set. The first knowledge graph corresponds to the bidding document and the bidding document set.

[0010] A matching network model is used to obtain the matching degree set between the first knowledge representation information and at least one second knowledge representation information, and the target bidding document in at least one bidding document is determined based on the matching degree set;

[0011] Recommend target bidders corresponding to the target bid documents.

[0012] Optionally, before obtaining the first knowledge representation information corresponding to the bidding document on the first knowledge graph using the target knowledge representation model, the following steps are also included:

[0013] The Robotic Process Automation (RPA) system is used to obtain a set of tender documents and bid documents in the target document format.

[0014] The triplet extraction technique is used to obtain the triplet information set corresponding to the sets of bidding documents and tender documents;

[0015] Based on the triplet information set, a first knowledge graph corresponding to the sets of bidding documents and tender documents is established.

[0016] Optionally, a triplet extraction technique is used to obtain a set of triplet information corresponding to the sets of bidding documents and tender documents, including:

[0017] The triplet extraction technique is used to obtain the first triplet information set corresponding to the bidding documents and the second triplet information set corresponding to the bid documents.

[0018] The Jieba word segmentation model is used to segment the first triplet information set and the second triplet set to obtain the processed first triplet information set and the processed second triplet set.

[0019] The processed first triplet information set and the processed second triplet set are subjected to entity boundary string verification to obtain the triplet information set corresponding to the bidding document and bid document sets.

[0020] Optionally, before obtaining the first knowledge representation information corresponding to the bidding document using the target knowledge representation model, and obtaining the second knowledge representation information corresponding to at least one bidding document in the bidding documents, the method further includes:

[0021] Obtain a collection of bidding and tendering training documents;

[0022] Obtain the knowledge graph corresponding to the set of bidding training documents;

[0023] Based on distributed representation learning reasoning technology, vector space mapping is performed on at least one knowledge pair in the knowledge graph to establish a target knowledge representation model.

[0024] Optionally, before obtaining the first knowledge representation information corresponding to the bidding document using the target knowledge representation model, and obtaining the second knowledge representation information corresponding to at least one bidding document in the bidding documents, the method further includes:

[0025] Obtain the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set;

[0026] Based on the model performance parameters, a target knowledge representation model is determined from the initial knowledge representation model set. The target knowledge representation model is a distributed knowledge reasoning model.

[0027] Optionally, before employing a matching network model to obtain a set of matching degrees for the first knowledge representation information and at least one second knowledge representation information, the method further includes:

[0028] Obtain the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set;

[0029] The first knowledge representation information is adjusted using the user representation learning module and bidding attribute information to obtain the adjusted first knowledge representation information;

[0030] The second knowledge representation information corresponding to any bid document is adjusted using the user representation learning module and bid attribute information to obtain the adjusted second knowledge representation information;

[0031] Traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set.

[0032] Optionally, the target bid document in at least one bid document is determined based on the matching degree set, including:

[0033] Get the highest matching score in the matching score set;

[0034] From at least one bid document, obtain the bid document with the highest matching degree and determine the bid document as the target bid document.

[0035] Secondly, embodiments of this application provide an object recommendation device combining RPA and AI, including:

[0036] The document acquisition unit is used to acquire a set of bidding documents and tender documents through the Robotic Process Automation (RPA) system;

[0037] The information acquisition unit is used to acquire first knowledge representation information corresponding to the bidding document on the first knowledge graph using the target knowledge representation model, and to acquire second knowledge representation information corresponding to at least one bidding document in the bidding document set. The first knowledge graph corresponds to the bidding document and the bidding document set.

[0038] The document determination unit is used to use a matching network model to obtain a set of matching degrees between first knowledge representation information and at least one second knowledge representation information, and to determine the target bidding document in at least one bidding document based on the set of matching degrees.

[0039] The object recommendation unit is used to recommend target bidding objects corresponding to the target bidding documents.

[0040] Optionally, the apparatus further includes a target document acquisition unit, a set extraction unit, and a graph building unit, used before acquiring the first knowledge representation information corresponding to the bidding document on the first knowledge graph using the target knowledge representation model:

[0041] The target document acquisition unit is used to acquire a set of tender documents and bid documents in the target document format through the Robotic Process Automation (RPA) system.

[0042] The set extraction unit is used to obtain the set of triplet information corresponding to the sets of bidding documents and tender documents using triplet extraction technology;

[0043] The graph building unit is used to build the first knowledge graph corresponding to the sets of bidding documents and tender documents based on the triplet information set.

[0044] Optionally, the set acquisition unit includes a set acquisition subunit, a word segmentation processing subunit, and an audit processing subunit. The set acquisition unit is used to obtain the set of triple information corresponding to the sets of bidding documents and tender documents using triple extraction technology.

[0045] The set acquisition subunit is used to obtain the first triplet information set corresponding to the bidding document and the second triplet information set corresponding to the bid document set using triplet extraction technology.

[0046] The word segmentation processing subunit is used to segment the first triplet information set and the second triplet set using the Jieba word segmentation model to obtain the processed first triplet information set and the processed second triplet set.

[0047] The audit processing subunit is used to perform entity boundary string audit processing on the processed first triplet information set and the processed second triplet set to obtain the triplet information set corresponding to the bidding document and bid document sets.

[0048] Optionally, the apparatus further includes a set acquisition unit, a graph acquisition unit, and a model building unit, used before acquiring the first knowledge representation information corresponding to the bidding document using the target knowledge representation model, and acquiring the second knowledge representation information corresponding to at least one bidding document in the bidding documents:

[0049] The set acquisition unit is used to acquire a set of bidding training documents;

[0050] The knowledge graph acquisition unit is used to acquire the knowledge graph corresponding to the set of bidding training documents.

[0051] The model building unit is used to perform vector space mapping on at least one knowledge pair in the knowledge graph based on distributed representation learning reasoning technology to build a target knowledge representation model.

[0052] Optionally, the apparatus further includes a parameter acquisition unit and a model determination unit, used before acquiring the first knowledge representation information corresponding to the bidding document using the target knowledge representation model, and acquiring the second knowledge representation information corresponding to at least one bidding document in the bidding documents:

[0053] The parameter acquisition unit is used to acquire the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set.

[0054] The model determination unit is used to determine the target knowledge representation model from the initial knowledge representation model set based on the model performance parameters. The target knowledge representation model is a distributed knowledge reasoning model.

[0055] Optionally, the apparatus further includes an attribute information acquisition unit, an information adjustment unit, and a set traversal unit, used before acquiring the matching degree set of the first knowledge representation information and at least one second knowledge representation information using a matching network model:

[0056] The attribute information acquisition unit is used to acquire the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the set of bidding documents.

[0057] The information adjustment unit is used to adjust the first knowledge representation information using the user representation learning module and the bidding attribute information to obtain the adjusted first knowledge representation information;

[0058] The information adjustment unit is also used to adjust the second knowledge representation information corresponding to any bid document using the user representation learning module and bid attribute information to obtain the adjusted second knowledge representation information;

[0059] The set traversal unit is used to traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set.

[0060] Optionally, the document determination unit includes a matching degree acquisition subunit and a document matching subunit. The document determination unit is used to determine the target bid document in at least one bid document based on the matching degree set.

[0061] The matching score acquisition sub-unit is used to obtain the highest matching score in the matching score set.

[0062] The document matching subunit is used to obtain the bid document with the highest matching degree from at least one bid document and identify the bid document as the target bid document.

[0063] Thirdly, embodiments of this application provide a terminal combining RPA and AI, the terminal including a memory and a processor. The memory and the processor communicate with each other via an internal connection path. The memory stores instructions, and the processor executes the instructions stored in the memory. When the processor executes the instructions stored in the memory, it causes the processor to perform the method in any of the embodiments described above.

[0064] Fourthly, embodiments of this application provide a computer-readable storage medium that stores a computer program, wherein when the computer program is run on a computer, the methods in any of the above-described embodiments are executed.

[0065] The advantages or beneficial effects of the above technical solutions include at least the following:

[0066] In one or related embodiments, a Robotic Process Automation (RPA) system is used to obtain a set of bidding documents and a set of tender documents. A target knowledge representation model is employed to obtain first knowledge representation information corresponding to the bidding documents on a first knowledge graph, and second knowledge representation information corresponding to at least one tender document in the tender document set is obtained. The first knowledge graph corresponds to the set of bidding documents and the set of tender documents. A matching network model is used to obtain matching degree sets between the first knowledge representation information and at least one second knowledge representation information, and a target tender document is determined from at least one tender document based on the matching degree sets. Target tender objects corresponding to the target tender documents are then recommended. Therefore, by using a target knowledge representation model to obtain the knowledge representation information corresponding to the bidding documents and tender documents on a first knowledge graph, and by making recommendations based on the knowledge representation information, the matching degree between the bidding documents and target tender documents can be improved during object recommendation, thereby improving the accuracy of object recommendation.

[0067] The above overview is for illustrative purposes only and is not intended to be limiting in any way. In addition to the illustrative aspects, embodiments, and features described above, further aspects, embodiments, and features of this application will become readily apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0068] In the accompanying drawings, unless otherwise specified, the same reference numerals throughout the various drawings denote the same or similar parts or elements. These drawings are not necessarily constructed to scale. It should be understood that these drawings depict only some embodiments according to this application and should not be construed as limiting the scope of this application.

[0069] Figure 1 This illustration shows a background diagram of an object recommendation method that combines RPA and AI to implement IA according to an embodiment of this application;

[0070] Figure 2 This illustration shows a background architecture diagram of an object recommendation method that combines RPA and AI to implement IA according to an embodiment of this application;

[0071] Figure 3 This document illustrates a flowchart of an object recommendation method that combines RPA and AI to implement IA according to an embodiment of this application.

[0072] Figure 4 This document illustrates a flowchart of an object recommendation method that combines RPA and AI to implement IA according to an embodiment of this application.

[0073] Figure 5 This diagram illustrates a format conversion process according to an embodiment of this application.

[0074] Figure 6 This diagram illustrates a first knowledge graph according to an embodiment of this application.

[0075] Figure 7 This document illustrates a flowchart of an object recommendation method that combines RPA and AI to implement IA according to an embodiment of this application.

[0076] Figure 8 This document illustrates a flowchart of an object recommendation method that combines RPA and AI to implement IA according to an embodiment of this application.

[0077] Figure 9 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0078] Figure 10 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0079] Figure 11 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0080] Figure 12 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0081] Figure 13 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0082] Figure 14 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0083] Figure 15 This illustration shows a schematic diagram of an object recommendation device that combines RPA and AI to implement IA according to an embodiment of this application;

[0084] Figure 16 This diagram illustrates a structural block diagram of a terminal according to an embodiment of the present application. Detailed Implementation

[0085] The embodiments of this application are described in detail below. Examples of these embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application.

[0086] In the description of this application, the term "multiple" means two or more.

[0087] In the description of this application, the term "RPA" refers to the automatic execution of process tasks according to rules by simulating human operation on a computer through specific "robot software." An RPA system comprises at least three components: development tools, runtime tools, and a control center. The intelligent automation platform seamlessly integrates multiple capabilities such as RPA, Intelligent Document Processing (IDP), Conversational AI (CoAI), and Process Mining. It possesses five major functional categories: "business understanding," "process creation," "anywhere operation," "centralized management and control," and "human-machine collaboration." This enables enterprises to achieve end-to-end intelligent automation of business processes, replacing manual operations, further improving business efficiency, and accelerating digital transformation.

[0088] In the intelligent automation platform, the three components of this RPA system are named Process Creator, Process Robot, and Robot Commander. The Process Creator is a programming tool for developing processes, performing specific steps within the process such as automating interface operations, AI recognition, and data reading and writing. The Process Creator allows users to easily assemble automated processes that meet business needs using flowcharts and low-code methods, dragging and dropping individual steps with a mouse.

[0089] Once the RPA process is written, it is deployed within the process robot. It can be started manually as needed, or automatically when specific trigger conditions are met. Tasks are programmable, and processes are traceable.

[0090] Robot Commander is a platform for unified management of multiple process robots within an enterprise. It can quickly and in batches issue tasks and provide the process robots with the data, credentials, and documents required for operation. It can also monitor the operational status of process robots in real time or review their historical records.

[0091] The intelligent automation platform also provides artificial intelligence (AI) capabilities specifically designed for RPA. These AI capabilities constitute the fourth component of the intelligent automation platform, known as the intelligent document processing platform. The intelligent document processing platform is built upon deep learning algorithms such as Optical Character Recognition (OCR) and Natural Language Processing (NLP). It provides functions such as document recognition, classification, element extraction, verification, comparison, and error correction, automating daily document processing tasks for enterprises.

[0092] In the description of this application, the term "tender document" refers to the invitation to offer issued by the tendering party to potential bidders, which informs them of project requirements, tendering and bidding rules, and contract terms, and is the primary basis for project tendering and bidding activities.

[0093] In the description of this application, the term "tender document" refers to the responsive document prepared by the bidder in accordance with the requirements of the tender document.

[0094] In the description of this application, the term "tender object" refers to the object for which a tender document is prepared in accordance with the requirements of the tender document.

[0095] In the description of this application, the term "knowledge representation model" refers to a model used to represent semantic information of images, text, speech, etc., as low-dimensional dense entity vectors. Types of knowledge representation models include, but are not limited to, Structured Embedding (SE), Single Layer Model (SLM), Semantic Matching Energy (SME), Translational Distance Model, Bilinear Function Based Models, Neural Tensor Network (NTN), Matrix Factorization Models, Translation Models, and so on.

[0096] In the description of this application, the term "knowledge representation information" refers to low-dimensional dense entity vector information corresponding to semantic information obtained using a knowledge representation model.

[0097] In the description of this application, the term "knowledge graph" refers to a semantic network with entities and concepts as nodes and semantic relationships as edges. In the library and information science field, a knowledge graph is called a knowledge domain visualization or knowledge domain mapping map. It is a series of various graphics that display the development process and structural relationships of knowledge, using visualization techniques to describe knowledge resources and their carriers, and to mine, analyze, construct, and display knowledge and the interrelationships between them.

[0098] In the description of this application, the term "matching network model" refers to a model used to map objects to an embedding space that also encapsulates the label distribution, then project test objects into the same embedding space using different architectures, and then use cosine similarity to measure similarity, thereby achieving classification and detection performance.

[0099] In the description of this application, the term "document format" refers to a special encoding method used by a computer to store text information. Document formats include, but are not limited to, text format (txt), HTML format, Word format, Portable Document Format (PDF), etc.

[0100] In the description of this application, the term "triple extraction technique" refers to a technique used to jointly extract triple information of entities and relations, including the extraction of multiple relations between entities.

[0101] In the description of this application, the term "triple" refers to a set of the form ((x, y), z) (that is, a triple is an even pair whose first projective is also an even pair), often abbreviated as (x, y, z). Triples are a concept in data structures, a fundamental course in computer science, primarily used as a compression method for storing sparse matrices.

[0102] In the description of this application, the term "triple information" refers to information used to identify entities and relationships between entities. This triple information does not refer to any specific, fixed information. For example, when the triple information corresponding to at least one piece of data changes, the triple information may also change accordingly.

[0103] In the description of this application, the term "entity" generally refers to something that can exist independently, serving as the foundation of all attributes and the origin of all things. Entities are the most basic elements in a knowledge graph.

[0104] In the description of this application, the term "Jieba segmentation model" refers to a model used for segmenting Chinese text. This Jieba segmentation model can perform functions such as word segmentation, part-of-speech tagging, and keyword extraction on Chinese text, and supports custom dictionaries.

[0105] In the description of this application, the term "word segmentation" refers to the process of recombining a continuous sequence of characters into a sequence of words according to certain rules.

[0106] In the description of this application, the term "entity boundary string auditing process" refers to the process used to audit whether triple information is information of an instruction category.

[0107] In the description of this application, the term "model performance parameter" refers to configuration variables within a model. Model performance parameters can be estimated or learned from data. Model performance parameters are required when making model predictions. Model performance parameter values ​​define the model's functionality. Model performance parameters are typically stored as part of the learned model. Model performance parameters can be estimated using optimization algorithms, which are efficient searches for possible values ​​of the parameters.

[0108] In the description of this application, the term "distributed representation learning inference technique" refers to a technique that represents entity vectors in a low-dimensional dense vector space and then performs computation and inference.

[0109] With the development of science and technology and the increasing maturity of terminal technology, the convenience of users' production and life has been improved. In terminal application scenarios, users can use object recommendation applications to recommend objects based on the data they input.

[0110] According to some embodiments, Figure 1This diagram illustrates a background illustration of an object recommendation method for implementing IA (Information Technology) by combining RPA and AI, according to an embodiment of this application. Figure 1 As shown, users can click on the object recommendation application on the terminal. When the terminal detects that the user has clicked on the object recommendation application, it can display the object recommendation interface. Users can then input bidding documents based on this interface. Next, when the terminal detects that the user has clicked the object recommendation button, it can recommend objects based on the entered bidding documents.

[0111] According to some embodiments, Figure 2 This diagram illustrates the background architecture of an object recommendation method combining RPA and AI, according to one embodiment of this application. Figure 2 As shown, terminal 11 can upload the tender document submitted by the user to server 13 via network 12. When server 13 receives the tender document, it can determine the matching tender document based on historical user preference information and send the matching tender document to terminal 11 via network 12. When terminal receives the tender document sent by server 13, it can display the tender document on the display interface.

[0112] In some embodiments, in bidding scenarios, it is necessary to understand the interaction information between past winning bidders and projects, the attribute information of the companies and projects, and the background knowledge of the companies and projects, including company qualifications, investment amounts, etc., in order to make recommendations based on the bidding information provided by the user. However, in related technologies, when making user recommendations, the terminal only uses historical user preference information and interaction information, resulting in low accuracy in user recommendations.

[0113] It is easy to understand that this terminal includes, but is not limited to: wearable devices, handheld devices, personal computers, tablets, in-vehicle devices, smartphones, computing devices, or other processing devices connected to a wireless modem. In different networks, the terminal device may be called by different names, such as: user equipment, access terminal, user unit, user station, mobile station, mobile station, remote station, remote terminal, mobile device, user terminal, terminal, wireless communication device, user agent or user device, cellular phone, cordless phone, personal digital assistant (PDA), terminal device in 5G networks, or future evolved networks, etc. An operating system can be installed on this terminal. This operating system refers to the operating system that can run on the terminal; it is a program that manages and controls the terminal hardware and terminal applications, and is an indispensable system application in the terminal. This operating system includes, but is not limited to, Android, iOS, Windows Phone (WP), and Ubuntu Mobile.

[0114] These and other aspects of the embodiments of this application will become clear from the following description and accompanying drawings. In these descriptions and drawings, specific embodiments of the present application are specifically described to illustrate some ways of implementing the principles of the embodiments of the present application; however, it should be understood that the scope of the embodiments of the present application is not limited thereto. Rather, the embodiments of the present application include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.

[0115] The following describes, with reference to the accompanying drawings, an object recommendation method for implementing IA by combining RPA and AI according to embodiments of this application.

[0116] Figure 3 This document illustrates a flowchart of an object recommendation method for implementing IA by combining RPA and AI according to an embodiment of this application. Figure 3 As shown, the method may include the following steps:

[0117] Step S101: Obtain the tender documents and bid documents set through the Robotic Process Automation (RPA) system;

[0118] According to some embodiments, a bid document set refers to a collection comprised of at least one bid document. This bid document set does not specifically refer to a particular set of bid documents. For example, the bid document set may change accordingly when the content of the bid documents changes. Similarly, the bid document set may change accordingly when the number of bid documents included in it changes.

[0119] It is easy to understand that when the terminal makes object recommendations, it can obtain a set of bidding documents and tender documents through the Robotic Process Automation (RPA) system.

[0120] Step S102: Use the target knowledge representation model to obtain the first knowledge representation information corresponding to the bidding document on the first knowledge graph, and obtain the second knowledge representation information corresponding to at least one bidding document in the bidding document set;

[0121] According to some embodiments, the target knowledge representation model refers to a trained model used to acquire knowledge representation information on a first knowledge graph. This target knowledge representation model is not specifically a fixed model. For example, the target knowledge representation model can change when the first knowledge graph changes. The target knowledge representation model can also change when the terminal receives a model modification instruction for it.

[0122] In some embodiments, the first knowledge graph refers to the knowledge graph corresponding to the set of bidding documents and the set of tender documents. This first knowledge graph does not specifically refer to a fixed graph. For example, the first knowledge graph can change when the bidding documents change. It can also change when the set of tender documents changes.

[0123] In some embodiments, the first knowledge representation information refers to the knowledge representation information corresponding to the tender document. This first knowledge representation information does not specifically refer to any fixed information. For example, the first knowledge representation information may change when the tender document changes. It may also change when the first knowledge graph changes.

[0124] In some embodiments, the second knowledge representation information refers to the knowledge representation information corresponding to the tender document. This second knowledge representation information does not specifically refer to any fixed information. For example, the second knowledge representation information may change when the tender document changes. Similarly, the second knowledge representation information may change when the first knowledge graph changes.

[0125] It is easy to understand that when the terminal obtains the set of bidding documents and tender documents, it can use the target knowledge representation model to obtain the first knowledge representation information corresponding to the bidding documents on the first knowledge graph. Furthermore, the terminal can also obtain the second knowledge representation information corresponding to at least one tender document among the tender documents.

[0126] Step S103: Using a matching network model, obtain the matching degree set between the first knowledge representation information and at least one second knowledge representation information, and determine the target bidding document in at least one bidding document based on the matching degree set;

[0127] According to some embodiments, the matching degree refers to the degree of matching between the first knowledge representation information and the second knowledge representation information. This matching degree is not specifically defined by a fixed degree. For example, the matching degree may change when the first knowledge representation information changes. Similarly, the matching degree may change when the second knowledge representation information changes.

[0128] In some embodiments, the matching degree set refers to a set comprised of at least one matching degree between the first knowledge representation information and at least one second knowledge representation information. This matching degree set is not specifically a fixed set. For example, the matching degree set can change when the number of second knowledge representation information pieces changes. The matching degree set can also change when the matching degree changes.

[0129] In some embodiments, the target bid document refers to a bid document that matches the tender document, obtained from the bid document set based on the matching degree set. This target bid document does not specifically refer to a single, fixed document. For example, the target bid document may change when the bid document set changes. Similarly, the target bid document may change when the tender document changes.

[0130] It is easy to understand that when the terminal obtains the first knowledge representation information and at least one second knowledge representation information, the terminal can use a matching network model to obtain the matching degree set between the first knowledge representation information and at least one second knowledge representation information. Then, the terminal can determine the target bidding document in at least one bidding document based on the matching degree set.

[0131] Step S104: Recommend target bid objects corresponding to the target bid documents.

[0132] In some embodiments, the target bid object refers to the tendering object corresponding to the target bid document. This target bid object is not specifically defined as a fixed object. For example, the target bid object can change when the target bid document changes.

[0133] It is easy to understand that when the terminal obtains the target bidding document, the terminal can recommend the target bidding objects corresponding to the target bidding document.

[0134] In this embodiment, a Robotic Process Automation (RPA) system is used to obtain a set of bidding documents and tender documents. A target knowledge representation model is employed to obtain first knowledge representation information corresponding to the bidding documents from a first knowledge graph, and second knowledge representation information corresponding to at least one tender document from the tender document set is obtained. A matching network model is used to obtain matching degree sets between the first knowledge representation information and at least one second knowledge representation information, and a target tender document is determined from at least one tender document based on the matching degree sets. Target tender objects corresponding to the target tender documents are then recommended. Therefore, by using a target knowledge representation model to obtain the knowledge representation information corresponding to the bidding documents and tender documents from the first knowledge graph, and by making recommendations based on the knowledge representation information, the matching degree between the bidding documents and target tender documents can be improved during object recommendation, thereby increasing the accuracy of object recommendation.

[0135] Figure 4 This document illustrates a flowchart of an object recommendation method for implementing IA by combining RPA and AI according to an embodiment of this application. Figure 4 As shown, the method may include the following steps:

[0136] Step S201: Obtain the set of bidding documents and tender documents through the Robotic Process Automation (RPA) system;

[0137] According to some embodiments, when a terminal obtains a set of bidding documents and tender documents based on RPA, the terminal can use the RPA system to extract information from the source, including the bidding documents and tender document sets, as well as the document format corresponding to the bidding documents and the document format corresponding to at least one tender document in the tender document set.

[0138] In some embodiments, the information extraction source refers to the source of the tender documents and bid document collection. This information extraction source does not specifically refer to any single, fixed information. Information extraction sources include, but are not limited to, public resource trading websites, core industry bidding websites, regional bidding websites, etc. Having multiple information extraction sources ensures comprehensive coverage of information data and allows for real-time access to the latest tender documents and bid document collections.

[0139] In some embodiments, when a terminal retrieves a set of tender documents and bid documents from an information extraction source using an RPA system, the terminal can retrieve web pages from the website addresses in the information extraction source. Once the terminal retrieves the web pages, it can extract the tender documents and bid document sets, as well as the document formats corresponding to the tender documents and at least one bid document from the bid document set, from the web page source code.

[0140] For example, when a terminal crawls the webpage of a core industry bidding website, it can extract the corresponding Word-format bidding document from the webpage source code. Similarly, when a terminal crawls a regional bidding website, it can extract the corresponding PDF-format bidding document from the webpage source code.

[0141] It is easy to understand that when the terminal makes object recommendations, it can obtain a set of bidding documents and tender documents through the Robotic Process Automation (RPA) system.

[0142] Step S202: Obtain the set of tender documents and bid documents in the target document format through the Robotic Process Automation (RPA) system;

[0143] According to some embodiments, the target document format refers to the document format corresponding to the tender document selected by the RPA system and the document format corresponding to at least one tender document in the set of tender documents. This target document format does not have a fixed format. For example, when the document format selected by the RPA system changes, the target document format can change accordingly. The target document format can also change when the terminal receives a format modification instruction for the target document format.

[0144] In some embodiments, when a terminal obtains a set of tender documents and bid documents through a Robotic Process Automation (RPA) system, if the terminal determines that the document format corresponding to the tender document and the document format corresponding to at least one bid document in the bid document set are not the target document format, the terminal can use the RPA system to convert the tender documents and bid document set into formats to obtain the tender documents and bid document set in the target document format.

[0145] In some embodiments, when a terminal performs format conversion on a collection of bidding documents and tender documents through an RPA system, the terminal can utilize the document format conversion tool built into the RPA system to convert the bidding documents. For example, when the terminal obtains a bidding document in Word format, it can use the Python Win32 library to call the underlying macro language of Word (Visual Basic for Applications, VBA) to convert the Word bidding document into a PDF bidding document.

[0146] For example, if the target document format set on the terminal is PDF, and the terminal extracts a tender document (e.g., Word document A) from the source using the RPA system, the terminal can use the RPA system's built-in document format conversion tool to convert tender document A into a PDF document. Figure 5 As shown.

[0147] It is easy to understand that when a terminal obtains a set of tender documents and bid documents through a Robotic Process Automation (RPA) system, the terminal can obtain a set of tender documents and bid documents in the target document format through the RPA system.

[0148] Step S203: Use triplet extraction technology to obtain the triplet information set corresponding to the sets of bidding documents and tender documents;

[0149] According to some embodiments, a triplet information set refers to a set comprised of at least one triplet information. This triplet information set does not specifically refer to a fixed set. For example, the triplet information set may change when the tender documents change. It may also change when the set of bid documents changes.

[0150] For example, the triplet information set includes, but is not limited to, project type, project bidding announcement release date, project bidding company, project funding source, project location, project bidding scope, project construction period, project construction scale, project investment amount, project enterprise qualification, project personnel qualification, project enterprise credit registration requirements, project performance qualification, bidding project, bidding company, project quotation, bidding project, bidding company, project credit, project enterprise performance, this bid evaluation score, winning company, whether the project has won the bid, project discount rate, project awards, etc.

[0151] According to some embodiments, when the terminal uses triplet extraction technology to obtain the triplet information set corresponding to the bidding documents and the tender documents, the triplet extraction technology used by the terminal includes, but is not limited to, multi-head selection technology, multi-round question-and-answer technology, modular extraction technology, etc.

[0152] In some embodiments, when the terminal uses multi-head selection technology to obtain the triple information set corresponding to the bidding documents and the tender documents, the terminal can use a Conditional Random Field (CRF) to solve the entity recognition task, treating relation extraction as a multi-head selection problem, thereby realizing the function of identifying multiple potential relations for each entity.

[0153] In some embodiments, when a terminal uses multi-turn question-answering technology to obtain the triple information set corresponding to the bidding documents and tender documents, the terminal can treat entity-relation extraction as a multi-turn question-answering task. By constructing question models for entities and relations, and using text information (the text to be extracted) to answer the question template, the entity-relation can be obtained. For question-answering tasks, a span-based answer extraction method can be adopted.

[0154] In some embodiments, when the terminal uses modular extraction technology to obtain the triple information set corresponding to the bidding documents and the tender documents, the terminal can divide the entity-relation extraction into two modules. First, the head-entity extraction (HE) is extracted, and then the tail-entity and relation (TER) is extracted jointly, which can reduce entity overlap.

[0155] According to some embodiments, when a terminal uses triplet extraction technology to obtain triplet information sets corresponding to bidding documents and bid document sets, the terminal can use triplet extraction technology to obtain the first triplet information set corresponding to the bidding documents and the second triplet information set corresponding to the bid document sets. Then, the terminal can use the Jieba word segmentation model to perform word segmentation processing on the first and second triplet information sets, obtaining processed first and second triplet information sets. Finally, the terminal can perform entity boundary string verification processing on the processed first and second triplet information sets to obtain the triplet information sets corresponding to the bidding documents and bid document sets. Therefore, the terminal can remove garbled characters from the triplet information sets, retaining only triplet information of preset types, such as numbers, English letters, Chinese characters, etc., and remove duplicate triplet information from the triplet information sets. This can improve the information quality of the triplet information sets and increase the accuracy of object recommendations.

[0156] In some embodiments, the first triplet information set refers to the triplet information set corresponding to the tender document. This first triplet information set does not specifically refer to a fixed set. For example, the first triplet information set may change when the tender document changes. Similarly, the first triplet information set may also change when the triplet information changes.

[0157] In some embodiments, the second triplet information set refers to the triplet information set corresponding to the set of tender documents. This second triplet information set does not specifically refer to a fixed set. For example, when the set of tender documents changes, this second triplet information set may change. Similarly, when the triplet information changes, this second triplet information set may also change.

[0158] In some embodiments, when the terminal uses the Jieba segmentation model to segment the first triplet information set and the second triplet set, the terminal can use the part-of-speech tagging tool pos_seg to tag the triplet information in the first and second triplet information sets. Furthermore, by comparing the performance of the Jieba segmentation model and the part-of-speech tagging tool, useless noise entities in the first and second triplet information sets can be removed.

[0159] In some embodiments, the terminal performs entity boundary string verification on the processed first set of triplet information and the processed second set of triplet information. This reduces noise in the triplet information set and improves its quality. For example, when the terminal obtains the triplet information "A network_B province_power_co. Ltd._of_operation_status_is_yes", if the terminal sets "A network_B province_power_co. Ltd." as the required triplet information and "of_operation_status_is_yes" as irrelevant information, the terminal can remove this irrelevant information through entity boundary string verification.

[0160] In some embodiments, when the terminal uses triplet extraction technology to obtain the triplet information set corresponding to the bidding documents and tender documents, the terminal can also remove triplet information corresponding to objects with low frequency or no bidding activities by removing invalid features and sparse features in the triplet information set.

[0161] It is easy to understand that when the terminal obtains a set of tender documents and bid documents in the target document format, the terminal can use triplet extraction technology to obtain the triplet information set corresponding to the set of tender documents and bid documents.

[0162] Step S204: Based on the triplet information set, establish the first knowledge graph corresponding to the sets of bidding documents and tender documents;

[0163] According to some embodiments, when a terminal builds a first knowledge graph corresponding to a set of bidding documents and a set of tender documents based on a set of triplet information, the terminal can use a target deep learning model to obtain a subset of triplet information corresponding to the target entity in the set of triplet information, and merge the triplet information in the subset of triplet information. The terminal can then construct a first knowledge graph corresponding to the data set.

[0164] In some embodiments, the target deep learning model refers to the model by which the terminal acquires a subset of triple information corresponding to the target entity. This target deep learning model does not specifically refer to a particular fixed deep learning model. For example, when the type of deep learning model changes, the target deep learning model may also change accordingly. Similarly, when the name of the deep learning model changes, the target deep learning model may also change accordingly.

[0165] In some embodiments, the target entity refers to the entity information needed by the terminal when constructing the first knowledge graph. This target entity information does not specifically refer to any fixed information. For example, the target entity information may change when the set of triples changes. Similarly, the target entity information may change when the entity information changes.

[0166] In some embodiments, the first knowledge graph information refers to the information corresponding to the first knowledge graph. This first knowledge graph information does not specifically refer to any fixed information. For example, when the first knowledge graph changes, the first knowledge graph information may also change. When the target bidding attribute information changes, the first knowledge graph information may also change. When the target entity information changes, the first knowledge graph information may also change.

[0167] For example, Figure 6 This diagram illustrates a first knowledge graph according to one embodiment of this application. Figure 6 As shown, “Tender Document E2” is the winning bid document of “Tender Document E1”, the person in charge of “Tender Document E2” is “Person G2”, and “Tender Document E2” originates from “Organization F2”; “Tender Document E1” originates from “Organization F1”, and the executive officer of “Tender Document E1” is “Person G1”.

[0168] It is easy to understand that when the terminal obtains the triple information set corresponding to the bidding documents and the tender documents, the terminal can build the first knowledge graph corresponding to the bidding documents and the tender documents based on the triple information set.

[0169] Step S205: Obtain the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set;

[0170] According to some embodiments, the initial knowledge representation model refers to an untrained knowledge representation model. This initial knowledge representation model does not represent a fixed model. For example, the initial knowledge representation model can change when its performance parameters change. The initial knowledge representation model can also change when the terminal receives a model modification instruction for it.

[0171] In some embodiments, the initial knowledge representation model set refers to a collection comprised of at least one initial knowledge representation model. This initial knowledge representation model set is not specifically defined as a fixed set. For example, the initial knowledge representation model set can change when the initial knowledge representation models change. The initial knowledge representation model set can also change when the number of initial knowledge representation models changes.

[0172] It is easy to understand that when the terminal obtains the target knowledge representation model, the terminal can obtain the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set.

[0173] Step S206: Based on the model performance parameters, determine the target knowledge representation model from the initial knowledge representation model set;

[0174] According to some embodiments, when a terminal determines a target knowledge representation model from an initial knowledge representation model set based on model performance parameters, the terminal evaluates the learning efficiency and learning effect of the initial knowledge representation model based on the model performance parameters corresponding to that initial knowledge representation model. Furthermore, the terminal can select the target knowledge representation model to be used based on the learning efficiency and learning effect corresponding to at least one initial knowledge representation model in the initial knowledge representation model set.

[0175] For example, when the initial knowledge representation model is a transfer distance model, it can be a Translating Embeddings for Modeling Multi-relational Data (TransE) model, a Flexible Translation (TransF) model, a Learning Entity and Relation Embeddings (TransR) model, a Translating on Hyperplanes (TransH) model, or a Knowledge-Driven Human Activity Understanding (HAKE) model. When the initial knowledge representation model is a bilinear model, it can be a semantic matching model such as RESCAL, DistMult, ComplEx, HolE, SimplE, or Analogy. After evaluating the learning efficiency and effectiveness of these initial knowledge representation models, the terminal can choose to adopt the TransE model as the target knowledge representation model.

[0176] It is easy to understand that when the terminal obtains the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set, the terminal can determine the target knowledge representation model in the initial knowledge representation model set based on the model performance parameters.

[0177] Step S207: Use the target knowledge representation model to obtain the first knowledge representation information corresponding to the bidding document on the first knowledge graph, and obtain the second knowledge representation information corresponding to at least one bidding document in the bidding documents;

[0178] According to some embodiments, when a terminal uses a target knowledge representation model to obtain knowledge representation information on a first knowledge graph, the terminal can use the target knowledge representation model, based on a distributed representation learning inference method, to map the symbolic representation in the triple information to a vector space and perform numerical representation. Therefore, the curse of dimensionality can be reduced, while the implicit associations between entities and relations in the triple information can be captured, and the computation is direct and the training speed is fast.

[0179] For example, when the terminal obtains the triple information "State Grid_Jiangxi Province_Electricity", the terminal can represent the triple information as knowledge representation information (0, 1, 2) based on the distributed representation learning reasoning method.

[0180] It is easy to understand that when the terminal obtains the target knowledge representation model, it can use the target knowledge representation model to obtain the first knowledge representation information corresponding to the bidding document on the first knowledge graph. Furthermore, the terminal can also obtain the second knowledge representation information corresponding to at least one bidding document within the bidding documents.

[0181] Step S208: Using a matching network model, obtain the matching degree set between the first knowledge representation information and at least one second knowledge representation information, and determine the target bidding document in at least one bidding document based on the matching degree set;

[0182] According to some embodiments, when a terminal uses a matching network model to obtain matching degree sets between first knowledge representation information and at least one second knowledge representation information, the terminal can use the first knowledge representation information as fixed data for the matching network model, and sequentially input the second knowledge representation information from the second knowledge representation information set into the matching network model. Then, the terminal can obtain matching degree sets between the first knowledge representation information and at least one second knowledge representation information.

[0183] In some embodiments, when the terminal uses a matching network model to obtain a matching degree set between the first knowledge representation information and at least one second knowledge representation information, the terminal may also use an attention mechanism to improve the accuracy of obtaining the matching degree set.

[0184] In some embodiments, an attention mechanism refers to a special structure embedded in a machine learning model that automatically learns and calculates the contribution of input data to output data. This attention mechanism does not refer to a specific fixed structure. For example, the attention mechanism can change when the terminal receives a modification instruction for it.

[0185] It is easy to understand that when the terminal obtains the first knowledge representation information and at least one second knowledge representation information, the terminal can use a matching network model to obtain the matching degree set between the first knowledge representation information and at least one second knowledge representation information. Then, the terminal can determine the target bidding document in at least one bidding document based on the matching degree set.

[0186] Step S209: Recommend target bid objects corresponding to the target bid documents.

[0187] According to some embodiments, when a terminal determines at least one target bid document among bid documents based on a matching degree set, the terminal can determine at least two target bid documents. Then, the terminal can display these at least two target bid documents in a sequential order on the display page.

[0188] For example, the terminal can sort at least two matching scores in the matching score set according to their size. Then, the terminal can select the bidding documents corresponding to the top 10 matching scores as the target bidding documents, and thus, the terminal can display these 10 target bidding documents in chronological order on the display page.

[0189] In some embodiments, when the terminal recommends a target bidding object corresponding to a target bidding document, the terminal can also provide an interpretable and referable basis for the target bidding object through a recommendation explanation module, thereby improving the accuracy of object recommendation while making the recommendation result highly interpretable.

[0190] In some embodiments, the recommendation explanation module refers to a module used to explain to the user in a clear and understandable way why the target bidding document is recommended to them. This recommendation explanation module is not specifically defined as a single, fixed module. For example, the recommendation explanation module may change when the target bidding document changes. The recommendation explanation module may also change when the terminal receives a module modification instruction for it.

[0191] It is easy to understand that when the terminal obtains the target bidding document, the terminal can recommend the target bidding objects corresponding to the target bidding document.

[0192] In this embodiment, a Robotic Process Automation (RPA) system is used to obtain a set of tender documents and bid documents in the target document format. This improves the matching accuracy of the RPA system for tender documents and bid documents in different formats, thereby enhancing the accuracy of object recommendation. Secondly, a triplet extraction technique is employed to obtain a set of triplet information corresponding to the tender documents and bid documents. Based on this triplet information set, a first knowledge graph corresponding to the tender documents and bid documents is established. This improves the accuracy of the first knowledge graph acquisition, thereby enhancing the accuracy of bid object recommendation. Furthermore, the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set are obtained. Based on these performance parameters, a target knowledge representation model is determined from the initial knowledge representation model set. Therefore, by selecting the required target knowledge representation model from the initial knowledge representation model set, the accuracy of object recommendation can be improved.

[0193] Figure 7 This document illustrates a flowchart of an object recommendation method for implementing IA by combining RPA and AI according to an embodiment of this application. Figure 7 As shown, the method may include the following steps:

[0194] Step S301: Obtain the tender documents and bid documents set through the Robotic Process Automation (RPA) system;

[0195] The specific process is as described above and will not be repeated here.

[0196] Step S302: Obtain the set of bidding training documents;

[0197] According to some embodiments, the bidding training document refers to a document used to train a knowledge representation model, enabling the knowledge representation model to express prior knowledge. This bidding training document does not specifically refer to a single, fixed document. For example, the bidding training document can change when the knowledge representation model changes. It can also change when the sets of bidding documents and tender documents change.

[0198] In some embodiments, the bidding training document set refers to a collection comprised of at least one bidding training document. This bidding training document set does not specifically refer to a fixed set. For example, the bidding training document set can change when the bidding training documents change. It can also change when the knowledge representation model changes.

[0199] It is easy to understand that when the terminal obtains the target knowledge representation model, the terminal can obtain the set of bidding training documents.

[0200] Step S303: Obtain the knowledge graph corresponding to the set of bidding training documents;

[0201] According to some embodiments, when a terminal obtains a set of bidding training documents, it can acquire the bidding attribute information and entity information corresponding to the set of bidding training documents. Then, the terminal can use a Robotic Process Automation (RPA) system and an Optical Character Recognition (OCR) model to obtain the knowledge graph information corresponding to the bidding attribute information and entity information. Finally, the terminal can display the knowledge graph corresponding to this knowledge graph information.

[0202] In some embodiments, bidding attribute information refers to the attribute information corresponding to the bidding training document set. This bidding attribute information includes, but is not limited to, bidding information, tender information, industry information, regional information, etc. For example, the bidding attribute information corresponding to "Bidding Document M" in the bidding training document set is bidding information. The bidding attribute information corresponding to "Tender Document N" is tender information.

[0203] In some embodiments, entity information refers to the information of entities corresponding to the bidding training document set. Entity information includes, but is not limited to, object attribute information, data attribute information, relational attribute information, etc. Object attribute information refers to the entity name or attribute name that abstracts and describes entities of the same type. Object attribute information includes, but is not limited to, people, organizations, time, posts, accounts, etc. Data attribute information refers to the attribute information corresponding to the entity itself. For example, the attribute information corresponding to the entity "Person D" includes, but is not limited to, name D1, former name D2, age D3, etc. Relational attribute information refers to the mutual relationship information between entities.

[0204] In some embodiments, knowledge graph information refers to the information used in constructing the knowledge graph. This knowledge graph information does not specifically refer to any fixed information. For example, the knowledge graph information can change when entity information changes. Similarly, the knowledge graph information can change when bidding attribute information changes.

[0205] It is easy to understand that when the terminal obtains the set of bidding training documents, the terminal can obtain the knowledge graph corresponding to the set of bidding training documents.

[0206] Step S304: Based on distributed representation learning inference technology, perform vector space mapping on at least one knowledge pair in the knowledge graph to establish a target knowledge representation model;

[0207] In some embodiments, knowledge refers to the entity relationship information contained in the knowledge graph. This knowledge is not specific to any particular fixed set of knowledge. For example, this knowledge can change when the knowledge graph changes. Similarly, it can change when the set of bidding training documents changes.

[0208] In some embodiments, when a terminal performs vector space mapping on at least one knowledge pair in a knowledge graph, the terminal can obtain the mapping function corresponding to the entity relations in the knowledge graph based on distributed representation learning inference technology. Then, the terminal can map the entity relations to the vector space and represent the entity relations in numerical form. For example, the knowledge "winning the bid" can be mapped to 25, and the knowledge "bidding" can be mapped to the index value 68.

[0209] It is easy to understand that when the terminal obtains the knowledge graph corresponding to the set of bidding training documents, the terminal can perform vector space mapping on at least one knowledge pair in the knowledge graph based on distributed representation learning inference technology. Furthermore, the terminal can establish a target knowledge representation model.

[0210] Step S305: Use the target knowledge representation model to obtain the first knowledge representation information corresponding to the bidding document on the first knowledge graph, and obtain the second knowledge representation information corresponding to at least one bidding document in the bidding documents;

[0211] The specific process is as described above and will not be repeated here.

[0212] Step S306: Using a matching network model, obtain the matching degree set between the first knowledge representation information and at least one second knowledge representation information respectively;

[0213] The specific process is as described above and will not be repeated here.

[0214] Step S307: Obtain the highest matching degree in the matching degree set;

[0215] For example, when the terminal obtains a matching degree of 30 between the first knowledge representation information and the second knowledge representation information B1, a matching degree of 50 between the first knowledge representation information and the second knowledge representation information B2, and a matching degree of 70 between the first knowledge representation information and the second knowledge representation information B3, then the terminal can obtain that the matching degree between the first knowledge representation information and the second knowledge representation information B3 is the highest matching degree in the matching degree set.

[0216] It is easy to understand that when the terminal obtains a set of matching degrees between the first knowledge representation information and at least one second knowledge representation information, the terminal can obtain the highest matching degree in the set of matching degrees.

[0217] Step S308: Among at least one bid document, obtain the bid document with the highest matching degree and determine the bid document as the target bid document;

[0218] For example, when the terminal obtains that the matching degree between the first knowledge representation information and the second knowledge representation information B3 is the highest matching degree in the matching degree set, the terminal can obtain the bid document b3 corresponding to the highest matching degree from at least one bid document. Then, the terminal can determine bid document b3 as the target bid document.

[0219] It is easy to understand that when the terminal obtains the highest matching degree in the matching degree set, the terminal can obtain the bidding document corresponding to the highest matching degree in at least one bidding document and determine the bidding document as the target bidding document.

[0220] Step S309: Recommend target bidders corresponding to the target bid documents.

[0221] The specific process is as described above and will not be repeated here.

[0222] In this embodiment, by acquiring a set of bidding training documents and the corresponding knowledge graph, and using distributed representation learning inference technology, vector space mapping is performed on at least one knowledge pair in the knowledge graph to establish a target knowledge representation model. This improves the accuracy of the target knowledge representation model and consequently enhances the accuracy of user recommendations. Secondly, by acquiring the highest matching degree from the matching degree set, and identifying the bidding document with the highest matching degree among at least one bidding document, this method determines the target bidding document. Therefore, by identifying the bidding document with the highest matching degree as the target bidding document, the accuracy of target bidding document determination is improved.

[0223] Figure 8 This document illustrates a flowchart of an object recommendation method for implementing IA by combining RPA and AI according to an embodiment of this application. Figure 8 As shown, the method may include the following steps:

[0224] Step S401: Obtain the tender documents and bid documents set through the Robotic Process Automation (RPA) system;

[0225] The specific process is as described above and will not be repeated here.

[0226] Step S402: Use the target knowledge representation model to obtain the first knowledge representation information corresponding to the bidding document on the first knowledge graph, and obtain the second knowledge representation information corresponding to at least one bidding document in the bidding documents;

[0227] The specific process is as described above and will not be repeated here.

[0228] Step S403: Obtain the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set;

[0229] According to some embodiments, bidding attribute information refers to the attribute information corresponding to the bidding document. This bidding attribute information does not specifically refer to any single fixed piece of information. It includes, but is not limited to, bidding information, industry information, regional information, etc.

[0230] In some embodiments, bid attribute information refers to the attribute information corresponding to the bid document. This bid attribute information includes, but is not limited to, bid information, industry information, regional information, etc.

[0231] It is easy to understand that when the terminal obtains the first knowledge representation information corresponding to the bidding document and the second knowledge representation information corresponding to at least one bidding document in the bidding document set, the terminal can obtain the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set.

[0232] Step S404: Adjust the first knowledge representation information using the user representation learning module and bidding attribute information to obtain the adjusted first knowledge representation information;

[0233] According to some embodiments, a user representation learning module refers to a module used to automatically learn effective features from raw input data and convert the input information into effective feature representations. This user representation learning module does not specifically refer to a particular fixed module. The representation forms adopted by this user representation learning module include, but are not limited to, local representations, distributed representations, etc.

[0234] In some embodiments, local representation refers to representation as a one-hot vector. This discrete representation is highly interpretable, but one-hot vectors have a high dimensionality, cannot be extended, and are orthogonal to each other, making it impossible to calculate similarity.

[0235] In some embodiments, the distribution representation can typically be represented as a low-dimensional continuous dense vector, which has a much stronger representational power and the similarity is also easy to calculate.

[0236] In some embodiments, when the terminal obtains the bidding attribute information corresponding to the bidding document, the terminal can use a user representation learning module to fuse and multiply the feature representation corresponding to the bidding attribute information with the first knowledge representation information. Then, the terminal can obtain the adjusted first knowledge representation information.

[0237] It is easy to understand that when the terminal obtains the bidding attribute information corresponding to the bidding document, the terminal can use the user representation learning module and the bidding attribute information to adjust the first knowledge representation information to obtain the adjusted first knowledge representation information.

[0238] Step S405: Adjust the second knowledge representation information corresponding to any bid document using the user representation learning module and bid attribute information to obtain the adjusted second knowledge representation information;

[0239] According to some embodiments, when a terminal obtains the bidding attribute information corresponding to a bidding document, the terminal can use a user representation learning module to fuse and multiply the feature representation corresponding to the bidding attribute information with the second knowledge representation information. Then, the terminal can obtain the adjusted second knowledge representation information.

[0240] It is easy to understand that when the terminal obtains the bidding attribute information corresponding to any bidding document in the bidding document set, the terminal can use the user representation learning module and the bidding attribute information to adjust the second knowledge representation information corresponding to any bidding document to obtain the adjusted second knowledge representation information.

[0241] Step S406: Traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set;

[0242] It is easy to understand that when the terminal obtains the bidding attribute information corresponding to any bidding document in the bidding document set, the terminal can traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set.

[0243] Step S407: Using a matching network model, obtain the matching degree set between the adjusted first knowledge representation information and at least one adjusted second knowledge representation information, and determine the target bidding document in at least one bidding document based on the matching degree set;

[0244] The specific process is as described above and will not be repeated here.

[0245] Step S408: Recommend target bidders corresponding to the target bid documents.

[0246] The specific process is as described above and will not be repeated here.

[0247] In this embodiment, by obtaining the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set, the first knowledge representation information is adjusted using a user representation learning module and the bidding attribute information to obtain the adjusted first knowledge representation information. Then, the second knowledge representation information set is traversed to obtain the adjusted second knowledge representation information set. Therefore, by employing a multi-model fusion learning method to perform representation learning and fine-tuning on the knowledge representation information, the accuracy of object recommendation can be improved. Secondly, a matching network model is used to obtain the matching degree set between the adjusted first knowledge representation information and at least one adjusted second knowledge representation information, and based on the matching degree set, a target bidding document in at least one bidding document is determined. The target bidding object corresponding to the target bidding document is then recommended. Therefore, by making recommendations based on the adjusted knowledge representation information, the terminal can fully consider the bidding background knowledge when recommending objects, thereby improving the accuracy of object recommendation.

[0248] The following are embodiments of the apparatus described in this application, which can be used to execute the embodiments of the method described in this application. For details not disclosed in the apparatus embodiments of this application, please refer to the embodiments of the method described in this application.

[0249] Please see Figure 9 This is a schematic diagram of the structure of an object recommendation device combining RPA and AI to implement IA, according to an embodiment of this application. This object recommendation device combining RPA and AI to implement IA can be implemented as all or part of a device through software, hardware, or a combination of both. The object recommendation device 9000 combining RPA and AI to implement IA includes a document acquisition unit 9001, an information acquisition unit 9002, a document determination unit 9003, and an object recommendation unit 9004, wherein:

[0250] Document acquisition unit 9001 is used to acquire a set of bidding documents and tender documents through the Robotic Process Automation (RPA) system;

[0251] The information acquisition unit 9002 is used to acquire first knowledge representation information corresponding to the bidding document on the first knowledge graph using the target knowledge representation model, and to acquire second knowledge representation information corresponding to at least one bidding document in the bidding document set. The first knowledge graph corresponds to the bidding document and the bidding document set.

[0252] The document determination unit 9003 is used to use a matching network model to obtain a matching degree set between the first knowledge representation information and at least one second knowledge representation information, and to determine the target bidding document in at least one bidding document based on the matching degree set;

[0253] The object recommendation unit 9004 is used to recommend target bidding objects corresponding to target bidding documents.

[0254] Optional, Figure 10 This diagram illustrates the structure of an object recommendation apparatus that combines RPA and AI to implement IA according to an embodiment of this application. Figure 10 As shown, the object recommendation device 9000 also includes a target document acquisition unit 9005, a set extraction unit 9006, and a graph building unit 9007, used to acquire the first knowledge representation information corresponding to the bidding document on the first knowledge graph using the target knowledge representation model:

[0255] The target document acquisition unit 9005 is used to acquire a set of tender documents and bid documents in the target document format through the Robotic Process Automation (RPA) system.

[0256] Set extraction unit 9006 is used to obtain the set of triplet information corresponding to the sets of bidding documents and tender documents using triplet extraction technology;

[0257] The graph building unit 9007 is used to build the first knowledge graph corresponding to the sets of bidding documents and tender documents based on the triplet information set.

[0258] Optional, Figure 11 This diagram illustrates the structure of an object recommendation apparatus that combines RPA and AI to implement IA according to an embodiment of this application. Figure 11 As shown, the set acquisition unit 9006 includes a set acquisition subunit 9106, a word segmentation processing subunit 9206, and an audit processing subunit 9306. The set acquisition unit 9006 is used to acquire the set of triplet information corresponding to the sets of bidding documents and tender documents using triplet extraction technology:

[0259] Set acquisition subunit 9106 is used to acquire the first triplet information set corresponding to the bidding document and the second triplet information set corresponding to the bid document set using triplet extraction technology;

[0260] The word segmentation processing subunit 9206 is used to perform word segmentation processing on the first triplet information set and the second triplet set using the Jieba word segmentation model to obtain the processed first triplet information set and the processed second triplet set.

[0261] The audit processing subunit 9306 is used to perform entity boundary string audit processing on the processed first triplet information set and the processed second triplet set to obtain the triplet information set corresponding to the bidding document and the tender document set.

[0262] Optional, Figure 12 This diagram illustrates the structure of an object recommendation apparatus that combines RPA and AI to implement IA according to an embodiment of this application. Figure 12As shown, the object recommendation device 9000 further includes a set acquisition unit 9008, a graph acquisition unit 9009, and a model building unit 9010, used before acquiring the first knowledge representation information corresponding to the bidding document using the target knowledge representation model, and acquiring the second knowledge representation information corresponding to at least one bidding document in the bidding documents:

[0263] Set acquisition unit 9008 is used to acquire a set of bidding training documents;

[0264] The graph acquisition unit 9009 is used to acquire the knowledge graph corresponding to the set of bidding training documents.

[0265] The model building unit 9010 is used to perform vector space mapping on at least one knowledge pair in the knowledge graph based on distributed representation learning inference technology to build a target knowledge representation model.

[0266] Optional, Figure 13 This diagram illustrates the structure of an object recommendation apparatus that combines RPA and AI to implement IA according to an embodiment of this application. Figure 13 As shown, the object recommendation device 9000 further includes a parameter acquisition unit 9011 and a model determination unit 9012, used before acquiring the first knowledge representation information corresponding to the bidding document using the target knowledge representation model, and acquiring the second knowledge representation information corresponding to at least one bidding document in the bidding documents:

[0267] The parameter acquisition unit 9011 is used to acquire the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set;

[0268] The model determination unit 9012 is used to determine the target knowledge representation model from the initial knowledge representation model set based on the model performance parameters. The target knowledge representation model is a distributed knowledge reasoning model.

[0269] Optional, Figure 14 This diagram illustrates the structure of an object recommendation apparatus that combines RPA and AI to implement IA according to an embodiment of this application. Figure 14 As shown, the object recommendation device 9000 further includes an attribute information acquisition unit 9013, an information adjustment unit 9014, and a set traversal unit 9015, used before acquiring a matching degree set of first knowledge representation information and at least one second knowledge representation information using a matching network model:

[0270] The attribute information acquisition unit 9013 is used to acquire the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set.

[0271] The information adjustment unit 9014 is used to adjust the first knowledge representation information using the user representation learning module and the bidding attribute information to obtain the adjusted first knowledge representation information;

[0272] The information adjustment unit 9014 is also used to adjust the second knowledge representation information corresponding to any bid document using the user representation learning module and the bid attribute information to obtain the adjusted second knowledge representation information;

[0273] Set traversal unit 9015 is used to traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set.

[0274] Optional, Figure 15 This diagram illustrates the structure of an object recommendation apparatus that combines RPA and AI to implement IA according to an embodiment of this application. Figure 15 As shown, the document determination unit 9003 further includes a matching degree acquisition subunit 9103 and a document matching subunit 9203. The document determination unit 9003 is used to determine the target bid document in at least one bid document based on the matching degree set:

[0275] The matching degree acquisition subunit 9103 is used to obtain the highest matching degree in the matching degree set;

[0276] The document matching subunit 9203 is used to obtain the bid document with the highest matching degree from at least one bid document and determine the bid document as the target bid document.

[0277] The functions of each module in each device in the embodiments of this application can be found in the corresponding descriptions in the above methods, and will not be repeated here.

[0278] In this embodiment, the document acquisition unit obtains a set of bidding documents and tender documents through a Robotic Process Automation (RPA) system. The information acquisition unit uses a target knowledge representation model to obtain first knowledge representation information corresponding to the bidding documents on a first knowledge graph, and obtains second knowledge representation information corresponding to at least one tender document in the tender documents. The first knowledge graph corresponds to the set of bidding documents and tender documents. The document determination unit uses a matching network model to obtain a matching degree set between the first knowledge representation information and at least one second knowledge representation information, and determines the target tender document in at least one tender document based on the matching degree set. The object recommendation unit recommends the target tender object corresponding to the target tender document. Therefore, by using a target knowledge representation model to obtain the knowledge representation information corresponding to the bidding documents and tender documents on the first knowledge graph, and making recommendations based on the knowledge representation information corresponding to the bidding documents and tender documents, the terminal can fully consider the background knowledge of bidding when making object recommendations, thereby improving the accuracy of object recommendations.

[0279] Figure 16 This diagram illustrates a structural block diagram of a terminal according to an embodiment of this application. Figure 16 As shown, the terminal includes a memory 1610 and a processor 1620. The memory 1610 stores a computer program that can run on the processor 1620. When the processor 1620 executes the computer program, it implements the object recommendation method described in the above embodiments. The number of memories 1610 and processors 1620 can be one or more.

[0280] The terminal also includes:

[0281] The communication interface 1630 is used to communicate with external devices and exchange and transmit data.

[0282] If the memory 1610, processor 1620, and communication interface 1630 are implemented independently, they can be interconnected via a bus to communicate with each other. This bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. This bus can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 16 The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0283] Optionally, in a specific implementation, if the memory 1610, processor 1620 and communication interface 1630 are integrated on a single chip, the memory 1610, processor 1620 and communication interface 1630 can communicate with each other through an internal interface.

[0284] This application provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method provided in this application.

[0285] This application also provides a chip, which includes a processor for calling and executing instructions stored in a memory, causing a communication device on which the chip is installed to perform the method provided in this application.

[0286] This application also provides a chip, including: an input interface, an output interface, a processor, and a memory. The input interface, output interface, processor, and memory are connected through an internal connection path. The processor is used to execute code in the memory. When the code is executed, the processor is used to execute the method provided in the application embodiment.

[0287] It should be understood that the aforementioned processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. General-purpose processors can be microprocessors or any conventional processor. It is worth noting that the processor can be a processor supporting the Advanced Reduced Instruction Set Computing (RISC) machine (ARM) architecture.

[0288] Further, optionally, the aforementioned memory may include read-only memory and random access memory, and may also include non-volatile random access memory. The memory may be volatile or non-volatile, or may include both. Non-volatile memory may include read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which serves as an external cache. Many forms of RAM are available by way of example, but not limitation. Examples include static random access memory (SRAM), dynamic random access memory (DRAM), synchronous dynamic random access memory (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous linked dynamic random access memory (SLDRAM), and direct rambus RAM (DR RAM).

[0289] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. A computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the flow or function according to this application is generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transferred from one computer-readable storage medium to another.

[0290] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of those different embodiments or examples.

[0291] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this application, "a plurality of" means two or more, unless otherwise explicitly specified.

[0292] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process. Furthermore, the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functionality involved.

[0293] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus or device (such as a computer-based system, a processor-included system or other system that can fetch and execute instructions from, an instruction execution system, apparatus or device).

[0294] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. All or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware, the program being stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiments.

[0295] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. This storage medium can be a read-only memory, a disk, or an optical disk, etc.

[0296] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any person skilled in the art can easily conceive of various variations or substitutions within the technical scope disclosed in this application, and these should all be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for intelligent automation of object recommendation by combining a Robotic Process Automation (RPA) system and AI, characterized in that, include: Obtain the collection of bidding documents and tender documents through a Robotic Process Automation (RPA) system; A target knowledge representation model is used to obtain first knowledge representation information corresponding to the bidding document on the first knowledge graph, and to obtain second knowledge representation information corresponding to at least one bidding document in the bidding document set. The first knowledge graph corresponds to the bidding document and the bidding document set. A matching network model is used to obtain a matching degree set between the first knowledge representation information and at least one second knowledge representation information, and the target bidding document in the at least one bidding document is determined based on the matching degree set; The system recommends target bid objects corresponding to the target bid documents; wherein, when recommending target bid objects, the system provides interpretable and referable basis for the target bid objects through the recommendation explanation module; Before using a matching network model to obtain the matching degree set of the first knowledge representation information and the at least one second knowledge representation information respectively, the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set are obtained. The first knowledge representation information is adjusted using the user representation learning module and the bidding attribute information to obtain the adjusted first knowledge representation information; the second knowledge representation information corresponding to any bidding document is adjusted using the user representation learning module and the bidding attribute information to obtain the adjusted second knowledge representation information. Traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set.

2. The method according to claim 1, characterized in that, Before obtaining the first knowledge representation information corresponding to the bidding document on the first knowledge graph using the target knowledge representation model, the method further includes: The Robotic Process Automation (RPA) system is used to obtain a set of tender documents and bid documents in the target document format. The triplet extraction technique is used to obtain the triplet information set corresponding to the bidding document and the set of tender documents; Based on the triplet information set, a first knowledge graph corresponding to the bidding documents and the tender documents set is established.

3. The method according to claim 2, characterized in that, The step of using triplet extraction technology to obtain the triplet information set corresponding to the tender documents and the set of bid documents includes: The triplet extraction technique is used to obtain the first triplet information set corresponding to the bidding documents and the second triplet information set corresponding to the bid documents. The Jieba word segmentation model is used to segment the first triplet information set and the second triplet set to obtain the processed first triplet information set and the processed second triplet set. The processed first triplet information set and the processed second triplet set are subjected to entity boundary string verification processing to obtain the triplet information set corresponding to the bidding document and the tender document set.

4. The method according to claim 1, characterized in that, Before obtaining the first knowledge representation information corresponding to the bidding documents on the first knowledge graph using the target knowledge representation model, and before obtaining the second knowledge representation information corresponding to at least one bidding document in the bidding document set, the method further includes: Obtain a collection of bidding and tendering training documents; Obtain the knowledge graph corresponding to the set of bidding training documents; Based on distributed representation learning reasoning technology, at least one knowledge pair in the knowledge graph is mapped into a vector space to establish a target knowledge representation model.

5. The method according to claim 1, characterized in that, Before obtaining the first knowledge representation information corresponding to the bidding documents on the first knowledge graph using the target knowledge representation model, and before obtaining the second knowledge representation information corresponding to at least one bidding document in the bidding document set, the method further includes: Obtain the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set; Based on the model performance parameters, a target knowledge representation model is determined from the initial knowledge representation model set, and the target knowledge representation model is a distributed knowledge reasoning model.

6. The method according to claim 1, characterized in that, The step of determining the target bid document among the at least one bid document based on the matching degree set includes: Obtain the highest matching degree in the matching degree set; From the at least one bid document, obtain the bid document corresponding to the highest matching degree, and determine the bid document as the target bid document.

7. An object recommendation device that combines a Robotic Process Automation (RPA) system and AI to achieve intelligent automation, characterized in that, include: The document acquisition unit is used to acquire a set of bidding documents and tender documents through the Robotic Process Automation (RPA) system; An information acquisition unit is used to acquire first knowledge representation information corresponding to the bidding document on a first knowledge graph using a target knowledge representation model, and to acquire second knowledge representation information corresponding to at least one bidding document in the bidding document set, wherein the first knowledge graph corresponds to the bidding document and the bidding document set. The document determination unit is used to use a matching network model to obtain a matching degree set between the first knowledge representation information and the at least one second knowledge representation information, and to determine the target bidding document in the at least one bidding document based on the matching degree set; The object recommendation unit is used to recommend target bidding objects corresponding to the target bidding document; wherein, when recommending the target bidding object, the recommendation explanation module provides an explainable and referable basis for the target bidding object; The device further includes an attribute information acquisition unit, an information adjustment unit, and a set traversal unit, prior to acquiring the matching degree sets of the first knowledge representation information and the at least one second knowledge representation information using a matching network model: The attribute information acquisition unit is used to acquire the bidding attribute information corresponding to the bidding document and the bidding attribute information corresponding to any bidding document in the bidding document set. The information adjustment unit is used to adjust the first knowledge representation information using the user representation learning module and the bidding attribute information to obtain the adjusted first knowledge representation information; and to adjust the second knowledge representation information corresponding to any bidding document using the user representation learning module and the bidding attribute information to obtain the adjusted second knowledge representation information. The set traversal unit is used to traverse the second knowledge representation information set to obtain the adjusted second knowledge representation information set.

8. The apparatus according to claim 7, characterized in that, The device further includes a target document acquisition unit, a set extraction unit, and a graph building unit, prior to the information acquisition unit acquiring the first knowledge representation information corresponding to the bidding document: The target document acquisition unit is used to acquire a set of tender documents and bid documents in the target document format through the Robotic Process Automation (RPA) system. The set extraction unit is used to obtain the set of triplet information corresponding to the set of bidding documents and the set of tender documents by using triplet extraction technology. The graph building unit is used to build a first knowledge graph corresponding to the bidding documents and the tender documents based on the triplet information set.

9. The apparatus according to claim 8, characterized in that, The set extraction unit includes a set acquisition subunit, a word segmentation processing subunit, and an audit processing subunit. When the set extraction unit acquires the triplet information set corresponding to the tender document and the bid document set: The set acquisition subunit is used to acquire the first triplet information set corresponding to the bidding document and the second triplet information set corresponding to the bid document set using triplet extraction technology. The word segmentation processing subunit is used to perform word segmentation processing on the first triplet information set and the second triplet set using the Jieba word segmentation model to obtain the processed first triplet information set and the processed second triplet set. The audit processing subunit is used to perform entity boundary string audit processing on the processed first triplet information set and the processed second triplet set to obtain the triplet information set corresponding to the bidding document and the tender document set.

10. The apparatus according to claim 7, characterized in that, The device further includes a set acquisition unit, a graph acquisition unit, and a model building unit. Before the information acquisition unit acquires the first knowledge representation information corresponding to the bidding documents on the first knowledge graph using a target knowledge representation model, and acquires the second knowledge representation information corresponding to at least one bidding document in the bidding document set: The set acquisition unit is used to acquire a set of bidding training documents; The graph acquisition unit is used to acquire the knowledge graph corresponding to the bidding training document set; The model building unit is used to perform vector space mapping on at least one knowledge pair in the knowledge graph based on distributed representation learning reasoning technology to establish a target knowledge representation model.

11. The apparatus according to claim 7, characterized in that, The device further includes a parameter acquisition unit and a model determination unit. Before the information acquisition unit acquires the first knowledge representation information corresponding to the bidding documents on the first knowledge graph using a target knowledge representation model, and acquires the second knowledge representation information corresponding to at least one bidding document in the bidding document set: The parameter acquisition unit is used to acquire the model performance parameters corresponding to each initial knowledge representation model in the initial knowledge representation model set. The model determination unit is used to determine a target knowledge representation model from the initial knowledge representation model set based on the model performance parameters. The target knowledge representation model is a distributed knowledge reasoning model.

12. A terminal that combines a Robotic Process Automation (RPA) system with AI to achieve intelligent automation, characterized in that: include: A processor and a memory, wherein instructions are stored in the memory and loaded and executed by the processor to implement the method as claimed in any one of claims 1 to 6.

13. A computer-readable storage medium storing a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images