Multi-dimensional knowledge base driven cost unit price generation method and system

Abstract

This invention relates to the field of engineering cost calculation technology, and discloses a method and system for generating unit prices driven by a multi-dimensional knowledge base. The method includes: acquiring a shared basic knowledge layer and a user-owned private knowledge extension layer; receiving a list description and extracting unknown terms, then replacing the unknown terms with equivalent terms through semantic similarity calculation to generate a standardized list description; sequentially performing quota retrieval in the user-owned private knowledge extension layer and the shared basic knowledge layer to obtain a candidate quota set; calculating the engineering cost unit price based on the quota with the highest matching confidence, combined with regional price coefficients and time adjustment coefficients; receiving feedback and writing the corrected quota and terminology mapping relationship into the user-owned private knowledge extension layer. This invention achieves multi-user knowledge isolation management and solves the matching failure problem caused by the lack of terminology in emerging industries.

Description

Technical Field

[0001] This invention relates to the field of engineering cost calculation technology, and more specifically, to a method and system for generating unit prices of engineering costs driven by a multi-dimensional knowledge base. Background Technology

[0002] In SaaS-based engineering cost estimation platforms, the platform needs to serve multiple enterprise users whose businesses span different industries, including traditional municipal and building construction, as well as emerging industries such as hydrogen energy infrastructure and low-altitude economic support facilities. Existing technologies construct a knowledge base encompassing multi-dimensional lists and quotas, including industry, region, project type, building type, and building area. The platform uses a unified knowledge base version to provide list matching and cost calculation services to all enterprise users. However, this approach suffers from the following technical problems: the unified multi-dimensional knowledge base cannot differentiate between different enterprise users' terminology usage habits and quota mapping preferences; a single enterprise user's list-quota matching correction may interfere with the matching results of other enterprise users; emerging industries lack historical list-quota pairing records in the knowledge base, causing quota matching failures due to the inability of the list matching module to recognize industry-specific terminology; and when enterprise users report quota matching errors and submit corrections, the corrected knowledge is directly updated to the global knowledge base, polluting the matching results of other enterprise users and causing mutual interference between knowledge updates across multiple users. Summary of the Invention

[0003] This invention provides a method and system for generating unit prices of construction costs driven by a multi-dimensional knowledge base, which solves the technical problems in related technologies such as the inability to balance platform-shared knowledge with the personalized needs of enterprises, inconsistent terminology leading to low accuracy in quota matching, and the lack of enterprise-specific knowledge accumulation mechanisms.

[0004] This invention discloses a method for generating unit prices for construction costs driven by a multi-dimensional knowledge base, comprising the following steps: obtaining a hierarchical knowledge base structure, wherein the hierarchical knowledge base structure includes a shared basic knowledge layer and a user-private knowledge extension layer, wherein the shared basic knowledge layer contains a platform-level multi-dimensional list-quota relationship and a set of standard terms, and the user-private knowledge extension layer stores terminology mapping relationships and quota matching preference records specific to corresponding enterprise users; receiving a new project list description and project information submitted by an enterprise user, extracting unknown terms from the new project list description, calculating the semantic similarity between the unknown terms and the standard terms in the shared basic knowledge layer, and replacing the unknown terms with the semantically closest equivalent terms based on the semantic similarity. The process involves generating a standardized list description; based on the standardized list description and the project information, quota retrieval is performed sequentially in the user's private knowledge extension layer and the shared basic knowledge layer, with the retrieval priority of the user's private knowledge extension layer being higher than that of the shared basic knowledge layer, to obtain a set of candidate quotas that match the standardized list description; based on the quota with the highest matching confidence in the candidate quota set, and combined with the price adjustment coefficient in the project information, a unit price for engineering cost is calculated and generated; feedback information from enterprise users regarding the unit price for engineering cost is received, and when the feedback information indicates a quota matching error, the matching relationship between the corrected quota and the corresponding list description is written into the user's private knowledge extension layer.

[0005] Furthermore, the multidimensional list-quota relationship in the shared basic knowledge layer refers to the correspondence data between list items and quota items organized according to the dimensions of industry, region, project type, building type and building area; the standard terminology set includes standardized terms of various mature industries and their semantic relationships, and the semantic relationships are expressed as synonyms, hierarchical or related associations between terms.

[0006] Furthermore, the storage content of the user private knowledge extension layer includes two types of data: the first type is the mapping relationship between the enterprise user's industry-specific terms and the standard terms of the shared basic knowledge layer; the second type is the list-quota matching preference record confirmed by the enterprise user.

[0007] Furthermore, the step of extracting unknown terms from the description of the newly created project list includes: performing word segmentation on the text of the description of the newly created project list, extracting noun phrases and professional term candidates; performing a matching query between the extracted term candidates and the standard term set of the shared basic knowledge layer; and marking term candidates that cannot find a match in the standard term set as unknown terms.

[0008] Furthermore, calculating the semantic similarity between the unknown term and the standard terms in the shared basic knowledge layer includes: converting the unknown term and the standard terms in the shared basic knowledge layer into semantic vector representations respectively; calculating the cosine similarity between the semantic vector of the unknown term and the semantic vector of each standard term; and selecting the standard term with the highest cosine similarity as the equivalent term of the unknown term.

[0009] Furthermore, quota retrieval is performed sequentially in the user private knowledge extension layer and the shared basic knowledge layer, including: firstly, searching in the user private knowledge extension layer for whether there is a quota preference record that matches the description of the normalized list; if it exists, the quota corresponding to the quota preference record is directly returned as the matching result; if it does not exist, multi-dimensional retrieval is performed in the shared basic knowledge layer according to the dimensions of industry, region, project type, building type and building area, and a set of candidate quotas that meet the dimensional constraints is returned.

[0010] Furthermore, the quotas in the candidate quota set are sorted according to the matching confidence score, which is calculated based on the text similarity and dimensional matching degree between the list description and the quota description.

[0011] Furthermore, the price adjustment coefficient includes a regional price coefficient and a time adjustment coefficient; the calculation of the unit price of the project cost includes: obtaining the benchmark unit price of the quota corresponding to the quota with the highest matching confidence in the candidate quota set; multiplying the benchmark unit price of the quota by the regional price coefficient and the time adjustment coefficient to obtain the final unit price of the project cost.

[0012] Furthermore, the step of writing the matching relationship between the modified quota and the corresponding list description into the user's private knowledge extension layer also includes: writing the mapping relationship between the unknown term and its equivalent term into the user's private knowledge extension layer; wherein, the update of the user's private knowledge extension layer adopts an incremental writing method, and the newly written matching relationship does not overwrite the original data in the shared basic knowledge layer; when the same user submits different modified quotas multiple times for the same list description, the user's private knowledge extension layer retains the most recently submitted modified quota as the valid matching preference.

[0013] This invention provides a multi-dimensional knowledge base-driven cost unit price generation system, comprising: a hierarchical knowledge base acquisition module for acquiring a hierarchical knowledge base structure, the hierarchical knowledge base structure including a shared basic knowledge layer and a user-private knowledge extension layer; a terminology bridging processing module for receiving a new project list description and project information submitted by an enterprise user, extracting unknown terms and replacing the unknown terms with equivalent terms based on semantic similarity to generate a standardized list description; a hierarchical quota retrieval module for performing quota retrieval in the user-private knowledge extension layer and the shared basic knowledge layer sequentially based on the standardized list description and the project information to obtain a candidate quota set; a unit price generation module for calculating and generating the engineering cost unit price based on the quota with the highest confidence in the candidate quota set, combined with a price adjustment coefficient; and a feedback update module for receiving feedback information from the enterprise user, and when the feedback information indicates a quota matching error, writing the corrected quota and the corresponding list description matching relationship into the enterprise user's user-private knowledge extension layer.

[0014] This invention achieves isolated management of knowledge among multiple users by constructing a hierarchical knowledge base structure consisting of a shared basic knowledge layer and a user-private knowledge extension layer, thus solving the technical problem of mutual interference between quota matching preferences of different enterprise users. Through terminology bridging processing, it maps unknown terms in emerging industries to equivalent terms in mature industries, resolving the problem of quota matching failures caused by missing terms in emerging industries. By writing corrective knowledge feedback from enterprise users into the user-private knowledge extension layer instead of the shared basic knowledge layer, it avoids contaminating the matching results of other enterprise users through knowledge updates, achieving the technical effect of improving the accuracy and applicability of engineering cost unit price generation in multi-user scenarios. Attached Figure Description

[0015] Figure 1 This is a flowchart of the multi-dimensional knowledge base-driven cost unit price generation method provided in this embodiment of the invention; Figure 2 This is a term semantic similarity matching analysis diagram provided in the embodiments of the present invention; Figure 3 This is a candidate quota matching confidence evaluation diagram provided in an embodiment of the present invention; Figure 4 This is a flowchart of the engineering cost unit price calculation provided in an embodiment of the present invention; Figure 5 This is a flowchart of the hierarchical knowledge retrieval path provided in an embodiment of the present invention; Figure 6 This is a comparison and analysis chart of the standard unit price provided in the embodiments of the present invention; Figure 7 This is a growth trend chart of the user's private knowledge layer provided in an embodiment of the present invention; Figure 8This is a weight distribution diagram for calculating matching confidence provided in an embodiment of the present invention. Detailed Implementation

[0016] This implementation method relates to the field of engineering cost calculation technology, specifically to a method for generating unit prices driven by a multi-dimensional knowledge base.

[0017] In SaaS-based engineering cost estimation platforms, the platform needs to serve multiple enterprise users whose businesses span different industries, including traditional municipal and building construction, as well as emerging industries such as hydrogen energy infrastructure and low-altitude economy supporting facilities. Current technologies construct a knowledge base covering multiple dimensions such as industry, region, project type, building type, and building area, encompassing bill of quantities and quota relationships. The platform uses a unified version of this knowledge base to provide bill of quantities matching and cost calculation services to all enterprise users.

[0018] The above technical solutions have the following technical problems in practical applications: First, the unified multidimensional knowledge base cannot distinguish the terminology usage habits and quota mapping preferences of different enterprise users. The list-quota matching corrections accumulated by one enterprise user in a specific industry project may interfere with the matching results of other enterprise users in different industry projects. Second, emerging industries lack historical list-quota matching records in the knowledge base. The list matching module fails to recognize the unique terminology of emerging industries, resulting in quota matching failure and thus unit price generation failure. Third, when enterprise users report quota matching errors and submit corrections, the corrected knowledge is directly updated to the global knowledge base, which will pollute the matching results of other enterprise users and cause mutual influence between knowledge updates among multiple users.

[0019] At least one embodiment of the present invention discloses a method for generating unit prices of construction costs driven by a multi-dimensional knowledge base, such as... Figure 1 As shown, it includes the following steps: Step 1: Obtain the hierarchical knowledge base structure; The system acquires a shared basic knowledge layer and a user-private knowledge extension layer. The shared basic knowledge layer contains a platform-level multidimensional list-quota relationship and a set of standard terms for various mature industries. The user-private knowledge extension layer stores the terminology mapping relationship and quota matching preference records specific to the corresponding enterprise users.

[0020] It should be noted that the multidimensional list-quota relationship in the shared basic knowledge layer refers to the correspondence data between list items and quota items organized according to dimensions such as industry, region, project type, building type, and building area. The standard terminology set in the shared basic knowledge layer includes standardized terms from various mature industries and their semantic relationships, which are expressed as synonyms, hierarchical relationships, or related associations between terms.

[0021] It should be noted that the user's private knowledge extension layer is initialized to empty and will be incrementally written to in subsequent steps based on feedback from enterprise users. The stored content of the user's private knowledge extension layer includes two types of data: the first type is the mapping relationship between the enterprise user's industry-specific terminology and the standard terminology of the shared basic knowledge layer; the second type is the list-quota matching preference record confirmed by the enterprise user.

[0022] Step 2: Receive the list description and perform terminology extraction and bridging processing; The system receives descriptions and information of newly created projects submitted by enterprise users, extracts unknown terms from the descriptions, calculates the semantic similarity between the unknown terms and standard terms in the shared basic knowledge layer, replaces the unknown terms with the most semantically similar mature industry equivalent terms based on the semantic similarity, and generates a standardized list description after term bridging.

[0023] It should be noted that unknown terms refer to terms that appear in the description of a new project list but do not have a matching item in the standard terminology set of the shared basic knowledge layer. These are usually terms specific to emerging industries.

[0024] Furthermore, the extraction of unknown terms is achieved through text segmentation and terminology recognition. The specific process is as follows: the description text of the newly created project list is segmented to extract noun phrases and professional term candidates. The extracted term candidates are matched with the standard terminology set of the shared basic knowledge layer. Term candidates that cannot be matched in the standard terminology set of the shared basic knowledge layer are marked as unknown terms.

[0025] Furthermore, the semantic similarity calculation adopts a vector space model. The specific process is as follows: the unknown term and the standard term in the shared basic knowledge layer are converted into semantic vector representations respectively, the cosine similarity between the unknown term vector and each standard term vector is calculated, and the standard term with the highest cosine similarity is selected as the equivalent term of the unknown term.

[0026] Step 3: Perform hierarchical knowledge retrieval and quota matching; Based on the standardized list description and project information, quota retrieval is performed sequentially in the user's private knowledge extension layer and the shared basic knowledge layer. The retrieval priority of the user's private knowledge extension layer is higher than that of the shared basic knowledge layer, and a set of candidate quotas matching the standardized list description is obtained.

[0027] It should be noted that the execution logic of the hierarchical retrieval is as follows: First, the user's private knowledge extension layer is searched to see if there is a quota preference record that matches the description of the standardized list. If there is, the quota corresponding to the preference record is returned directly as the matching result. If there is no matching record in the user's private knowledge extension layer, the shared basic knowledge layer is searched in multiple dimensions such as industry, region, project type, building type and building area to return a set of candidate quotas that meet the dimensional constraints.

[0028] Furthermore, the quotas in the candidate quota set are sorted according to the matching confidence score, which is calculated based on the textual similarity and dimensional matching degree between the list description and the quota description.

[0029] Step 4: Generate the unit price for the project cost; Based on the quota with the highest confidence level matched from the candidate quota set, and combined with the regional price coefficient and time adjustment coefficient in the project information, the unit price of the project cost is calculated and generated, and the unit price of the project cost is returned to the enterprise user.

[0030] It should be noted that the formula for calculating the unit price of the project cost is as follows:

[0031] in, This indicates the final unit price of the project cost. This represents the standard unit price. Indicates the regional price coefficient. This represents the time adjustment factor.

[0032] Step 5: Receive feedback and update the user's private knowledge extension layer; Receive feedback from enterprise users regarding the unit price of engineering costs. When the feedback indicates an error in the quota matching, receive the revised quota submitted by the enterprise user, write the matching relationship between the revised quota and the corresponding list description into the enterprise user's user private knowledge extension layer, and simultaneously write the mapping relationship between the unknown terms identified in step 2 and their equivalent terms into the enterprise user's user private knowledge extension layer.

[0033] It should be noted that the update of the user's private knowledge extension layer adopts an incremental writing method. The newly written matching relationship does not overwrite the original data in the shared basic knowledge layer, and only takes effect in the subsequent retrieval process of the enterprise user.

[0034] Furthermore, when the same enterprise user submits different revised quotas multiple times for the same list description, the user's private knowledge extension layer retains the most recently submitted revised quota as the valid matching preference.

[0035] This implementation achieves isolated management of knowledge among multiple users by constructing a hierarchical knowledge base structure consisting of a shared basic knowledge layer and a user-private knowledge extension layer, thus avoiding interference between quota matching preferences of different enterprise users. Through terminology bridging processing, unknown terms in emerging industries are mapped to equivalent terms in mature industries, resolving the quota matching failure problem caused by missing terms in emerging industries. By writing corrective knowledge feedback from enterprise users into the user-private knowledge extension layer instead of the shared basic knowledge layer, knowledge updates are prevented from contaminating the matching results of other enterprise users.

[0036] The embodiments of the present invention have been described above. However, the embodiments are not limited to the specific implementation methods described above. The specific implementation methods described above are merely illustrative and not restrictive. Those skilled in the art can make more equivalent embodiments under the guidance of the present embodiments, and all of them are within the protection scope of the present embodiments.

[0037] A certain engineering cost SaaS platform serves multiple construction companies. Company A primarily undertakes traditional building and municipal engineering projects, while Company B focuses on hydrogen energy infrastructure construction. In March 20XX, Company B constructed a hydrogen refueling station project on the platform. Located in an industrial park in East China, the project covers an area of ​​800 square meters and is classified as new energy infrastructure. When compiling the bill of quantities, Company B entered the item "high-pressure hydrogen storage tank foundation pouring," requiring the platform to automatically generate the unit price for this item. Because the hydrogen energy industry is an emerging field, the platform's shared knowledge base lacks historical matching records for this term. Therefore, the system needs to generate the unit price through terminology bridging and hierarchical knowledge retrieval, and then write the corrected knowledge into Company B's private knowledge layer based on user feedback.

[0038] The platform initializes a layered knowledge base structure for User Company B. The shared basic knowledge layer contains list-quota relationship data for mature industries such as traditional municipal engineering, building construction, and petrochemical engineering, covering the standard quota library for East China, and storing standardized terminology and semantic relationships for each mature industry. User Company B's private knowledge extension layer is initially empty and is prepared to receive the company's proprietary terminology mapping and quota matching preferences accumulated in hydrogen energy projects.

[0039] Table 1. Examples of Standard Terminology for the Shared Basic Knowledge Layer:

[0040] Table 2 Initial state of User B Company's private knowledge extension layer:

[0041] The system receives a new project list description, "High-pressure hydrogen storage tank foundation pouring," submitted by User Company B. Simultaneously, it retrieves project information: industry: hydrogen energy; region: East China; project type: new energy infrastructure; building area: 800 square meters. The system performs text segmentation on the list description, extracting the noun phrases "high-pressure hydrogen storage tank" and "foundation pouring." It then performs a matching query against the shared basic knowledge layer's standard terminology set. Finding no match for "high-pressure hydrogen storage tank" in the standard terminology set, it marks it as an unknown term.

[0042] The system converts the unknown term "high-pressure hydrogen storage tank" into a semantic vector and calculates its cosine similarity with the standard term vectors in the shared basic knowledge layer.

[0043] Table 3. Results of semantic similarity calculation for unknown terms:

[0044] The system selects the standard term "tank foundation construction" (similarity 0.78) with the highest cosine similarity as the equivalent term for the unknown term "high-pressure hydrogen storage tank". The original list description "high-pressure hydrogen storage tank foundation pouring" is replaced with the standardized list description "tank foundation construction (pouring type)". The term mapping relationship "high-pressure hydrogen storage tank → tank foundation construction" is generated and will be written into the user's private layer later.

[0045] Table 4 Terminology Bridging Conversion Records:

[0046] The system first searches for matching records for "storage tank foundation construction (pouring type)" in User B's private knowledge extension layer. Since this layer is initially empty, the search yields no results. The system then switches to the shared basic knowledge layer and performs a multi-dimensional search based on dimensions such as industry (petrochemical engineering - storage tank related), region (East China), project type (industrial facilities), and building area (800 square meters), returning a set of candidate quotas that meet the dimensional constraints.

[0047] Table 5: Search Results of Candidate Quotas for Shared Basic Knowledge Layer

[0048] The system calculates the matching confidence score for each candidate quota using the following formula:

[0049] Taking the fixed quota QUOTA201 as an example:

[0050] The system selects the quota QUOTA201 with the highest matching confidence as the matching result.

[0051] The system is based on the benchmark unit price of quota QUOTA201. Yuan / cubic meter, combined with the price coefficient for East China region based on project information. Time adjustment factor for the first quarter of 20XX The unit price of the project cost is generated according to the unit price calculation formula.

[0052] Table 6 Unit Price Calculation Parameters:

[0053] Unit price calculation process:

[0054] The system generated a final project cost of 799.68 yuan per cubic meter and returned it to user company B.

[0055] Table 7: Results of Unit Price Generation for Engineering Costs

[0056] After reviewing the unit price generation results, the engineers of User Company B discovered that the high-pressure hydrogen storage tank foundation in the hydrogen energy project, due to its high bearing pressure, requires the use of C35 high-strength concrete instead of C30 concrete. Therefore, the appropriate quota should be QUOTA256 (C35 concrete pouring for tank foundation, base unit price 720.00 yuan / cubic meter). User Company B submitted feedback through the platform's feedback function, indicating the quota mismatch, and submitted a revised quota QUOTA256 along with the reasons for the correction.

[0057] Table 8 User Feedback and Correction Records:

[0058] After receiving the feedback, the system will write the corrected quota matching relationship "Storage tank foundation construction (pouring) → QUOTA256" into the private knowledge extension layer of user company B, and at the same time write the term mapping relationship "High pressure hydrogen storage tank → Storage tank foundation construction" identified in step 2 into the user's private knowledge extension layer.

[0059] Table 9: Status of User B Company's Private Knowledge Extension Layer after Update:

[0060] When User Company B subsequently submits a list containing "high-pressure hydrogen storage tank foundation pouring," the system will prioritize retrieving the quota preference record from User Company B's private knowledge extension layer during the hierarchical search in step 3, and directly return the corrected quota QUOTA256 to avoid duplicate matching errors. This corrected knowledge only applies to User Company B and will not affect the quota matching results of other users such as User Company A in traditional municipal projects.

[0061] Throughout the implementation process, the data begins with the original bill of quantities submitted by User Company B, describing "high-pressure hydrogen storage tank foundation pouring" and project information (industry: hydrogen energy, region: East China, building area: 800 square meters). The system performs word segmentation and terminology recognition on the bill of quantities description, extracting the unknown term "high-pressure hydrogen storage tank." It then bridges this term to the standard term "tank foundation construction" by calculating the semantic vector cosine similarity (maximum value 0.78), generating a standardized bill of quantities description "tank foundation construction (pouring type)." The system sequentially searches the user's private layer (initially empty, no results) and the shared foundation layer in the hierarchical knowledge base. From the shared foundation layer, it retrieves a set of candidate quotas based on industry, region, and other dimensions, selecting the quota with the highest matching confidence (0.89) QUOTA201 (base price 680.00 yuan / cubic meter). Combining the regional price coefficient of 1.12 and the time adjustment coefficient of 1.05, it follows the formula... The system calculates and returns a unit price of 799.68 yuan / cubic meter. After User Company B reports a quota matching error and submits a revised quota QUOTA256, the system writes the terminology mapping relationship "high-pressure hydrogen storage tank → tank foundation construction" and the quota preference "tank foundation construction (pouring) → QUOTA256" into User Company B's private knowledge extension layer. This achieves knowledge isolation and incremental learning, ensuring that subsequent similar lists directly match the revised quota and avoiding cross-user knowledge contamination.

Claims

1. A method for generating unit prices for construction costs driven by a multi-dimensional knowledge base, characterized in that, Includes the following steps: Obtain a hierarchical knowledge base structure, which includes a shared basic knowledge layer and a user-private knowledge extension layer. The shared basic knowledge layer contains a platform-level multi-dimensional list-quota relationship and a set of standard terms. The user-private knowledge extension layer stores terminology mapping relationships and quota matching preference records specific to the enterprise user. Receive a new project list description and project information submitted by the enterprise user; extract unknown terms from the new project list description; calculate the semantic similarity between the unknown terms and the standard terms in the shared basic knowledge layer; replace the unknown terms with the semantically closest equivalent terms based on the semantic similarity; and generate a standardized list description. The standardized list description and the project information are sequentially searched for quotas in the user's private knowledge extension layer and the shared basic knowledge layer, with the search priority of the user's private knowledge extension layer being higher than that of the shared basic knowledge layer. A set of candidate quotas matching the standardized list description is obtained. Based on the quota with the highest matching confidence in the candidate quota set, and combined with the price adjustment coefficient in the project information, the unit price of the project cost is calculated and generated. Feedback information from enterprise users on the unit price of the project cost is received. When the feedback information indicates that the quota matching is incorrect, the matching relationship between the corrected quota and the corresponding list description is written into the user's private knowledge extension layer.

2. The method according to claim 1, characterized in that, The multidimensional list-quota relationship in the shared basic knowledge layer refers to the correspondence data between list items and quota items organized according to the dimensions of industry, region, project type, building type and building area; the standard terminology set includes standardized terms of various mature industries and their semantic relationships, and the semantic relationships are expressed as synonyms, hierarchical or related associations between terms.

3. The method according to claim 1, characterized in that, The storage content of the user private knowledge extension layer includes two types of data: the first type is the mapping relationship between the industry-specific terms of the enterprise user and the standard terms of the shared basic knowledge layer; The second category is the list of fixed-quota matching preferences confirmed by the enterprise user.

4. The method according to claim 1, characterized in that, The step of extracting unknown terms from the description of the new project list includes: performing word segmentation on the text of the description of the new project list, extracting noun phrases and professional term candidates; performing a matching query between the extracted term candidates and the standard term set of the shared basic knowledge layer; and marking term candidates that cannot find a match in the standard term set as unknown terms.

5. The method according to claim 1, characterized in that, The step of calculating the semantic similarity between the unknown term and the standard terms in the shared basic knowledge layer includes: converting the unknown term and the standard terms in the shared basic knowledge layer into semantic vector representations respectively; calculating the cosine similarity between the semantic vector of the unknown term and the semantic vector of each standard term; and selecting the standard term with the highest cosine similarity as the equivalent term of the unknown term.

6. The method according to claim 1, characterized in that, The step of sequentially performing quota retrieval in the user's private knowledge extension layer and the shared basic knowledge layer includes: firstly, searching in the user's private knowledge extension layer for whether there is a quota preference record that matches the description of the normalized list; if there is, directly returning the quota corresponding to the quota preference record as the matching result; if there is not, performing a multi-dimensional search in the shared basic knowledge layer according to the dimensions of industry, region, project type, building type and building area, and returning a set of candidate quotas that meet the dimensional constraints.

7. The method according to claim 1, characterized in that, The quotas in the candidate quota set are sorted according to the matching confidence score, which is calculated based on the text similarity and dimensional matching degree between the list description and the quota description.

8. The method according to claim 1, characterized in that, The price adjustment coefficient includes a regional price coefficient and a time adjustment coefficient; The calculation of the unit price of the project cost includes: obtaining the benchmark unit price of the quota corresponding to the quota with the highest matching confidence in the candidate quota set; The final project cost unit price is obtained by multiplying the benchmark unit price by the regional price coefficient and the time adjustment coefficient.

9. The method according to claim 1, characterized in that, The step of writing the matching relationship between the modified quota and the corresponding list description into the user's private knowledge extension layer further includes: writing the mapping relationship between the unknown term and its equivalent term into the user's private knowledge extension layer; wherein, the update of the user's private knowledge extension layer adopts an incremental writing method, and the newly written matching relationship does not overwrite the original data in the shared basic knowledge layer; when the same user submits different modified quotas multiple times for the same list description, the user's private knowledge extension layer retains the most recently submitted modified quota as the valid matching preference.

10. A multi-dimensional knowledge base-driven cost unit price generation system, used to execute the method described in any one of claims 1 to 9, characterized in that, include: A layered knowledge base acquisition module is used to acquire the layered knowledge base structure, which includes a shared basic knowledge layer and a user private knowledge extension layer. The terminology bridging processing module is used to receive the new project list description and project information submitted by enterprise users, extract unknown terms and replace the unknown terms with equivalent terms based on semantic similarity, and generate a standardized list description. The hierarchical quota retrieval module is used to perform quota retrieval in the user private knowledge extension layer and the shared basic knowledge layer in sequence based on the standardized list description and the project information to obtain a candidate quota set; The unit price generation module is used to calculate and generate the unit price of the project cost based on the quota with the highest confidence level matched in the candidate quota set and combined with the price adjustment coefficient. The feedback update module is used to receive feedback information from enterprise users. When the feedback information indicates that the quota matching is incorrect, the corrected quota and the matching relationship between the quota and the corresponding list description are written into the enterprise user's private knowledge extension layer.

Images