Universal digital project function similarity comparison method
By building an intelligent project review system and adopting a digital project function similarity comparison method, the problem of repeated functions of enterprises in digital project management is solved, automatic judgment and repetitive calculation are realized, and work efficiency is improved.
Patent Information
- Application Number
- CN202411939255.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-12-26
- Publication Date
- 2025-04-18
AI Technical Summary
Major enterprises lack effective repetitive review methods in digital project management, resulting in the problem of duplicate construction of digital project functions.
A project intelligent review system is built, and a general digital project function similarity comparison method is used to realize the accurate identification and repetitive calculation of the digital project function through the function recognition in step S1, the functional objectification processing in step S2 and the functional similarity calculation in step S3.
It solves the pain points of repetitive review of digital projects, realizes automatic judgment and repetitive calculation of digital project functions, reduces manual repetitive work, and improves work efficiency.
Smart Images
Figure CN119850137A_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of artificial intelligence and relates to a general digital project functional similarity comparison method. Background Art
[0002] In the context of the booming development of various industries in China, with the rapid rise of large enterprises in various fields and the rapid expansion of their businesses, digital transformation has become a key strategy to improve efficiency and quality. It is widely used in all aspects of corporate operations, from manufacturing to customer service, helping companies maintain their leading position in the fierce market competition.
[0003] At present, major enterprises have built digital project management systems, including investment direction decision-making and planning, project plan management, project execution monitoring, project effectiveness evaluation, etc. However, there is still a lack of effective digital support for the effectiveness, necessity, and repeatability review of project investment. Although there are relatively mature cases of repeatability review of papers and patents in society, there is still a lack of relevant methodological research on the repeatability review of digital project functions in enterprises.
[0004] As the scale of investment and number of digital projects of major enterprises increase year by year, enterprises have more urgent needs for digital project management and higher requirements for the degree of digitalization. Existing digitalization may have duplication of construction between various management units of enterprises, so enterprises need to be able to identify the content of digital project construction and coordinate and arrange it in an overall manner.
[0005] Therefore, the repetitive review of digital projects in major enterprises has become a long-standing pain point. The crux of the problem lies in the technical difficulty of how to identify the functions of digital projects and realize their repetitive calculations, which cannot be solved. Summary of the invention
[0006] In order to solve the pain point problem of repetitive review of digital projects in major enterprises, the present invention constructs a project intelligent review system to realize the repetitive review function of digital projects.
[0007] The technical solution adopted by the present invention is a general method for comparing the functional similarity of digital projects. The key is that the above comparison method is completed by step S1 of identifying the digital project functions, step S2 of objectifying the functions, and step S3 of calculating the functional similarity:
[0008] Step S1 is to identify and divide the project into functions, construct a function list and describe the function list in a hierarchical manner, with the lowest level of hierarchical description being function points;
[0009] The processing process in step S2 is to convert the function list into function objects, function title objects and function description objects with unique codes respectively, including three aspects: S2-1 function object processing, S2-2 function title object processing and S2-3 function description object processing; the above-mentioned function objects include title, description, search path, parent node, level, whether it is a leaf node, and child node object array, and then store the function objects in the function library; the above-mentioned function title object is to split the function title into business subject, business action and business limitation; the above-mentioned function description object is a detailed function description of the function title; the above-mentioned function point object is the lowest level function object in the function object and can only be used as a leaf node. The function title of the function point object is split to obtain the function point title object, and the function point title object is described in detail to obtain the function point description object;
[0010] In step S3, similarity calculation is performed in a bottom-up manner to obtain function title similarity, function description similarity and function similarity respectively, and finally obtain similarity values of all function objects.
[0011] Specifically, the above-mentioned identification in step S1 is to identify the project by project number or title, recursively search for the function point node and record the hierarchical path as the retrieval path of the function point; the above-mentioned functional division in step S1 is divided in the order of title, functional classification, and functional point, and the above-mentioned functional points are described according to hierarchical requirements, hierarchical functions, functional descriptions and status content.
[0012] Specifically, in step S2, the business subject mentioned above is the object or field of the function, the business action mentioned above is the specific operation or display method performed on the business subject, and the business limitation mentioned above is a description of the supplement, enhancement or limitation of the function.
[0013] Specifically, the bottom-up approach in step S3 is to first calculate the similarity of each function point object, and then construct the upper-level function similarity step by step based on the similarity results of the function points. The similarity results of the above function points are calculated based on the function title similarity and function description similarity results, among which the similarity calculation carried out on a single function is the single function similarity.
[0014] More specifically, the function title similarity is calculated based on the business subject similarity, business action similarity, and business limitation similarity. The similarity between the current function title object and other function title objects in the function library is calculated and the highest value is taken as the current function title similarity result; the function description similarity result is calculated using a semantic similarity algorithm, using one or more of the function description object, function point object, and function object as input.
[0015] Preferably, the above-mentioned single function similarity is calculated according to Formula 1 and Formula 2:
[0016]
[0017] In Formula 1 or Formula 2, SFS represents the similarity of a single function, FTS represents the similarity of function title, and W FTS represents the weight of FTS; FPCS represents the similarity of function point description, W FPCS represents the weight of FPCS; Formula 1 is for the functional similarity calculation of function F, and Formula 2 is for the similarity calculation of several secondary functions F 11 , F 12 , ..., F 1n The first-level function F1 is composed of;
[0018] The calculation formulas for the above function title similarity are Formula 3 and Formula 4:
[0019]
[0020] In Formula 3 or Formula 4, FTS represents the similarity of functional titles, BE represents the similarity of business entities, and W BE represents the weight of BE; BA represents the similarity of business actions, W BA represents the weight of BA; BQ represents the business-limited similarity, W BQ represents the weight of BQ; BEQ represents the comprehensive subject similarity, W BEQ Represents the weight of BEQ.
[0021] Preferably, the above W FTS +W FPCS =1,W BE +W BQ =1,W BEQ +W BA =1.
[0022] Furthermore, the calculation order of the function similarity calculation in step S3 is to first calculate the function point title similarity, the function point description similarity, and the function point similarity in sequence, and then calculate the function similarity of the previous level of the function point in the order of function title similarity, function description similarity, and function similarity, and finally obtain the function similarity step by step.
[0023] Furthermore, the evaluation principle of the results of function similarity, function title similarity and function description similarity is a scoring principle, specifically measuring the similarity in percentage, 0 points means there is no correlation at all, 100 points means the two are completely consistent, and the degree of similarity is defined for different score ranges.
[0024] It should be noted that the calculation principles of functional similarity are the principle of burden reduction, the principle of short circuit and the principle of optimality:
[0025] The above-mentioned principle of reducing the burden is that when computing power is limited, the function list participating in the function similarity calculation and comparison should be as small as possible;
[0026] The above short-circuit principle is that when the function title similarity is too low, the algorithm triggers the short-circuit effect and no longer calculates the function description similarity. The function point score will directly adopt the title similarity calculation result;
[0027] The above-mentioned optimal principle is that the function title similarity adopts an algorithm that enhances the ability to replace synonyms and antonyms, and the function description similarity adopts a semantic similarity algorithm that enhances the ability to understand the meaning of the entire paragraph, and uses the corresponding corpus for continuous optimization.
[0028] The beneficial effects of the present invention are:
[0029] The present invention solves the pain point of repetitive review of digital projects of major enterprises, builds a project intelligent review system, and realizes the repetitive review function of digital projects. More importantly, the present invention overcomes the technical difficulty of how to identify the functions of digital projects and realize their repetitive calculation.
[0030] The present invention proposes a universal method for comparing the functional similarity of digital projects. By studying the decomposition of digital project functions, the main features that distinguish each function are found, and a method for calculating the functional similarity in a scientific and quantitative way is discussed, providing a basis for automatic judgment of digital project functions.
[0031] The general digital project function similarity comparison method proposed in the present invention points out how to accurately identify the functions of digital projects in a scientific, simple and effective way, replacing manual repetitive work, so that staff can focus on the review of the business content achieved by the digital project itself. BRIEF DESCRIPTION OF THE DRAWINGS
[0032] Figure 1 This is a schematic diagram of the main components of the digitization project.
[0033] Figure 2 Schematic diagram of the calculation sequence of functional similarity - comparing functional titles.
[0034] Figure 3 Schematic diagram of the calculation sequence of functional similarity - comparing functional content.
[0035] Figure 4 Schematic diagram of the calculation sequence of functional similarity - comparing functional points.
[0036] Figure 5 Schematic diagram of the calculation sequence of functional similarity - comparing secondary functions.
[0037] Figure 6Schematic diagram of the calculation sequence of functional similarity - comparing first-level functions.
[0038] Figure 7 Flowchart for functional similarity calculation. DETAILED DESCRIPTION
[0039] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0040] A digital project is mainly composed of four parts: basic project information, requirements, functions, and architecture. The basic project information describes the project name, unit, necessity, expected results, implementation scope, etc.; the requirements describe the problems that the project needs to solve from a business perspective; the functions describe the construction content of the current project from an implementation perspective; and the architecture describes the architectural implementation in terms of business, application, technology, data, security, etc.
[0041] Among them, function occupies a core position in the project. It is the key to realizing project requirements and reflecting project value. It is directly related to whether the final product can meet user expectations and solve practical problems.
[0042] The main components of the digitalization project are as follows: Figure 1 shown.
[0043] As the main body that carries the content of digital project construction, function has the characteristics of clear content, comprehensive description and hierarchical structure.
[0044] Based on the above characteristics, the comparison of digital project function similarity is completed through three steps: step S1 digital project function identification, step S2 function object processing and step S3 function similarity calculation, specifically:
[0045] Step S1: Identification of digital project functions:
[0046] This step is to identify and divide the project into functions, build a function list and describe it hierarchically. The function list can be divided into multiple levels of description, where the lowest level of description is function point. Function point can be described according to the requirements, primary function, secondary function, function description, status, etc. If necessary, third-level function, fourth-level function, etc. can be added;
[0047] When identifying the function list, due to different management requirements of each unit, the level and granularity of the function division are different, but structurally, they all follow the "title → function classification → function point" method. Therefore, identification is to identify the project by project number or title, recursively search for the function point node and record the hierarchical path as the retrieval path of the function point; the function division is divided in the order of title, function classification, and function point, and the function point is described according to hierarchical requirements, hierarchical functions, function descriptions, and status content.
[0048] Tables 1 to 3 are three typical function list formats:
[0049] Format 1: The table is divided into primary and secondary functions, and multiple minimum function points are maintained in the function description. The specific function point is "batch import", and its search path is "XX declaration → large XX declaration → batch import".
[0050] Table 1: Classic function list format 1
[0051]
[0052] Format 2: The table distinguishes between primary, secondary and tertiary functions. Among them, the tertiary function is the smallest function point, and the search path for the specific function point "land use planning approval" is "XX approval → XX government docking approval → land use planning approval".
[0053] Table 2: Classic function list format 2
[0054]
[0055] Format 3: The table distinguishes between primary, secondary, and tertiary functions, and maintains multiple minimum function points in the function description. The search path for the specific function point "Vehicle Mileage Overview" is "XX Resource Overview → Company Vehicle Resource Overview → Vehicle Condition Overview → Vehicle Mileage Overview".
[0056] Table 3: Classic function list format 3
[0057]
[0058] Step S2: Function objectification processing:
[0059] This step needs to create a function point object by using the function points, search paths and other information obtained from the identification function list. The processing process is to convert the function list into a function object with a unique code, a function title object and a function description object. It is specifically divided into three aspects: S2-1 function object processing, S2-2 function title object processing and S2-3 function description object processing.
[0060] S2-1 Functional object processing:
[0061] Functional objects can be functions at any level, including the lowest level function points.
[0062] The function object includes title, description, search path, parent node, level, whether it is a leaf node, and child node object array. The format of the function object is as follows:
[0063]
[0064]
[0065] Store the function object in the function library. For an example of the function library, see Table 4:
[0066] Table 4: Function library example
[0067]
[0068] Among them, the function point object is the lowest level function object in the function object and can only be a leaf node. The function point object is the smallest independent unit that can be used for functional similarity comparison. The difference between it and the function object is that it can only be a leaf node.
[0069] S2-2 Function Title Object Processing:
[0070] The function title is a condensed reflection of the function's business content, and is characterized by being concise, clear, hierarchical, and containing keywords, and usually contains business keywords.
[0071] In order to carry out comparative calculation of function titles, the function titles are divided into three parts: business subject, business action, and business limitation:
[0072] Among them, the business subject is the object or field of the function, and its content is usually a noun. The business subject it represents can be a physical object or a virtual object. Physical objects include employees, vehicles, and factories, and virtual objects include defects, indicators, and reports.
[0073] Business actions are specific operations or presentation methods performed on business topics. Their contents are usually verbs or nouns, such as management, registration, approval, review, examination, verification, viewing, statistics, analysis, and judgment.
[0074] Business limitation is a description of the supplement, enhancement or limitation of the function, that is, it further clarifies the specific aspects or conditions that the function focuses on, and plays a role of supplementation, enhancement, and limitation. Its content can be adjectives or nouns, such as main grid, power distribution, 110kV, railway, trunk, rural area, industrialization, new energy, Industry 4.0, high risk, automation, etc.
[0075] The feature header object format is as follows:
[0076]
[0077] Since there may be multiple ways of semantic understanding and multiple business entities to be identified, all identification methods are applied in this case.
[0078] When comparing, the compared function title (current function title) and the compared function title (the compared item in the library) are grouped together, and the final function title similarity takes the highest value of the calculation results of the group of function titles. Taking the title "Vehicle Mileage Overview" as an example, in the subsequent title calculation, the two business entities "Mileage" and "Vehicle" can be associated at the same time for comparison calculation.
[0079] Method 1: "Driving mileage" is the business subject, "Overview" is the business action, and "Vehicle" is the business limitation.
[0080] Method 2: "Vehicle" is the business subject, "Overview" is the business action, and "Driving Mileage" is the business limitation.
[0081] S2-3 Functional Description Object Processing:
[0082] The function description object is a detailed description of the function title, a specific explanation and expansion of the function title, and describes more detailed business content, expected results, implementation methods, interaction methods, etc. The function description is usually a paragraph of text with rich content, complex meaning, and strong professionalism. The comparison calculation is carried out using a semantic similarity algorithm that has been trained and tuned with project management and related business field corpus.
[0083] The format of the function description object is as follows:
[0084]
[0085] Step S3: Functional similarity calculation:
[0086] In this step, similarity calculation is carried out in a bottom-up manner, and finally the similarity values of all functional objects are obtained.
[0087] The overall logic of function similarity calculation in this embodiment is as follows:
[0088] 1) Functional similarity calculation method
[0089] Displayed in a tree-like structure, a primary function can be split into multiple secondary functions, a secondary function can be split into multiple tertiary functions, and so on recursively until it is no longer possible to split, at which point all leaf nodes, that is, those without subordinate nodes, are function points.
[0090] The calculation of function similarity is carried out in a bottom-up manner. The core is to calculate the similarity of each function point object. Based on the similarity results of the function points, the upper-level function similarity is constructed step by step, and finally the similarity value of all function objects is obtained. The similarity of the function point is calculated based on the results of the function title similarity and the function description similarity. The function title similarity is based on the calculation results of the business subject similarity, business action similarity, and business limitation similarity. The calculation calculates the similarity between the current function title object and other function title objects in the function library and takes the highest value as the function title similarity result; the function description similarity result is calculated using the semantic similarity algorithm, using one or more of the function description object, function point object, and function object as input.
[0091] The calculation sequence of functional similarity is shown in the attached Figure 2 ~Attached Figure 6 shown.
[0092] 2) Functional similarity calculation process
[0093] The functional similarity calculation flow chart is as follows Figure 7 shown.
[0094] The calculation order of function similarity is to first calculate the function point title similarity, function point description similarity, and function point similarity in sequence, and then calculate the function similarity of the previous level of the function point in the order of function title similarity, function description similarity, and function similarity, and finally obtain the function similarity by proceeding step by step.
[0095] 3) Functional similarity algorithm scoring rules
[0096] The evaluation principle of functional similarity is a scoring principle, that is, the functional similarity is measured in percentage in the calculation, 0 means there is no correlation at all, and 100 means that the two are completely consistent.
[0097] For specific scoring rules, please refer to Table 5:
[0098] Table 5: Scoring rules table
[0099]
[0100]
[0101] The following is an explanation of the calculation principle of functional similarity:
[0102] 1) Burden reduction principle: the comparison range should be as small as possible
[0103] Since the underlying calculation is based on the semantic similarity algorithm, when computing power is limited, the list of functions involved in the functional similarity calculation and comparison should be as small as possible.
[0104] 2) Short-circuit principle: compare the results in advance
[0105] When the function title similarity is calculated too low, i.e., less than 70 points, the algorithm triggers a short-circuit effect and no longer calculates the function description similarity. The function point score will directly adopt the title similarity calculation result.
[0106] When the function title similarity is high enough, that is, ≥95 points, the algorithm triggers a short-circuit effect and no longer calculates the function description similarity. The function point score will directly adopt the title similarity calculation result.
[0107] 3) Optimum principle: matching algorithm with scenario
[0108] The similarity of functional titles requires high precision and high efficiency, so an algorithm that enhances the ability to replace synonyms and antonyms should be used.
[0109] Functional description similarity requires semantic understanding rather than literal text matching, so a semantic similarity algorithm that enhances the ability to understand the meaning of the entire text is used. The corresponding algorithm needs to be continuously tuned using the corresponding corpus.
[0110] The following is an explanation of the functional similarity calculation formula:
[0111] The key terms and English abbreviations involved in this embodiment are shown in Table 6:
[0112] Table 6: Comparison table of key terms and English abbreviations
[0113]
[0114] 1) Single function similarity calculation
[0115] Single function similarity is a similarity calculation based on a single function. The "function" can be a "function" of any level, including first-level functions, second-level functions, etc., as well as the lowest-level function points.
[0116] Assume there is a function F, whose function title similarity is FTS(F) and function point description similarity is FPCS(F). Set the weight w of factor FTS(F) FTS is 0.7, the weight of the factor FPCS(F) is W FPCS is 0.3, W FTS +W FPCS =1.
[0117] The calculation method of single function similarity (SFS) is shown in Formula 1:
[0118] SFS(F)=w FTS FTS(F)+w FPCS FPCS(F) Formula 1
[0119] Assume that the first-level function F1 consists of several second-level functions F 11 , F 12 , ..., F 1n The SFS of the first-level function F1 is expressed as SFS(F1). The title similarity of F1 itself is FTS(F1), and its weight w FTS is 0.7.
[0120] The weight W of the mean similarity of the subordinate functions of F1 FPCS is 0.3, W FTS +W FPCS =1.
[0121]
[0122] In Formula 1 or Formula 2, SFS represents the similarity of a single function, FTS represents the similarity of function title, and W FTS represents the weight of FTS; FPCS represents the similarity of function point description, W FPCS represents the weight of FPCS; Formula 1 is for the functional similarity calculation of function F, and Formula 2 is for the similarity calculation of several secondary functions F 11 , F 12 , ..., F 1n The first-level function F1 is composed of.
[0123] 2) Function title similarity calculation
[0124] According to formula 3 and formula 4:
[0125] ETS=w BEQ ·BEQ+w BA ·BA Formula 3
[0126]
[0127] Among them, BE is the business subject similarity, BA is the business action similarity, and BQ is the business qualification similarity. Set the weight W of factor BE BE is 0.7, and the weight of factor BQ is W BQ is 0.3, W BE +W BQ =1. Factor BEQ W BEQ The weight is 0.7, and the weight of factor BA is W BA is 0.3, W BEQ +W BA =1.
[0128] First, since BE and BQ are highly correlated, the comprehensive subject similarity BEQ is calculated. Then, the functional title similarity FTS is calculated by BEQ and BA.
[0129] Example
[0130] The present invention is further described with a specific embodiment:
[0131] A certain enterprise consists of a parent company M and several subsidiaries. It is now conducting a functional duplication check on the newly built project of subsidiary A, "Company A - Line Loss Monitoring and Assisted Analysis and Decision-making Tool" (referred to as Project 1). The main comparison object is the historical project of subsidiary B, "Company B - Distribution Network Line Loss Data Governance and Analysis Tool" (referred to as Project 2).
[0132] Step S1: Identification of digital project functions:
[0133] The functional similarity calculation is performed on the above two digital projects to check whether there is any duplication of functions between the two projects. The functional identification content of the digital projects in this embodiment is shown in Table 7.
[0134] Table 7: Functional identification content of digital project
[0135]
[0136] See Table 8 for a list of functions for Project 1:
[0137] Table 8: Function list of project 1
[0138]
[0139]
[0140] See Table 9 for the function list of Project 2:
[0141] Table 9: Function list of project 2
[0142]
[0143] Step S2: Function objectification processing:
[0144] S2-1 Functional object processing:
[0145] Identify the function lists of Project 1 and Project 2 and convert them into function objects. The results are shown in Table 10.
[0146] Table 10: Functional object processing results
[0147]
[0148]
[0149] S2-2 Function Title Object Processing:
[0150] Identify the function titles of each level of functions of Project 1 and Project 2, and convert them into function title objects. The results are shown in Table 11.
[0151] Table 11: Function title object processing results
[0152]
[0153]
[0154] S2-3 Functional Description Object Processing:
[0155] Identify the functional descriptions of each level of functions of Project 1 and Project 2 and convert them into functional description objects. The results are shown in Table 12.
[0156] Table 12: Functional description object processing results
[0157]
[0158]
[0159] Step S3: Functional similarity calculation:
[0160] Use the function objects, function title objects, and function description objects of Project 1 and Project 2 as inputs, and perform the function title similarity, function description similarity, and function similarity calculations in order. Here we select the process of the "Line loss anomaly analysis by substation area" function in Project 1 and the "Line loss anomaly analysis by substation area" function in Project 2.
[0161] S3-1 Function title similarity calculation
[0162] The business subject similarity, business action similarity, and business limitation similarity of the function title "Daily / Monthly Line Loss Monitoring by Component" and the function titles of Project 2 are calculated respectively, and the function title similarity FTS calculation is performed according to Formula 3 and Formula 4.
[0163] The "Line loss anomaly analysis by substation area" function in Project 1 has three title object recognition methods, and the "Line loss anomaly analysis by substation area" function in Project 2 has two title object recognition methods. Therefore, the final title similarity calculation of the two needs to be calculated 3×2=6 times in total.
[0164] According to the rules, the highest similarity value of a group of compared function titles is taken as the final score. Therefore, the similarity score of the function title "Line loss abnormality analysis in each substation area" in project 1 and the function title "Line loss abnormality analysis in substation area" in project 2 is 1, indicating that they are completely consistent.
[0165] The calculation results of function title similarity are shown in Table 13.
[0166] Table 13: Function title similarity calculation results
[0167]
[0168]
[0169] S3-2 Single function similarity calculation
[0170] Since the function-title similarity calculation result is 1, which is greater than 0.95 and satisfies the short-circuit condition, it is directly judged that the two functions are completely identical. The function similarity score is equal to the title similarity, SFS=FTS=1.
[0171] If the FTS does not trigger a short-circuit condition, it should be calculated according to Equation 1 and the results are shown in Table 14.
[0172] Table 14: Functional similarity calculation results
[0173]
[0174]
[0175] According to Table 5, the functions compared in this embodiment are very similar and are suspected to be the same function.
Claims
1. A general method for comparing functional similarity of digital projects, characterized in that: The comparison method is completed by step S1 of identifying the function of the digital item, step S2 of processing the function object, and step S3 of calculating the function similarity: Step S1 is to identify and divide the project into functions, construct a function list and describe the function list in a hierarchical manner, with the lowest level of hierarchical description being function points; The processing process in step S2 is to convert the function list into function objects, function title objects and function description objects with unique codes respectively, including three aspects: S2-1 function object processing, S2-2 function title object processing and S2-3 function description object processing; the function object includes title, description, search path, parent node, level, whether it is a leaf node, and child node object array, and then the function object is stored in the function library; the function title object is to split the function title into business subject, business action and business limitation; the function description object is a detailed function description of the function title; the function point object is the lowest level function object in the function object and can only be used as a leaf node. The function title of the function point object is split to obtain the function point title object, and the function point title object is described in detail to obtain the function point description object; In step S3, similarity calculation is performed in a bottom-up manner to obtain function title similarity, function description similarity and function similarity respectively, and finally obtain similarity values of all function objects.
2. A general digital project functional similarity comparison method according to claim 1, characterized in that: The identification described in step S1 is to identify the project by project number or title, recursively search for the function point node and record the hierarchical path as the retrieval path of the function point; the function division described in step S1 is divided in the order of title, function classification, and function point, and the function point is described according to hierarchical requirements, hierarchical functions, function descriptions and status content.
3. A general digital project functional similarity comparison method according to claim 1, characterized in that: The business subject described in step S2 is the object or field of the function, the business action is the specific operation or display method performed on the business subject, and the business limitation is a description of the supplement, enhancement or limitation of the function.
4. A general method for comparing functional similarity of digital items according to claim 1, characterized in that: The bottom-up approach described in step S3 is to first calculate the similarity of each function point object, and then construct the upper-level function similarity step by step based on the similarity results of the function points. The similarity results of the function points are calculated based on the function title similarity and the function description similarity results. The similarity calculation carried out on a single function is the single function similarity.
5. A universal digital project functional similarity comparison method according to claim 4, characterized in that: The function title similarity is calculated based on the business subject similarity, business action similarity, and business limitation similarity. The similarity between the current function title object and other function title objects in the function library is calculated and the highest value is taken as the current function title similarity result; the function description similarity result is calculated using a semantic similarity algorithm, using one or more of the function description object, function point object, and function object as input.
6. A universal digital project functional similarity comparison method according to claim 4, characterized in that: The single function similarity is calculated according to Formula 1 and Formula 2: SFS(F)=w FTS FTS(F)+w FPCS FPCS(F) Official 1 In Formula 1 or Formula 2, SFS represents the similarity of a single function, FTS represents the similarity of function title, and W FTS represents the weight of FTS; FPCS represents the similarity of function point description, W FPCS represents the weight of FPCS; Formula 1 is for the functional similarity calculation of function F, and Formula 2 is for the similarity calculation of several secondary functions F 11 , F 12 , ..., F 1n The first-level function F1 is composed of; The calculation formulas for the function title similarity are Formula 3 and Formula 4: FTS = w BEQ ·BEQ+w BA ·BA Official 3 In Formula 3 or Formula 4, FTS represents the similarity of functional titles, BE represents the similarity of business entities, and W BE represents the weight of BE; BA represents the similarity of business actions, W BA represents the weight of BA; BQ represents the business-limited similarity, W BQ Indicates the weight of BQ; BEQ represents the comprehensive subject similarity, W BEQ Represents the weight of BEQ.
7. A universal digital project functional similarity comparison method according to claim 6, characterized in that: The W FTS +W FPCS =1,W BE +W BQ =1,W BEQ +W BA =1.
8. A universal digital project functional similarity comparison method according to claim 1, characterized in that: The calculation order of the function similarity calculation in step S3 is to first calculate the function point title similarity, function point description similarity, and function point similarity in sequence, and then calculate the function similarity of the previous level of the function point in the order of function title similarity, function description similarity, and function similarity, and finally obtain the function similarity step by step.
9. A universal digital project functional similarity comparison method according to claim 1, characterized in that: The evaluation principle of the results of function similarity, function title similarity and function description similarity is a scoring principle. Specifically, the similarity is measured in percentage. 0 means there is no correlation at all, and 100 means the two are completely consistent. The degree of similarity is defined for different score ranges.
10. A universal digital project functional similarity comparison method according to claim 1, characterized in that: The calculation principles of functional similarity are the burden reduction principle, the short-circuit principle and the optimal principle: The principle of reducing the burden is that when computing power is limited, the function list involved in the function similarity calculation and comparison should be as small as possible; The short-circuit principle is that when the function title similarity is too low, the algorithm triggers the short-circuit effect and no longer calculates the function description similarity. The function point score will directly adopt the title similarity calculation result; The optimal principle is that the function title similarity adopts an algorithm that enhances the ability to replace synonyms and antonyms, and the function description similarity adopts a semantic similarity algorithm that enhances the ability to understand the meaning of the entire paragraph, and is continuously optimized using the corresponding corpus.