CN116861924A - Project risk early warning method and system based on artificial intelligence - Google Patents
Project risk early warning method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN116861924A CN116861924A CN202310953637.1A CN202310953637A CN116861924A CN 116861924 A CN116861924 A CN 116861924A CN 202310953637 A CN202310953637 A CN 202310953637A CN 116861924 A CN116861924 A CN 116861924A
- Authority
- CN
- China
- Prior art keywords
- item
- semantic understanding
- semantic
- training
- feature vector
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 47
- 239000013598 vector Substances 0.000 claims description 149
- 238000012549 training Methods 0.000 claims description 65
- 238000004891 communication Methods 0.000 claims description 19
- 238000012098 association analyses Methods 0.000 claims description 16
- 238000011156 evaluation Methods 0.000 claims description 11
- 239000011159 matrix material Substances 0.000 claims 1
- 230000001902 propagating effect Effects 0.000 claims 1
- 238000004458 analytical method Methods 0.000 abstract description 16
- 238000005516 engineering process Methods 0.000 abstract description 8
- 238000012502 risk assessment Methods 0.000 abstract description 6
- 235000019580 granularity Nutrition 0.000 description 62
- 230000006870 function Effects 0.000 description 16
- 238000007726 management method Methods 0.000 description 11
- 230000011218 segmentation Effects 0.000 description 9
- 230000008569 process Effects 0.000 description 8
- 238000004140 cleaning Methods 0.000 description 5
- 238000013135 deep learning Methods 0.000 description 5
- 238000007781 pre-processing Methods 0.000 description 5
- 238000013528 artificial neural network Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 210000002569 neuron Anatomy 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 238000012935 Averaging Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000003058 natural language processing Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 238000007796 conventional method Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000013136 deep learning model Methods 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 230000004927 fusion Effects 0.000 description 2
- 239000000463 material Substances 0.000 description 2
- 238000013468 resource allocation Methods 0.000 description 2
- 238000013179 statistical model Methods 0.000 description 2
- 238000012384 transportation and delivery Methods 0.000 description 2
- 238000012800 visualization Methods 0.000 description 2
- 230000002411 adverse Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000002457 bidirectional effect Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000002372 labelling Methods 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 238000003908 quality control method Methods 0.000 description 1
- 230000000306 recurrent effect Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 230000009466 transformation Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/30—Semantic analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/3331—Query processing
- G06F16/334—Query execution
- G06F16/3344—Query execution using natural language analysis
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
- G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
- G06F16/33—Querying
- G06F16/338—Presentation of query results
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computational Linguistics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Databases & Information Systems (AREA)
- Audiology, Speech & Language Pathology (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application discloses an artificial intelligence-based project risk early warning method and system, which are used for carrying out semantic analysis on text data related to a project through an artificial intelligence-based semantic understanding technology when project risk is predicted, so that the condition and the background of the project are better understood, and the project risk is accurately assessed to carry out project early warning. By the method, subjective analysis of experts can be avoided, automatic project risk assessment and early warning are achieved, and accordingly project teams are helped to make corresponding decisions timely and accurately.
Description
Technical Field
The application relates to the field of risk early warning, in particular to an artificial intelligence-based project risk early warning method and system.
Background
The risk early warning means that potential risks are recognized, evaluated and monitored in the process of operating the project or the service, and corresponding measures are timely found and taken to avoid or reduce adverse effects of the risks on the project or the service. In project management, risk early warning is an important link that can help a project team identify and evaluate various risks that a project may face, including technical risks, market risks, financial risks, and the like. Through risk early warning, project team can in time take corresponding risk response strategy, reduces the risk of project failure.
However, conventional project risk early warning schemes generally rely on expert analysis of project data, which often relies on expertise and experience of the expert, and has problems of subjectivity and limitation, so that different experts may have different opinions and judgments, resulting in inconsistency of early warning results. Moreover, the traditional project risk early warning method often requires a great deal of manpower and time investment, so that early warning efficiency is low. Meanwhile, the conventional method is difficult to process a large amount of unstructured text data, and cannot fully utilize various information sources in the project. Some existing project risk early warning schemes predict risk by being based on statistical models or rules, which often only identify known risk types, and it is difficult to accurately early warn of emerging risks or complex risks.
Accordingly, an optimized project risk early warning scheme is desired.
Disclosure of Invention
The present application has been made to solve the above-mentioned technical problems. The embodiment of the application provides an artificial intelligence-based project risk early warning method and system, which are used for carrying out semantic analysis on text data related to a project through an artificial intelligence-based semantic understanding technology when carrying out project risk prediction so as to better understand the condition and background of the project, thereby more accurately evaluating the project risk for carrying out project early warning. By the method, subjective analysis of experts can be avoided, automatic project risk assessment and early warning are achieved, and accordingly project teams are helped to make corresponding decisions timely and accurately.
According to one aspect of the present application, there is provided an artificial intelligence based project risk early warning method, comprising:
acquiring text data related to an item to be evaluated, wherein the text data comprises an item target, a range, a progress, a cost, a quality, a resource, communication and risk;
carrying out semantic association analysis on text data related to the evaluated item to obtain item semantic features; and
and determining a risk level tag based on the item semantic understanding features.
According to another aspect of the present application, there is provided an artificial intelligence based item risk early warning system, comprising:
the data acquisition module is used for acquiring text data related to the evaluated item, wherein the text data comprises an item target, a range, a progress, a cost, a quality, a resource, communication and risk;
the semantic association analysis module is used for carrying out semantic association analysis on the text data related to the evaluated item so as to obtain item semantic features; and
and the risk level generation module is used for determining a risk level label based on the item semantic understanding characteristics.
Compared with the prior art, the project risk early warning method and system based on the artificial intelligence provided by the application have the advantages that when project risk is predicted, the text data related to the project is subjected to semantic analysis by the semantic understanding technology based on the artificial intelligence, so that the condition and the background of the project are better understood, and the project risk is accurately estimated to perform project early warning. By the method, subjective analysis of experts can be avoided, automatic project risk assessment and early warning are achieved, and accordingly project teams are helped to make corresponding decisions timely and accurately.
Drawings
The above and other objects, features and advantages of the present application will become more apparent by describing embodiments of the present application in more detail with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
FIG. 1 is a flow chart of an artificial intelligence based project risk early warning method according to an embodiment of the application;
FIG. 2 is a system architecture diagram of an artificial intelligence based project risk early warning method according to an embodiment of the present application;
FIG. 3 is a flow chart of a training phase of an artificial intelligence based project risk early warning method according to an embodiment of the application;
FIG. 4 is a flowchart of substep S2 of an artificial intelligence based project risk early warning method according to an embodiment of the present application;
FIG. 5 is a flowchart of substep S21 of an artificial intelligence based project risk early warning method according to an embodiment of the present application;
FIG. 6 is a flowchart of substep S22 of an artificial intelligence based project risk early warning method according to an embodiment of the present application;
FIG. 7 is a block diagram of an artificial intelligence based project risk early warning system in accordance with an embodiment of the present application.
Detailed Description
Hereinafter, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
As used in the specification and in the claims, the terms "a," "an," "the," and/or "the" are not specific to a singular, but may include a plurality, unless the context clearly dictates otherwise. In general, the terms "comprises" and "comprising" merely indicate that the steps and elements are explicitly identified, and they do not constitute an exclusive list, as other steps or elements may be included in a method or apparatus.
Although the present application makes various references to certain modules in a system according to embodiments of the present application, any number of different modules may be used and run on a user terminal and/or server. The modules are merely illustrative, and different aspects of the systems and methods may use different modules.
A flowchart is used in the present application to describe the operations performed by a system according to embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed in order precisely. Rather, the various steps may be processed in reverse order or simultaneously, as desired. Also, other operations may be added to or removed from these processes.
Hereinafter, exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some embodiments of the present application and not all embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Conventional project risk early warning schemes generally rely on expert analysis of project data, which often relies on expert expertise and experience, and has subjective and limiting problems, so that different experts may have different comments and judgments, resulting in inconsistency of early warning results. Moreover, the traditional project risk early warning method often requires a great deal of manpower and time investment, so that early warning efficiency is low. Meanwhile, the conventional method is difficult to process a large amount of unstructured text data, and cannot fully utilize various information sources in the project. Some existing project risk early warning schemes predict risk by being based on statistical models or rules, which often only identify known risk types, and it is difficult to accurately early warn of emerging risks or complex risks. Accordingly, an optimized project risk early warning scheme is desired.
In the technical scheme of the application, an artificial intelligence-based project risk early warning method is provided. FIG. 1 is a flow chart of an artificial intelligence based project risk early warning method according to an embodiment of the application; fig. 2 is a system architecture diagram of an artificial intelligence-based project risk early warning method according to an embodiment of the present application. As shown in fig. 1 and 2, the project risk early warning method based on artificial intelligence according to the embodiment of the application comprises the following steps: s1, acquiring text data related to an evaluated item, wherein the text data comprises an item target, a range, a progress, a cost, a quality, a resource, communication and risk; s2, carrying out semantic association analysis on text data related to the evaluated item to obtain item semantic features; and S3, determining a risk level label based on the item semantic understanding features.
Specifically, in step S1, text data related to the item under evaluation is obtained, wherein the text data includes item targets, ranges, progress, costs, quality, resources, communications, and risks. It will be appreciated that through analysis and evaluation of the above data, knowledge of the status, progress and risk of the project can be aided in making corresponding decisions and adjustments. In addition, the text data may also be used for project reporting, communication, and knowledge management, facilitating collaboration and learning by a project team.
Notably, the project goal is to explicitly define the final goal and the expected outcome of the project. It describes the specific achievements that the project is to achieve, as well as the key steps and milestones required to accomplish these achievements. Project objectives are generally consistent with organizational strategic objectives, with the aim of meeting the needs and desires of stakeholders. Project scope refers to the definition of the work content and deliverables contained by a project. It defines the boundaries of the project, specifying the tasks and goals that the project team needs to accomplish. Project progress refers to the progress of a project over time, i.e., the progress made by a team of projects in completing project work and delivering deliverables. Project costs refer to the resources and fees required in the execution of a project. It includes direct costs and indirect costs. Project quality refers to whether the outcome of the project delivery meets expected requirements and criteria. Project quality management is the process of ensuring that the outcome of a project delivery is of a desired quality level. Project resources refer to various resources required in the process of executing a project, including human resources, material resources, financial resources, equipment resources, and the like. Efficient management and rational configuration of project resources is one of the keys to project success. Project communication refers to the activity of information exchange and communication between members of a project team and between parties associated with the project during execution of the project. Good project communication can promote team cooperation, ensure understanding and consensus of project targets, solve problems in time and improve project execution efficiency. Item risk refers to an uncertain factor that may negatively impact the goal of an item. Item risk management aims to identify, evaluate and address item risks to minimize the impact of risks on items.
Accordingly, in one possible implementation, text data related to the item under evaluation may be obtained by, for example: collecting project plans and documents: collecting project documents such as project plans, project rules, demand documents, scope specifications, progress plans, cost estimates, quality plans, communication plans, risk management plans and the like; evaluating project objectives and scopes: carefully reading the project rules and the range specifications, knowing the targets and the ranges of the projects, and extracting relevant text data; analyzing project progress and cost: reviewing project schedule and cost estimates, identifying key milestones, work packages, resource allocations and cost budgets for the project, and extracting relevant text data; inspection item quality plan and quality report: checking a quality plan, a test plan and a quality report of the project, knowing a quality target and a quality control activity of the project, and extracting relevant text data; investigation project resource management: checking resource requirements and resource allocation conditions of the project, including human resources, material resources, financial resources and technical resources, and extracting relevant text data; analyzing the project communication record: the method comprises the steps of examining communication plans and communication records of projects, including meeting summary, communication mails and communication records of communication tools, knowing communication contents between project teams and between related parties, and extracting related text data; evaluating project risk management: the risk management program and risk registry of the project are reviewed, the identified risk, risk assessment results and risk countermeasures are known, and relevant text data is extracted.
Specifically, in step S2, semantic association analysis is performed on the text data related to the evaluated item to obtain an item semantic feature. In particular, in one specific example of the present application, as shown in fig. 4, the S2 includes: s21, carrying out semantic understanding based on character association on the text data related to the evaluated item to obtain a semantic understanding feature vector of the character granularity item; s22, carrying out semantic understanding based on word association on the text data related to the evaluated item to obtain a semantic understanding feature vector of the item with the word granularity; and S23, fusing the character granularity project semantic understanding feature vector and the word granularity project semantic understanding feature vector to obtain a multi-granularity project semantic understanding feature vector as the project semantic feature.
Correspondingly, the S21 carries out semantic understanding based on character association on the text data related to the evaluated item to obtain a character granularity item semantic understanding feature vector. In particular, in one specific example of the present application, as shown in fig. 5, the S21 includes: s211, dividing the text data related to the evaluated item by taking characters as units to obtain a sequence of item description characters; and S212, enabling the sequence of the project description characters to pass through a semantic encoder comprising a Word2Vec model to obtain the character granularity project semantic understanding feature vector.
And S211, dividing the text data related to the evaluated item by character unit to obtain a sequence of item description characters. Considering that the text data related to the evaluated item contains a large amount of semantic information and the text data related to the evaluated item is composed of individual characters, the text data related to the evaluated item is further divided in character units to obtain a sequence of item description characters. In this way, the text data can be decomposed into a sequence of single characters, so that the text data related to the evaluated item can be processed more finely, for example, the semantic analysis at the character level or the format requirement of processing specific characters can be processed, the semantic understanding of the text data sheet can be facilitated, and the risk evaluation and early warning can be facilitated for the engineering.
Accordingly, in one possible implementation, the text data related to the item under evaluation may be divided in units of characters to obtain a sequence of item description characters, for example, by: collecting project description text data; the item description text data is divided in units of characters. This may be accomplished using string manipulation functions or libraries in a programming language; creating an empty character sequence for storing the divided project description character sequence; traversing the item description text data, and performing the following operation on each text data: converting the text data into a character sequence; adding a character sequence to the item description character sequence; after the traversal is completed, the item description character sequence is the item description sequence divided by taking the characters as units.
And S212, passing the sequence of the project description characters through a semantic encoder comprising a Word2Vec model to obtain the character granularity project semantic understanding feature vector. Considering that the sequence of the item description characters in the text data related to the item to be evaluated is data information divided by characters, each character has a semantic association relation of context. Therefore, in order to capture the semantic information of the text data related to the evaluated item, in the technical scheme of the application, the sequence of the item description characters needs to pass through a semantic encoder containing a Word2Vec model to obtain the character granularity item semantic understanding feature vector.
Accordingly, in one possible implementation, the sequence of item description characters may be passed through a semantic encoder comprising a Word2Vec model to obtain the character granularity item semantic understanding feature vector, for example, by: a corpus required to train the Word2Vec model was prepared. The sequence of project description characters is input as a corpus. Ensuring that the corpus contains enough samples to train out meaningful word vectors; the Word2Vec model was trained using the training corpus. The existing Word2Vec library may be used or an own training algorithm may be implemented. The training process learns the semantic representation of the characters according to the character sequences in the corpus; loading a trained Word2Vec model; the following operations are performed for each item description character: inputting the characters into a Word2Vec model to obtain corresponding character vectors; adding the character vector into the semantic understanding feature vector of the item description; after the traversal is completed, the semantic understanding feature vector of the item description is the combination of all character vectors.
It should be noted that, in other specific examples of the present application, the text data related to the evaluated item may be further understood based on the semantic understanding of the semantic association of the character to obtain a semantic understanding feature vector of the character granularity item, for example: collecting project-related text data: collecting text data associated with the project, including documents, reports, mail, meeting notes, and the like; data cleaning and pretreatment: cleaning and preprocessing the collected text data, including removing special characters and punctuation marks, converting into lowercase letters and the like; segmentation of the character sequence: dividing the preprocessed text data into character sequences, each character serving as a unit; constructing a character-level corpus: constructing a character-level corpus of the segmented character sequence for subsequent semantic understanding; training a character-level semantic model: training the character-level corpus by using machine learning or deep learning technology to construct a character-level semantic model. Character-level based bag of words models, such as FastText, or deep learning models, such as CharCNN, etc., may be used; character association semantic understanding: and carrying out character association semantic understanding on the item-related text data by using the trained character-level semantic model. The model can calculate the similarity between characters and find out the characters with similar semantic features; constructing a character granularity item semantic understanding feature vector: and constructing a character granularity item semantic understanding feature vector according to the result of the character association semantic understanding. Each character may be represented as a vector, the dimensions of which may be the word vector representation of the character, the semantic features of the character, etc.; visualization and analysis: the constructed character granularity item semantic understanding feature vector is visualized and analyzed so as to better understand the character level semantic features and association of the item.
Correspondingly, the S22 carries out semantic understanding based on word association on the text data related to the evaluated item to obtain a semantic understanding feature vector of the item with word granularity. In particular, in one specific example of the present application, as shown in fig. 6, the S22 includes: s221, dividing the text data related to the evaluated item by word unit to obtain a sequence of item description words; and S222, respectively passing the sequence of the item description words through the semantic encoders containing the Word2Vec model to obtain the Word granularity item semantic understanding feature vector.
And S221, dividing the text data related to the evaluated item by word unit to obtain a sequence of item description words. Considering that sequences divided in units of characters alone often fail to capture word-level semantic information, semantic understanding of the text data related to the item under evaluation may be inaccurate or lack context consistency. Therefore, in the technical scheme of the application, the text data related to the evaluated item is further divided by word units to obtain the sequence of the item description words, so that the semantic information and the context relation of the words in the text data related to the evaluated item can be better captured, the accuracy and the consistency of the subsequent understanding of the engineering situation and the background can be improved, and the risk evaluation and the early warning of the engineering can be facilitated.
Accordingly, in one possible implementation, the text data related to the item under evaluation may be divided in word units to obtain a sequence of item description words, for example, by: collecting project description text data; and carrying out word segmentation operation on the item description text data, and dividing the text data into word sequences. The word segmentation may be performed using natural language processing tools or libraries, such as NLTK, spaCy, etc.; creating an empty word sequence for storing the divided project description word sequence; traversing the item description text data, and performing the following operation on each text data: performing word segmentation operation, dividing text data into word sequences, and adding the word sequences into the item description word sequences; after the traversal is completed, the item description word sequence is the item description sequence divided by word units.
And S222, respectively passing the sequence of the item description words through the semantic encoder comprising the Word2Vec model to obtain the Word granularity item semantic understanding feature vector. Considering that the sequence of the item descriptor in the text data related to the item to be evaluated is data information divided by words, the words have a semantic association relation of context. Therefore, in order to capture the semantic information of the text data related to the evaluated item, in the technical scheme of the application, the sequence of the item description words needs to pass through a semantic encoder containing a Word2Vec model to obtain the Word granularity item semantic understanding feature vector.
It should be appreciated that the Word2Vec model is a technique for mapping words or characters to vector representations that can map similar words or characters into similar vector spaces to facilitate semantic understanding. After the sequence of the project description characters and the sequence of the project description words respectively pass through a Word2Vec model to map similar characters and words into a similar vector space respectively to obtain a sequence of input character vectors and a sequence of Word vectors, respectively performing context semantic association coding by using a context semantic understanding module of a semantic encoder, such as a bidirectional long-short term memory network model, so as to extract context semantic association characteristic information between character granularity and context semantic association characteristic information between Word granularity in the text data related to the evaluated project, namely semantic understanding characteristic information between character granularity and Word granularity-based associated semantic understanding characteristic information in the text data related to the evaluated project.
Accordingly, in one possible implementation manner, the sequence of the item description words may be respectively passed through the semantic encoder including the Word2Vec model to obtain the Word granularity item semantic understanding feature vector, for example: preparing a Word2Vec model: first, a semantic encoder containing a Word2Vec model needs to be prepared. Word2Vec is a technique for converting words into vector representations that can capture semantic relationships between words; loading a Word2Vec model: and loading the pre-trained Word2Vec model into a memory for subsequent use. The existing Word2Vec model can be used, and the training can be performed by using own data; word segmentation: the project description text is segmented and split into individual words. This may be done using word segmentation tools or natural language processing libraries; acquiring a word vector: and for each term, acquiring a corresponding term vector by querying a Word2Vec model. The word vector is a real number vector and represents semantic information of the word; constructing a feature vector: word vectors of all words are combined together according to a certain sequence to form a feature vector. Word vectors may be combined using simple weighted averaging or stitching, etc.; obtaining item semantic understanding feature vectors: through the steps, each item description text can obtain a corresponding item semantic understanding feature vector. This feature vector may be used for subsequent project management tasks such as similarity calculation, classification, etc.
It should be noted that, in other specific examples of the present application, the text data related to the evaluated item may be further understood based on word association semantics to obtain a word-granularity item semantic understanding feature vector in other manners, for example: collecting text data related to the item being evaluated: collecting text data related to the evaluated project, including project plans, demand documents, meeting records and the like; data preprocessing: preprocessing the collected text data, including punctuation removal, conversion to lower case letters and the like, for subsequent processing; word segmentation: dividing the preprocessed text data into word sequences using an appropriate word segmentation tool; constructing a word association semantic understanding model: based on the existing text data, a word association semantic understanding model is constructed. Word2Vec, gloVe and other pre-trained Word vector models can be used, and a Word vector model can be trained by self; word vector encoding: and converting each word into a corresponding word vector representation by using the constructed word association semantic understanding model. These word vectors may capture word-to-word semantic associations; constructing a semantic understanding feature vector of a word granularity item: for each word in the text data of the evaluated item, the corresponding word vectors are combined to form the item semantic understanding feature vector with the word granularity. Simple averaging or weighted averaging methods may be used; application item semantic understanding feature vector: after obtaining the term-granularity project semantic understanding feature vector, the term-granularity project semantic understanding feature vector can be used for various project management tasks, such as project classification, similarity calculation, information retrieval and the like.
Correspondingly, the S23 fuses the character granularity item semantic understanding feature vector and the word granularity item semantic understanding feature vector to obtain a multi-granularity item semantic understanding feature vector as the item semantic feature. That is, in the technical scheme of the application, the character granularity item semantic understanding feature vector and the word granularity item semantic understanding feature vector are fused, so that the context semantic association feature information based on the character granularity and the context semantic association feature information based on the word granularity of the text data related to the evaluated item are fused, and the multi-granularity item semantic understanding feature vector is obtained. Therefore, the text data related to the evaluated item can be fully understood semantically, so that more characteristic information about the condition, the background and the like of the item is captured, and the risk level of the item is evaluated.
Accordingly, in one possible implementation, the character granularity item semantic understanding feature vector and the word granularity item semantic understanding feature vector may be fused to obtain a multi-granularity item semantic understanding feature vector as the item semantic feature, for example: acquiring semantic understanding feature vectors of character granularity items: and for each item description text, obtaining a corresponding character granularity item semantic understanding feature vector through a semantic encoder at a character level. This feature vector captures semantic information for each character; acquiring semantic understanding feature vectors of word granularity items: according to the steps, the semantic understanding feature vector of the corresponding word granularity item is obtained through a semantic encoder of the word granularity. This feature vector captures semantic information for each word; fusion of feature vectors: and fusing the semantic understanding feature vector of the character granularity item and the semantic understanding feature vector of the word granularity item. Two feature vectors can be connected together by using a simple splicing mode to form a multi-granularity project semantic understanding feature vector; obtaining item semantic features: through the steps, each item description text can obtain a multi-granularity item semantic feature vector. This feature vector integrates semantic information of character granularity and word granularity, describing the semantics of the item more fully.
It should be noted that, in other specific examples of the present application, the text data related to the evaluated item may also be subjected to semantic association analysis in other manners to obtain semantic features of the item, for example: collecting project-related text data: collecting text data such as documents, reports, mails, meeting records and the like related to the project; data cleaning and pretreatment: cleaning and preprocessing the collected text data, including removing special characters and punctuation marks, converting into lowercase letters and the like; word segmentation and part-of-speech tagging: dividing the text data into words, dividing sentences into words, and labeling each word with part of speech for subsequent semantic analysis; constructing a corpus: constructing the text data after cleaning and preprocessing into a corpus for subsequent semantic association analysis; training a semantic model: the language library is trained by using machine learning or deep learning technology to construct a semantic model. A bag of words model based approach, such as TF-IDF, or a deep learning model, such as Word2Vec, BERT, etc., may be used; semantic association analysis: and carrying out semantic association analysis on the item-related text data by using the trained semantic model. The similarity between the texts can be calculated by using the model, and texts with similar semantic features can be found out; extracting item semantic features: and extracting semantic features of the item according to the result of the semantic association analysis. The texts with similar semantic features can be classified into one type through a clustering algorithm, or key features of the project can be extracted through methods such as key word extraction; visualization and analysis: the extracted semantic features of the item are visualized and analyzed to better understand the key features and semantic associations of the item.
Specifically, in step S3, a risk level tag is determined based on the item semantic understanding feature. In particular, in one specific example of the present application, the multi-granularity item semantic understanding feature vector is passed through a classifier to obtain a classification result, where the classification result is used to represent a risk level tag. According to the embodiment of the application, the multi-granularity item semantic understanding feature vector is passed through a classifier to obtain a classification result, wherein the classification result is used for representing a risk level label. In particular, the label of the classifier is a risk grade label of the project, and after the classification result is obtained, the project can be evaluated and early-warned based on the classification result so as to help a project team to accurately make corresponding decisions in time. Specifically, using a plurality of full-connection layers of the classifier to perform full-connection coding on the multi-granularity item semantic understanding feature vector so as to obtain a coding classification feature vector; and passing the coding classification feature vector through a Softmax classification function of the classifier to obtain the classification result.
A classifier is a deep neural network based classifier that can perform classification tasks by learning features and patterns of large amounts of data. Compared with the traditional machine learning classifier, the deep learning classifier has stronger expressive power and classification accuracy when processing large-scale and complex data sets.
Fully connected layers (Fully Connected Layer), also known as dense connected layers or fully connected layers, are a common layer type in deep neural networks. In the fully connected layer, each neuron has a connection to all neurons in the previous layer, each connection having a weight parameter. This means that each neuron in the fully connected layer receives the inputs of all neurons in the previous layer, and performs a weighted summation by the weight parameters, and then a nonlinear transformation by the activation function.
The Softmax classification function is a commonly used multi-class classification function that converts an input vector into an output vector that represents the probabilities of the respective classes. In deep learning, a Softmax function is typically used for the output of the last layer, converting the output of the neural network into a class probability distribution, and then making a classification decision based on the probability magnitude. Specifically, the output vector of the network may be input into a Softmax function, to obtain an output vector representing the probability of each category, and then the category with the highest probability is selected as the final classification result.
It is worth mentioning that in other specific examples of the application, the risk level tag may also be determined by other means based on the item semantic understanding feature, for example: item description text data is collected and converted into semantic understanding feature vectors. Text data may be encoded into semantic feature vectors using natural language processing techniques such as word embedding (word embedding) or pre-trained language models; a set of sample data of known risk levels is prepared, which samples should contain item descriptions and corresponding risk level tags. Marking can be performed by an expert according to project characteristics and historical experience; sample data of known risk levels are divided into training and test sets. Typically, most of the samples can be used for training and a small part for testing to evaluate the performance of the model; a deep learning classifier model is trained using the training set data. An appropriate model structure, such as a Convolutional Neural Network (CNN) or a Recurrent Neural Network (RNN), may be selected and a Softmax classification function used as the last layer activation function. Optimizing model parameters by using a cross entropy loss function in the training process; the performance of the classifier is evaluated using the test set data. Indexes such as accuracy, recall rate, F1 value and the like can be calculated to measure the classification performance of the model. If the model performance is not ideal, attempts may be made to adjust the model structure, super parameters, or use more training data to refine the model; when the model training is completed and passed through the assessment, the model can be used to predict the risk level of the new project. The semantic understanding features of the new item are input into the model, which will output a vector representing the probability of each risk level. The risk level with the highest probability may be selected as the final prediction result.
It should be appreciated that the semantic encoder containing the Word2Vec model and the classifier need to be trained prior to making inferences using the neural network model described above. That is, the project risk early warning method based on artificial intelligence of the application further comprises a training stage for training the semantic encoder including the Word2Vec model and the classifier.
Fig. 3 is a flowchart of a training phase of an artificial intelligence based project risk early warning method according to an embodiment of the present application. As shown in fig. 3, the project risk early warning method based on artificial intelligence according to the embodiment of the application includes: a training phase comprising: s110, training data are acquired, wherein the training data comprise training text data related to an evaluated item, and the true value of the risk level label; s120, dividing the training text data related to the evaluated item by taking characters as units to obtain a training item description character sequence; s130, dividing the training text data related to the evaluated item by word units to obtain a training item description word sequence; s140, respectively passing the sequence of the training item description character and the sequence of the training item description Word through the semantic encoder comprising the Word2Vec model to obtain a training character granularity item semantic understanding feature vector and a training Word granularity item semantic understanding feature vector; s150, fusing the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector to obtain a training multi-granularity item semantic understanding feature vector; s160, enabling the training multi-granularity project semantic understanding feature vector to pass through the classifier to obtain a classification loss function value; s170, calculating a common manifold implicit similarity factor of the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector to obtain a common manifold implicit similarity loss function value; s180, training the semantic encoder comprising the Word2Vec model and the classifier by taking the weighted sum of the classification loss function value and the common manifold implicit similarity loss function value as the loss function value and transmitting the gradient descent direction.
In particular, in the technical scheme of the application, the character granularity item semantic understanding feature vector and the word granularity item semantic understanding feature vector respectively express text semantic features of different granularities of text data related to an evaluated item, and the consideration that homologous data is in a base is taken into considerationWhen the semantic difference of each unit text in different granularity division and the associated feature difference in the text context associated direction caused by the semantic difference are fused, the multi-granularity item semantic understanding feature vector still expects to promote the geometric monotonicity of the high-dimensional feature manifold of the fused high-dimensional feature distribution when the feature expression of each of the semantic understanding feature vector of the character granularity item and the semantic understanding feature vector of the word granularity item is in the feature manifold difference in the high-dimensional feature space, so that the convergence difficulty of the multi-granularity item semantic understanding feature vector in classification regression through a classifier is avoided. Based on this, the applicant of the present application considers that the semantic understanding feature vector of the item of the character granularity and the semantic understanding feature vector of the item of the word granularity have a common manifold of feature manifold of the homologous text semantic at the time of feature extraction though semantic feature encoding is performed based on different granularities on the basis of homologous data, so manifold geometric constraint of the overall feature distribution of the semantic understanding feature vector of the item of the multi-granularity, that is, the semantic understanding feature vector V for the item of the character granularity, can be performed based on the common manifold representation 1 And the word granularity item semantic understanding feature vector V 2 The implicit similarity factor of the common manifold of feature vectors is introduced as a loss function, specifically expressed as:
‖·‖ 2 representing the two norms of the vector, anFrobe representing a matrixSquare root of nius norm, eigenvector V 1 And V 2 All in the form of column vectors, w 1 、w 2 、w 3 And alpha is a weight super parameter. Here, the common manifold implicit similarity factor can semantically understand the feature vector V with the character granularity item 1 And the word granularity item semantic understanding feature vector V 2 The structural association between the feature vectors represents the common manifold of the respective feature manifolds under the cross dimension, and the common constraint of manifold structural factors such as the difference, the correspondence, the relativity and the like of the feature vectors is shared by the same factorization weight, so that the distribution similarity of geometric derivative structural representations depending on the common manifold is measured to realize the semantic understanding of the feature vector V of the character granularity project 1 And the word granularity item semantic understanding feature vector V 2 The nonlinear geometric monotonicity of the fusion features of the multi-granularity item semantic understanding feature vector is improved, and the convergence effect is improved when the multi-granularity item semantic understanding feature vector is subjected to classification regression through a classifier. Therefore, the condition and the background of the project are more fully understood by the semantic understanding technology of artificial intelligence, subjective analysis of experts can be avoided, automatic project risk assessment and early warning are realized, corresponding decisions are timely and accurately made by project teams, and project management efficiency and accuracy are improved.
In summary, the method for early warning the risk of the project based on the artificial intelligence is explained, and when the project risk is predicted, the condition and the background of the project are better understood by carrying out semantic analysis on text data related to the project through semantic understanding technology based on the artificial intelligence, so that the project risk is accurately estimated for early warning the project. By the method, subjective analysis of experts can be avoided, automatic project risk assessment and early warning are achieved, and accordingly project teams are helped to make corresponding decisions timely and accurately.
Further, an artificial intelligence-based project risk early warning system is provided.
FIG. 7 is a block diagram of an artificial intelligence based project risk early warning system in accordance with an embodiment of the present application. As shown in fig. 7, an artificial intelligence based project risk early warning system 300 according to an embodiment of the present application includes: a data collection module 310, configured to obtain text data related to an item under evaluation, where the text data includes an item goal, a scope, a progress, a cost, a quality, a resource, a communication, and a risk; the semantic association analysis module 320 is configured to perform semantic association analysis on text data related to the evaluated item to obtain item semantic features; and a risk level generation module 330 for determining a risk level tag based on the item semantic understanding feature.
As described above, the artificial intelligence based item risk early warning system 300 according to the embodiment of the present application may be implemented in various wireless terminals, for example, a server or the like having an artificial intelligence based item risk early warning algorithm. In one possible implementation, the artificial intelligence based project risk early warning system 300 according to embodiments of the present application may be integrated into a wireless terminal as a software module and/or hardware module. For example, the artificial intelligence based project risk early warning system 300 may be a software module in the operating system of the wireless terminal or may be an application developed for the wireless terminal; of course, the artificial intelligence based item risk early warning system 300 could equally be one of many hardware modules of the wireless terminal.
Alternatively, in another example, the artificial intelligence based item risk early warning system 300 and the wireless terminal may be separate devices, and the artificial intelligence based item risk early warning system 300 may be connected to the wireless terminal through a wired and/or wireless network and transmit interactive information in a agreed data format.
The foregoing description of the embodiments of the present disclosure has been presented for purposes of illustration and description, and is not intended to be exhaustive or limited to the embodiments disclosed. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the various embodiments described. The terminology used herein was chosen in order to best explain the principles of the embodiments, the practical application, or the improvement of technology in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.
Claims (9)
1. An artificial intelligence-based project risk early warning method is characterized by comprising the following steps:
acquiring text data related to an item to be evaluated, wherein the text data comprises an item target, a range, a progress, a cost, a quality, a resource, communication and risk;
carrying out semantic association analysis on the text data related to the evaluated item to obtain item semantic features; and
and determining a risk level tag based on the item semantic understanding features.
2. The artificial intelligence based item risk early warning method according to claim 1, wherein performing semantic association analysis on the text data related to the evaluated item to obtain item semantic understanding features, comprises:
carrying out semantic understanding based on character association on the text data related to the evaluated item to obtain a semantic understanding feature vector of the character granularity item;
carrying out semantic understanding based on word association on the text data related to the evaluated item to obtain a semantic understanding feature vector of the item with word granularity; and
and merging the character granularity item semantic understanding feature vector and the word granularity item semantic understanding feature vector to obtain a multi-granularity item semantic understanding feature vector as the item semantic feature.
3. The artificial intelligence based item risk early warning method of claim 2, wherein performing character-based semantic understanding on the text data related to the item under evaluation to obtain a character-granularity item semantic understanding feature vector, comprises:
dividing the text data related to the evaluated item by taking characters as units to obtain a sequence of item description characters; and
and passing the sequence of the item description characters through a semantic encoder comprising a Word2Vec model to obtain the character granularity item semantic understanding feature vector.
4. The artificial intelligence based item risk early warning method of claim 3, wherein performing word-association based semantic understanding on the text data related to the item under evaluation to obtain a word-granularity item semantic understanding feature vector, comprising:
dividing the text data related to the evaluated item by word unit to obtain a sequence of item description words; and
and respectively passing the sequence of the item description words through the semantic encoder containing the Word2Vec model to obtain the Word granularity item semantic understanding feature vector.
5. The artificial intelligence based item risk early warning method of claim 4, wherein determining a risk level tag based on the item semantic understanding feature comprises: and passing the multi-granularity item semantic understanding feature vector through a classifier to obtain a classification result, wherein the classification result is used for representing a risk level label.
6. The artificial intelligence based project risk early warning method according to claim 5, further comprising a training step of: training the semantic encoder comprising Word2Vec model and the classifier.
7. The artificial intelligence based project risk early warning method according to claim 6, wherein the training step comprises:
acquiring training data, wherein the training data comprises training text data related to an evaluated item and a true value of the risk level label;
dividing the training text data related to the evaluated item by taking characters as units to obtain a sequence of training item description characters;
dividing the training text data related to the evaluated item by word units to obtain a training item description word sequence;
respectively passing the sequence of the training item description characters and the sequence of the training item description words through the semantic encoder comprising the Word2Vec model to obtain training character granularity item semantic understanding feature vectors and training Word granularity item semantic understanding feature vectors;
fusing the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector to obtain a training multi-granularity item semantic understanding feature vector;
Passing the training multi-granularity item semantic understanding feature vector through the classifier to obtain a classification loss function value; and
calculating a common manifold implicit similarity factor of the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector to obtain a common manifold implicit similarity loss function value;
training the semantic encoder and the classifier comprising the Word2Vec model with a weighted sum of the classification loss function value and the common manifold implicit similarity loss function value as a loss function value and propagating in the direction of gradient descent.
8. The artificial intelligence based project risk early warning method of claim 7, wherein calculating a common manifold implicit similarity factor for the training character granularity project semantic understanding feature vector and the training word granularity project semantic understanding feature vector to obtain a common manifold implicit similarity loss function value comprises:
calculating a common manifold implicit similarity factor of the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector according to the following loss formula to obtain a common manifold implicit similarity loss function value;
Wherein, the loss formula is:
wherein V is 1 And V 2 Respectively the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector, and II 2 Representing the two norms of the vector, anRepresenting square root of Frobenius norm of matrix, wherein the training character granularity item semantic understanding feature vector and the training word granularity item semantic understanding feature vector are in column vector form, and w 1 、w 2 、w 3 And alpha is a weight superparameter, +.>Indicates vector multiplication, ++indicates multiplication by position point, ++>Representing difference by position +.>Representing the common manifold implicit similarity loss function value.
9. An artificial intelligence based project risk early warning system, comprising:
the data acquisition module is used for acquiring text data related to the evaluated item, wherein the text data comprises an item target, a range, a progress, a cost, a quality, a resource, communication and risk;
the semantic association analysis module is used for carrying out semantic association analysis on the text data related to the evaluated item so as to obtain item semantic features; and
and the risk level generation module is used for determining a risk level label based on the item semantic understanding characteristics.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310953637.1A CN116861924A (en) | 2023-07-31 | 2023-07-31 | Project risk early warning method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310953637.1A CN116861924A (en) | 2023-07-31 | 2023-07-31 | Project risk early warning method and system based on artificial intelligence |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116861924A true CN116861924A (en) | 2023-10-10 |
Family
ID=88235934
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310953637.1A Pending CN116861924A (en) | 2023-07-31 | 2023-07-31 | Project risk early warning method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116861924A (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117637153A (en) * | 2024-01-23 | 2024-03-01 | 吉林大学 | Informationized management system and method for patient safety nursing |
| CN117689278A (en) * | 2024-02-04 | 2024-03-12 | 新疆盛诚工程建设有限责任公司 | Construction quality intelligent management system and method |
| CN118136219A (en) * | 2024-05-06 | 2024-06-04 | 吉林大学 | Medical institution hospital feel management quality evaluation system and method |
-
2023
- 2023-07-31 CN CN202310953637.1A patent/CN116861924A/en active Pending
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117637153A (en) * | 2024-01-23 | 2024-03-01 | 吉林大学 | Informationized management system and method for patient safety nursing |
| CN117637153B (en) * | 2024-01-23 | 2024-03-29 | 吉林大学 | Informationized management system and method for patient safety nursing |
| CN117689278A (en) * | 2024-02-04 | 2024-03-12 | 新疆盛诚工程建设有限责任公司 | Construction quality intelligent management system and method |
| CN118136219A (en) * | 2024-05-06 | 2024-06-04 | 吉林大学 | Medical institution hospital feel management quality evaluation system and method |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109657947B (en) | Enterprise industry classification-oriented anomaly detection method | |
| CN116861924A (en) | Project risk early warning method and system based on artificial intelligence | |
| CN108459955B (en) | Software defect prediction method based on deep self-coding network | |
| CN112836509A (en) | Expert system knowledge base construction method and system | |
| CN115526236A (en) | Text network graph classification method based on multi-modal comparative learning | |
| Martínez-Plumed et al. | CASP-DM: context aware standard process for data mining | |
| CN117271767A (en) | Operation and maintenance knowledge base establishing method based on multiple intelligent agents | |
| CN117131449A (en) | Data management-oriented anomaly identification method and system with propagation learning capability | |
| CN117236665B (en) | Material production scheduling optimization method and system | |
| CN117422181B (en) | Fuzzy label-based method and system for early warning loss of issuing clients | |
| CN117333146A (en) | Manpower resource management system and method based on artificial intelligence | |
| Saetia et al. | Data-driven approach to equipment taxonomy classification | |
| CN117372144A (en) | Wind control strategy intelligent method and system applied to small sample scene | |
| CN115438190B (en) | Power distribution network fault auxiliary decision knowledge extraction method and system | |
| CN117237095A (en) | Small micro financial industry risk early warning and monitoring platform based on big data analysis | |
| CN112001484A (en) | Safety defect report prediction method based on multitask deep learning | |
| CN114757183B (en) | Cross-domain emotion classification method based on comparison alignment network | |
| Pattnaik et al. | A survey on machine learning techniques used for software quality prediction | |
| CN113468203B (en) | Financial user image drawing method based on recurrent neural network and attention mechanism | |
| Liu et al. | TD-GAT: Graph neural network for fault diagnosis knowledge graph | |
| CN113643141A (en) | Method, device and equipment for generating explanatory conclusion report and storage medium | |
| CN117852553B (en) | Language processing system for extracting component transaction scene information based on chat record | |
| CN117009883B (en) | Object classification model construction method, object classification method, device and equipment | |
| CN117455059A (en) | Industry trend evaluation system based on data acquisition | |
| CN117422311A (en) | Enterprise risk rating model construction method, system and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |