Enterprise digital evaluation method based on AI enhancement

Abstract

The invention provides an enterprise digital evaluation method based on AI enhancement. The method comprises the following steps: merging text data and structured data in a historical enterprise digital evaluation document into a feature matrix, carrying out model training by using the merged feature matrix and a digital evaluation result, and extracting feature importance of the historical enterprise digital evaluation document; new enterprise digital evaluation document data are input into the trained model, the model automatically distributes weights for new enterprise digital evaluation document data features according to feature importance learned from historical enterprise digital evaluation documents, and a digital evaluation result is obtained. According to the evaluation method, the efficiency and precision of digital evaluation are improved, detailed feature weight distribution is provided, an enterprise management layer can be helped to better understand key factors influencing the digital level, and the performance of the model can be further improved through continuous learning iteration.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to an enterprise digitalization evaluation method based on AI enhancement. Background Art

[0002] The rapid development of digital technology is changing the business environment and operation mode of enterprises. Digital transformation has become an important way for enterprises to achieve sustainable development. Digital transformation not only involves the application of technology and tools, but more importantly, it is about changing the way of thinking and organizational culture of enterprises to adapt to the rapidly changing market and consumer needs.

[0003] Digital transformation is an ongoing process, and enterprises need to regularly evaluate the performance of digital transformation in order to make timely adjustments and improvements. Evaluations are usually conducted through regular data analysis, surveys, and feedback mechanisms. Through these evaluations, enterprises can understand the actual effects of digital transformation, identify problems and bottlenecks, and take timely measures to improve them. Existing digital evaluation methods include traditional survey methods, which usually rely on manual methods such as questionnaires and expert interviews. Although these methods have certain intuitiveness and flexibility, they have limitations such as low efficiency, high cost, and strong subjectivity. There are also evaluation methods using data analysis tools, which can process and analyze a large amount of enterprise data to obtain relatively objective evaluation results. However, these evaluation methods can often only evaluate specific areas and lack flexibility, comprehensiveness, and comprehensiveness. Therefore, there is an urgent need for a digital evaluation method that is efficient, flexible, comprehensive, and accurate. Summary of the invention

[0004] The present invention describes an enterprise digital assessment method based on AI enhancement, which can solve the above technical problems.

[0005] According to a first aspect, a method for enterprise digitalization assessment based on AI enhancement is provided, the method comprising: Collect digital evaluation documents of historical enterprises, where the digital evaluation documents of historical enterprises include text data, structured data and digital evaluation results; Combine text data and structured data from historical enterprise digital assessment documents into a feature matrix; Use the merged feature matrix and digital evaluation results to train the model and extract the feature importance of historical enterprise digital evaluation documents; The new enterprise digital assessment document data is input into the trained model. The model automatically assigns weights to the new enterprise digital assessment document data features according to the feature importance learned from the historical enterprise digital assessment documents to obtain the digital assessment results.

[0006] In some embodiments, before merging the text data and the structured data into a feature matrix, the text data is preprocessed as follows: Segment new text data into words or phrases; Remove meaningless words from text data; Convert the processed text data into a numeric vector.

[0007] In some embodiments, before merging the text data and the structured data into a feature matrix, the structured data is cleaned and standardized.

[0008] In some embodiments, after extracting the feature importance of the historical enterprise digital assessment document, the feature importance is sorted, and then the feature name and the corresponding feature importance are stored in a dictionary.

[0009] In some embodiments, the method of automatically assigning weights to new enterprise digital assessment document data features according to learned feature importance includes: Traverse the features of the new enterprise digital assessment document data, and find the corresponding feature importance from the weight dictionary according to the feature name; The found feature importance is assigned to the corresponding features of the new enterprise digital evaluation document data.

[0010] In some embodiments, a visual report is generated based on the feature data stored in the dictionary.

[0011] In some embodiments, for features that exist in the training data but are missing in the new enterprise digital assessment document data, these features are added to the new enterprise digital assessment document data and the values ​​are set to 0; for features that exist in the new enterprise digital assessment document data but are not present in the training data, these features are retained in the new enterprise digital assessment document data.

[0012] In some embodiments, for features that exist in the new enterprise digitalization evaluation document data but do not exist in the training data, after retaining these features in the new enterprise digitalization evaluation document data, the method further includes: By manually adjusting the importance of the retained features; Alternatively, the model can be incrementally trained using new enterprise digital assessment document data through incremental learning methods, the model parameters can be updated, and the importance of features can be automatically adjusted through the model.

[0013] In some embodiments, a digital report is generated based on the digital evaluation results.

[0014] The present invention provides an enterprise digital assessment method based on AI enhancement. By merging the text data and structured data in the historical enterprise digital assessment documents into a feature matrix, using the merged feature matrix and the digital assessment results to perform model training, extracting the feature importance of the historical enterprise digital assessment documents, and inputting the new enterprise digital assessment document data into the trained model, the model automatically assigns weights to the new enterprise digital assessment document data features according to the feature importance learned from the historical enterprise digital assessment documents, and obtains the digital assessment results. This assessment method improves the efficiency and accuracy of digital assessment, and providing detailed feature weight allocation can help enterprise management better understand the key factors affecting the level of digitalization, and the model can further improve its performance through continuous learning iterations. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0016] Figure 1 A flowchart of an enterprise digital assessment method based on AI enhancement provided in an embodiment of this specification is shown; Figure 2 A schematic diagram of the structure of an enterprise digital evaluation device based on AI enhancement provided in an embodiment of this specification is shown. DETAILED DESCRIPTION

[0017] The solution provided in this specification is described below in conjunction with the accompanying drawings.

[0018] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in conjunction with the accompanying drawings.

[0019] In the description of the embodiments of the present application, words such as "exemplary", "for example" or "for example" are used to indicate examples, illustrations or descriptions. Any embodiment or design described as "exemplary", "for example" or "for example" in the embodiments of the present application should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of words such as "exemplary", "for example" or "for example" is intended to present related concepts in a concrete way.

[0020] In the description of the embodiments of the present application, the term "and / or" is merely a description of the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B may represent: A exists alone, B exists alone, and A and B exist at the same time. In addition, unless otherwise specified, the term "plurality" means two or more.

[0021] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. The terms "include", "comprises", "has" and their variations all mean "including but not limited to", unless otherwise specifically emphasized.

[0022] Figure 1 A schematic diagram of a method for enterprise digitalization evaluation based on AI enhancement provided in an embodiment of this specification is shown as follows: Figure 1 As shown, the method comprises the following steps: Step S101, collecting historical enterprise digital evaluation documents. Specifically, the historical enterprise digital evaluation documents include text data, structured data and digital evaluation results.

[0023] The data in the enterprise digital assessment document generally includes relevant data of the enterprise, such as business process data, financial system data, IT infrastructure data, customer experience data, employee skills and training data, compliance and risk management data, supply chain and logistics data, etc. And indicators formulated according to the above data dimensions, such as business process data further includes: degree of automation, process efficiency, error rate, etc. The indicators are defined as: the proportion of process automation, the time required to complete the process, and the frequency of errors in the process. With the digital development of the enterprise, quantified indicator data will be collected regularly. Scoring rules are formulated for each indicator based on meeting minutes, project documents, internal reports, customer feedback, public reports, and suggestions from leadership or experts, where the scoring rules include the weights of each indicator, and then documents are formulated based on the quantified indicator data. Among them, meeting minutes, project documents, internal reports, customer feedback, public reports, suggestions from leadership or experts, etc. are text data in the enterprise digital assessment document, and documents formulated based on quantified indicator data are structured data. The digital assessment results can be in the form of two categories, such as "transformation successful" or "transformation unsuccessful", or multiple categories, such as the digital transformation stage: "primary", "intermediate", "advanced", or it can be the digital transformation score of the enterprise, or a percentage value of digital maturity.

[0024] Structured data has high accuracy and comparability, and can directly reflect the actual operating conditions of an enterprise. Text data provides rich background information and contextual information, which helps to understand the meaning behind structured data, and can provide explanations and supplements to structured data, helping the model to better understand and interpret data.

[0025] Step S102: merging text data and structured data in the historical enterprise digital evaluation document into a feature matrix.

[0026] First, you need to vectorize the text data and structured data, then concatenate the text feature vectors and structured feature vectors by column, and then create a DataFrame containing all the features. Specifically, DataFrame is an important data structure in the pandas library, which is used to process and analyze structured data. It is similar to a table or Excel table, where data is organized in rows and columns. Each column can contain different types of data (such as integers, floating-point numbers, strings, etc.), and each column has a label. DataFrame provides a wealth of functions to manipulate data, and is very suitable for tasks such as data cleaning, transformation, aggregation, and visualization.

[0027] In one embodiment, before merging text data and structured data into a feature matrix, the text data is preprocessed by segmenting new text data into words or phrases, deleting common meaningless words, and converting the processed text data into a numerical vector.

[0028] Use natural language processing tools (such as NLTK, SpaCy, etc.) to segment text into words or phrases. The purpose of word segmentation is to divide continuous text into independent units, which can be words, phrases, or other meaningful fragments. Then, common meaningless words are deleted to reduce noise and improve the efficiency and accuracy of subsequent processing. Furthermore, the word vectorization method is used to convert the processed text data into numerical vectors. Some common word vectorization methods can be used, such as TF-IDF, Word2Vec, GloVe, etc.

[0029] In one implementation, before merging the text data and the structured data into a feature matrix, the structured data is cleaned and standardized.

[0030] First, perform data cleaning on the structured data. Data cleaning includes processing missing values, processing outliers, and processing duplicate values. If a feature has a large number of missing values ​​and the feature has little effect on the model, you can choose to delete the feature or delete the rows with missing values. For important features, you can use the mean, median, mode, or interpolation method to fill in the missing values. Use statistical methods such as Z-score, IQR, etc. to detect outliers in the data. You can delete outliers or replace them with reasonable values, such as mean and median. You can view all duplicate rows by indexing the Boolean series with the data frame, delete all duplicate rows from the data frame, and only keep the first row. Finally, standardize the structured data. You can use the Min-Max Scaling method to calculate the minimum and maximum values ​​of each column and scale each column of data to between 0 and 1. You can also use the Z-score standardization method to calculate the mean and standard deviation of each column and convert each column of data to a standard normal distribution.

[0031] Step S103, using the merged feature matrix and the digital evaluation results to perform model training, and extracting the feature importance of the historical enterprise digital evaluation documents.

[0032] Optionally, the model can choose to use decision tree, random forest, XGBoost or LightGBM.

[0033] The merged feature matrix data and digital evaluation results are divided into a training set and a test set. The feature matrix data is used as the model input, and the digital evaluation results are used as the model output. The model is trained using the training set, and tested using the test set. During the testing process, the model parameters are adjusted to further evaluate the model performance until a model with high evaluation performance is obtained. Finally, the feature importance of historical enterprise digital evaluation documents is extracted from the trained model.

[0034] In one implementation, after extracting the feature importance of the historical enterprise digital assessment document, the feature importance is sorted, and then the feature name and the corresponding feature importance are stored in a dictionary.

[0035] You need to combine the feature names and feature importances into a list of tuples containing key-value pairs, and build a new dictionary with the first element (feature name) of each tuple as the key and the second element (feature importance) as the value. Then sort by feature importance, for example, you can use the sorted function to sort the list, and finally store the sorted features in the dictionary.

[0036] Through the above steps, you can sort by feature importance and store the feature names and corresponding feature importance in a dictionary, so as to better understand and analyze which features contribute most to the predictive ability of the model.

[0037] In one implementation, a visualization report is generated based on the feature data stored in the dictionary.

[0038] To facilitate visualization, the data in the dictionary needs to be converted into a table format. The keys and values ​​in the dictionary can be extracted separately to form two lists: one containing the feature names and the other containing the feature importance. Choose a visualization tool to convert the table into a bar chart, pie chart, box plot, etc. Common visualization tools include: Matplotlib, Seaborn, Plotly, etc.

[0039] The combined feature matrix contains a large amount of structured and unstructured data, which enables the model to comprehensively consider all relevant information, rather than being limited to a single dimension or a few indicators. This can result in a more comprehensive and accurate evaluation result. The extraction of feature importance helps to understand how the model makes predictions and increases the transparency of the model. This is very important for explaining model results to stakeholders, especially when major decisions need to be made based on these results. By monitoring changes in key features, early warning signals can be issued in advance to help companies prepare and avoid potential risks.

[0040] Step S104: input the new enterprise digital evaluation document data into the trained model. The model automatically assigns weights to the new enterprise digital evaluation document data features according to the feature importance learned from the historical enterprise digital evaluation documents to obtain the digital evaluation results.

[0041] Before inputting the new enterprise digital evaluation document into the model, its text data and structured data need to be preprocessed, and the processed text data and structured data are merged into a feature matrix. The specific processing method can be referred to step S102, which will not be repeated here. After that, the feature matrix data is input into the trained model, and the model automatically assigns weights to the new enterprise digital evaluation document data features according to the feature importance learned from the historical enterprise digital evaluation documents, and obtains the digital evaluation results.

[0042] In one embodiment, the method for automatically assigning weights to features of new enterprise digitalization evaluation document data according to the learned feature importance of the model includes: traversing the features of the new enterprise digitalization evaluation document data, searching for the corresponding feature importance from the weight dictionary according to the feature name, and assigning the found feature importance to the corresponding feature of the new enterprise digitalization evaluation document data.

[0043] First, create a new column to store the weight of each feature in the new data, traverse each row and column of the new data, find the corresponding feature importance from the weight dictionary according to the feature name, and assign the found feature importance to the new column. Finally, check the processed data to ensure that the new column has contained the feature importance of each sample.

[0044] In one embodiment, for features that exist in the training data but are missing in the new enterprise digital evaluation document data, these features are added to the new enterprise digital evaluation document data and their values ​​are set to 0; for features that exist in the new enterprise digital evaluation document data but are not present in the training data, these features are retained in the new enterprise digital evaluation document data.

[0045] Get the feature list of the training data, traverse the feature list of the training data, check whether these features exist in the new data, if a feature is missing in the new data, add the feature to the new data and set its value to 0, if a feature in the new data is missing in the training data, keep the feature in the new data. To ensure that the feature order of the new data is consistent with the training data, the feature order of the new data can be rearranged.

[0046] Through the above steps, it can be ensured that the characteristics of the new data are consistent with the characteristics of the training data, thereby ensuring the consistency of the input data format of the model and improving the prediction accuracy of the model.

[0047] In one embodiment, for features that exist in new enterprise digitalization evaluation document data but do not exist in training data, after retaining these features in the new enterprise digitalization evaluation document data, it also includes: manually adjusting the importance of the retained features, or using the new enterprise digitalization evaluation document data to incrementally train the model through an incremental learning method, updating the parameters of the model, and automatically adjusting the importance of the features through the model.

[0048] Depending on the scenario or the requirements for data, two methods can be used to handle the situation where features exist in the new enterprise digital assessment document data but do not exist in the training data. One method is: after identifying the features that exist in the new data but do not exist in the training data, these new features can be manually assigned importance based on domain knowledge or expert opinions, and these manually assigned importances can be added to the weight dictionary. Another method is: incrementally train the model using new enterprise digital assessment document data through the incremental learning method. Specifically, the new data is added to the training data set to form a new training data set, and the new training data set is gradually input into the model for incremental training. The performance of the model is regularly evaluated to ensure that the model can adapt to the new features and maintain a high prediction accuracy. The model will automatically adjust its importance based on the performance of the new features in the data, and update the parameters of the model. The incrementally trained model is saved to a file for subsequent use.

[0049] Through the above steps, it is possible to ensure that the characteristics of the new enterprise digital assessment document data are consistent with those of the training data, and to process the characteristics that exist in the new data but not in the training data. The importance of the new features can be manually adjusted or the parameters of the model can be updated through the incremental learning method to ensure that the model can adapt to the new features and improve the prediction accuracy.

[0050] In one implementation, a digital report may be generated based on the digital evaluation results.

[0051] The window for generating digital reports can provide a variety of output formats and customization options to meet the needs of different users. Specifically, for scenarios that require detailed data and text descriptions, the window can generate a comprehensive analysis report to provide users with detailed data analysis and professional advice. For scenarios that require intuitive presentation of charts, the window can generate pie charts, bar charts, line charts, etc. to help quickly identify key information in the data. In addition, you can also choose different output formats such as PDF, Excel, PPT, etc. according to the needs of other scenarios.

[0052] By providing detailed data analysis results, digital reports help management make more informed investment decisions based on evidence, including which areas need to be strengthened and which projects may bring higher returns. By conducting an in-depth review of existing systems and processes, digital assessment reports can also help companies identify potential security vulnerabilities or other risk factors, and take preventive measures in advance to reduce losses.

[0053] Corresponding to the above method provided by the present invention, the present invention also provides a device. Figure 2 The following is a schematic diagram showing the structure of an enterprise digital evaluation device based on AI enhancement provided in an embodiment of this specification. Figure 2 As shown, the device comprises: The data collection module S201 is used to collect enterprise digital evaluation document data.

[0054] The data merging module S202 is used to merge the text data and structured data in the enterprise digital evaluation document into a feature matrix.

[0055] The model training module S203 is used to use the merged feature matrix and the digital evaluation results to perform model training and extract the feature importance of the historical enterprise digital evaluation documents.

[0056] The digital evaluation module S204 is used to input new enterprise digital evaluation document data into the trained model. The model automatically assigns weights to the new enterprise digital evaluation document data features according to the feature importance learned from the historical enterprise digital evaluation documents to obtain the digital evaluation results.

[0057] It is worth mentioning that Figure 2 The device shown is Figure 1 The method embodiments shown correspond to and can be applied to Figure 1 The method embodiments shown include data collection of enterprise digital assessment documents, merging of text data and structured data in enterprise digital assessment documents, model training using the merged feature matrix and digital assessment results, and evaluation of new enterprise digital assessment document data. Those skilled in the art should easily appreciate that, in combination with the units and algorithm steps of each example described in the embodiments disclosed herein, the present application can be implemented in the form of hardware or a combination of hardware and computer software. Whether a function is executed in the form of hardware or computer software driving hardware depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0058] According to another embodiment of the present invention, a computing device is provided, including a memory and a processor, wherein the memory stores executable code, and when the processor executes the executable code, the Figure 1 The method described.

[0059] According to another embodiment, there is also provided a computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute a combination of Figure 1 The method described.

[0060] Those skilled in the art should be aware that in one or more of the above examples, the functions described in the present invention can be implemented by hardware, software, firmware or any combination thereof. When implemented by software, these functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium.

[0061] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solution of the present invention should be included in the scope of protection of the present invention.

Claims

1. An enterprise digitalization evaluation method based on AI enhancement, characterized in that: The method comprises: Collect digital evaluation documents of historical enterprises, where the digital evaluation documents of historical enterprises include text data, structured data and digital evaluation results; Combine text data and structured data from historical enterprise digital assessment documents into a feature matrix; Use the merged feature matrix and digital evaluation results to train the model and extract the feature importance of historical enterprise digital evaluation documents; The new enterprise digital assessment document data is input into the trained model. The model automatically assigns weights to the new enterprise digital assessment document data features according to the feature importance learned from the historical enterprise digital assessment documents to obtain the digital assessment results.

2. The method according to claim 1, characterized in that Before merging text data and structured data into a feature matrix, the text data is preprocessed as follows: Segment new text data into words or phrases; Remove meaningless words from text data; Convert the processed text data into a numeric vector.

3. The method according to claim 1, characterized in that Before merging text data and structured data into a feature matrix, the structured data is cleaned and standardized.

4. The method according to claim 1, characterized in that: After extracting the feature importance of the historical enterprise digital assessment document, it is sorted by feature importance, and then the feature name and the corresponding feature importance are stored in a dictionary.

5. The method according to claim 4, characterized in that Methods for the model to automatically assign weights to new enterprise digital assessment document data features based on learned feature importance include: Traverse the features of the new enterprise digital assessment document data, and find the corresponding feature importance from the weight dictionary according to the feature name; The found feature importance is assigned to the corresponding features of the new enterprise digital evaluation document data.

6. The method according to claim 4, characterized in that Generate a visual report based on the feature data stored in the dictionary.

7. The method according to claim 1, characterized in that For features that exist in the training data but are missing in the new enterprise digitalization assessment document data, these features are added to the new enterprise digitalization assessment document data and their values ​​are set to 0. For features that exist in the new enterprise digitalization assessment document data but are not present in the training data, these features are retained in the new enterprise digitalization assessment document data.

8. The method according to claim 7, characterized in that For features that exist in the new enterprise digitalization evaluation document data but not in the training data, after retaining these features in the new enterprise digitalization evaluation document data, the following are also included: By manually adjusting the importance of the retained features; Alternatively, the model can be incrementally trained using new enterprise digital assessment document data through incremental learning methods, the model parameters can be updated, and the importance of features can be automatically adjusted through the model.

9. The method according to claim 1, characterized in that: Generate digital reports based on the digital assessment results.

Images