Software demand analysis method and device, equipment, medium and program product

Abstract

The invention provides a software demand analysis method which can be applied to the field of artificial intelligence. The method comprises the steps that demand information of target software is collected in a multi-modal mode, and initial data of m modals are obtained; wherein m is a positive integer greater than 1; based on the initial data of the m modalities, performing multi-modal demand analysis by using a first large language model to obtain information data of the m modalities; performing information fusion on the information data of the m modes to obtain software demand data; and performing demand optimization processing on the software demand data to obtain a software demand document corresponding to the target software. According to the software demand analysis method and device, intelligent demand analysis and automatic and intelligent processing of the target software are achieved, manual intervention is reduced, the processing efficiency is improved, and the invention further provides a software demand analysis device and device, a storage medium and a program product.

Description

Technical Field

[0001] The present disclosure relates to the field of artificial intelligence, and more specifically to a software requirements analysis method, apparatus, device, medium and program product. Background Art

[0002] Software requirements analysis is a key stage in the software development life cycle, and its purpose is to determine what functions the software system must complete and how to complete these functions. In the process of software development, the software requirements document obtained through analysis describes the software functional and non-functional requirements in detail, which helps to improve the transparency of the software development process and enables team members to better understand the progress of the project and the basis for decision-making. Good requirements document writing and management helps to ensure the consistency and accuracy of the project and improve development efficiency.

[0003] Generally, software requirements analysis is conducted in a traditional way, where humans are responsible for recording and writing. There is also a template method, which obtains software requirements information and the corresponding business type, determines the business requirements information, and then modifies the prompt word template based on this information to generate the target software requirements document. However, both the traditional recording method and the template method have the following disadvantages: (1) They mainly rely on text input, are limited by the text content, and have a single modality limitation; (2) They are highly dependent on humans, rely on human judgment and intervention, and have insufficient automation; (3) They are not dynamic enough, and cannot update the requirements document in real time to adapt to the dynamic changes in requirements. Summary of the invention

[0004] In view of the above problems, the present disclosure provides a software requirement analysis method, apparatus, device, medium and program product that improve the degree of automation and intelligence.

[0005] According to a first aspect of the present disclosure, a software requirement analysis method is provided, comprising: collecting requirement information of a target software in a multimodal manner to obtain initial data of m modes; wherein m is a positive integer greater than 1; based on the initial data of the m modes, performing multimodal requirement analysis using a first large language model to obtain information data of the m modes; performing information fusion on the information data of the m modes to obtain software requirement data; and performing requirement optimization processing on the software requirement data to obtain a software requirement document corresponding to the target software.

[0006] According to an embodiment of the present disclosure, the modalities include text modality, voice modality and visual modality, and the multimodal method of collecting the target software demand information to obtain the initial data of m modalities includes: using a text collection method to collect the user's demand text information for the target software to obtain the initial data of the text modality; using a voice method to collect the user's demand voice information for the target software, and using voice recognition technology to process the demand voice information to obtain the initial data of the voice modality; and collecting images and videos by capturing the user's behavior and environment to obtain the initial data of the visual modality.

[0007] According to an embodiment of the present disclosure, the multimodal demand analysis based on the initial data of the m modalities is performed using a first large language model to obtain information data of the m modalities, including: preprocessing the initial data of the m modalities; performing multimodal conversion processing on the preprocessed initial data of the m modalities to obtain feature data of the m modalities; and analyzing the feature data of the m modalities using the first large language model to obtain information data of the m modalities.

[0008] According to an embodiment of the present disclosure, the use of the first large language model to analyze the feature data of the m modalities to obtain the information data of the m modalities includes: using the first large language model to perform semantic sentiment analysis on the feature data of the text modality to obtain the information data of the text modality; using the first large language model to perform semantic understanding on the feature data of the speech modality to obtain the information data of the speech modality; and using the first large language model to perform content scene analysis on the feature data of the visual modality to obtain the information data of the visual modality.

[0009] According to an embodiment of the present disclosure, the software requirement data is subjected to requirement optimization processing to obtain a software requirement document corresponding to the target software, including: using a second language model to perform user requirement verification on the software requirement data, and refining the verified software requirement data; normalizing the refined software requirement data to obtain an initial requirement document; and using a third language model to prioritize the initial requirement document to obtain a software requirement document corresponding to the target software.

[0010] According to an embodiment of the present disclosure, the method of using the third largest language model to prioritize the initial requirement document to obtain the software requirement document corresponding to the target software includes: extracting key element features of the initial requirement document; based on the key element features, using the third largest language model to perform criticality assessment on each requirement content in the initial requirement document to obtain n criticality assessment levels corresponding to each requirement content; wherein n is a positive integer greater than 1; converting the n criticality assessment levels corresponding to each requirement content into a numerical score to obtain a priority score for each requirement content; and prioritizing the initial requirement document according to the priority score of each requirement content to obtain the software requirement document corresponding to the target software.

[0011] According to an embodiment of the present disclosure, the preprocessed initial data of m modes are subjected to multimodal conversion processing to obtain feature data of the m modes, including: extracting modal data features corresponding to the preprocessed initial data of the m modes to obtain feature information of the m modes; and converting the feature information of the m modes into a data structured form to obtain feature data of the m modes.

[0012] According to an embodiment of the present disclosure, the preprocessing of the initial data of the m types of modalities includes: performing modality cleaning on the initial data of the m types of modalities; and formatting the cleaned initial data of the m types of modalities.

[0013] According to an embodiment of the present disclosure, when the requirement information of the target software changes, the method further includes: collecting the changed requirement information data, performing multimodal requirement processing using the first language model to update the software requirement document; and tracking and managing requirement changes based on the software requirement document and the updated software requirement document using a fourth language model, and integrating the tracking results into the development process of the target software.

[0014] According to an embodiment of the present disclosure, the information fusion of the m types of modalities of information data to obtain software requirement data includes: integrating the m types of modalities of information data to obtain fused data; and knowledge-architecting the fused data to obtain the software requirement data.

[0015] The second aspect of the present disclosure provides a software requirement analysis device, including: a requirement collection module, used to collect requirement information of a target software in a multimodal manner to obtain initial data of m modes; wherein m is a positive integer greater than 1; a multimodal analysis module, used to perform multimodal requirement analysis based on the initial data of the m modes using a first large language model to obtain information data of the m modes; an information fusion module, used to perform information fusion on the information data of the m modes to obtain software requirement data; and a requirement optimization module, used to perform requirement optimization processing on the software requirement data to obtain a software requirement document corresponding to the target software.

[0016] According to an embodiment of the present disclosure, the modalities include text modalities, voice modalities and visual modalities, and the demand collection module includes: a first demand collection submodule, which is used to collect the user's demand text information for the target software by using a text collection method to obtain the initial data of the text modality; a second demand collection submodule, which is used to collect the user's demand voice information for the target software by using a voice method, and use voice recognition technology to process the demand voice information to obtain the initial data of the voice modality; and a third demand collection submodule, which is used to collect images and videos by capturing the user's behavior and environment to obtain the initial data of the visual modality.

[0017] According to an embodiment of the present disclosure, the multimodal analysis module includes: a first multimodal analysis submodule, used for preprocessing the initial data of the m modalities; a second multimodal analysis submodule, used for performing multimodal conversion processing on the preprocessed initial data of the m modalities to obtain feature data of the m modalities; and a third multimodal analysis submodule, used for analyzing the feature data of the m modalities using the first large language model to obtain information data of the m modalities.

[0018] According to an embodiment of the present disclosure, the third multimodal analysis submodule includes: the text modality analysis unit, which is used to perform semantic sentiment analysis on the feature data of the text modality using the first large language model to obtain the information data of the text modality; the speech modality analysis unit, which is used to perform semantic understanding on the feature data of the speech modality using the first large language model to obtain the information data of the speech modality; and the visual modality analysis unit, which is used to perform content scene analysis on the feature data of the visual modality using the first large language model to obtain the information data of the visual modality.

[0019] According to an embodiment of the present disclosure, the demand optimization module includes: a first demand optimization sub-module, which is used to use the second largest language model to perform user demand verification on the software demand data and refine the verified software demand data; a second demand optimization sub-module, which is used to normalize the refined software demand data to obtain an initial demand document; and a third demand optimization sub-module, which is used to use the third largest language model to prioritize the initial demand document to obtain a software demand document corresponding to the target software.

[0020] According to an embodiment of the present disclosure, the third requirement optimization submodule includes: an element feature extraction unit, which is used to extract key element features of the initial requirement document; a criticality assessment extraction unit, which is used to perform criticality assessment on each requirement content in the initial requirement document based on the key element features and using the third language model to obtain n criticality assessment levels corresponding to each requirement content; wherein n is a positive integer greater than 1; a priority scoring unit, which is used to convert the n criticality assessment levels corresponding to each requirement content into a numerical score to obtain a priority score for each requirement content; and a priority sorting unit, which is used to perform requirement priority sorting on the initial requirement document according to the priority score of each requirement content to obtain a software requirement document corresponding to the target software.

[0021] According to an embodiment of the present disclosure, the second multimodal analysis submodule includes: a modal feature extraction unit, used to extract modal data features corresponding to the initial data of the preprocessed m modes, and obtain feature information of the m modes; and a feature structuring unit, used to convert the feature information of the m modes into a data structured form to obtain feature data of the m modalities.

[0022] According to an embodiment of the present disclosure, the first multimodal analysis submodule includes: a cleaning unit for performing modal cleaning on the initial data of the m modalities; and a formatting unit for formatting the cleaned initial data of the m modalities.

[0023] According to an embodiment of the present disclosure, when the requirement information of the target software changes, the method also includes: an iterative update module, which is used to collect the changed requirement information data and use the first language model to perform multimodal requirement processing to update the software requirement document; and based on the software requirement document and the updated software requirement document, use the fourth language model to track and manage the requirement changes, and integrate the tracking results into the development process of the target software.

[0024] According to an embodiment of the present disclosure, the information fusion module includes: the first information fusion sub-module, used to integrate the information data of the m modes to obtain fused data; and the second information fusion sub-module, used to knowledge-architect the fused data to obtain the software requirement data.

[0025] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0026] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.

[0027] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor.

[0028] In the embodiments of the present disclosure, multimodal input processing is introduced for intelligent demand analysis of target software, so that demand analysis is no longer limited to text, but can process richer multimodal information, enhance the comprehensiveness and accuracy of demand analysis, and process multimodal data by applying a large language model, and fuse and optimize the information to obtain software requirement documents, thereby realizing automated and intelligent processing, reducing manual intervention, and improving processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] The above contents and other purposes, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0030] Figure 1 The application scenario diagram of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown;

[0031] Figure 2 A flowchart of a software requirement analysis method according to an embodiment of the present disclosure is schematically shown;

[0032] Figure 3 The multimodal data collection flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown;

[0033] Figure 4 A multimodal requirement analysis flow chart of a software requirement analysis method according to an embodiment of the present disclosure is schematically shown;

[0034] Figure 5A data preprocessing flow chart of a software requirement analysis method according to an embodiment of the present disclosure is schematically shown;

[0035] Figure 6 The multi-modal conversion processing flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown;

[0036] Figure 7 A data analysis flow chart of a software requirement analysis method according to an embodiment of the present disclosure is schematically shown;

[0037] Figure 8 The information fusion flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown;

[0038] Fig. 9 A flowchart of software requirement optimization according to an embodiment of the present disclosure is schematically shown;

[0039] Fig.10 A flowchart of the software requirement analysis method according to an embodiment of the present disclosure is schematically shown;

[0040] Fig.11 The iterative update flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown;

[0041] Fig.12 A structural block diagram of a software requirement analysis device according to an embodiment of the present disclosure is schematically shown; and

[0042] Fig.13 A block diagram of an electronic device suitable for implementing a software requirement analysis method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0043] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0044] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0045] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0046] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0047] The embodiment of the present disclosure provides a method for software requirement analysis, including: collecting the requirement information of the target software in a multimodal manner to obtain initial data of m modes; wherein m is a positive integer greater than 1; based on the initial data of the m modes, using the first large language model to perform multimodal requirement analysis to obtain information data of the m modes; performing information fusion on the information data of the m modes to obtain software requirement data; and performing requirement optimization processing on the software requirement data to obtain a software requirement document corresponding to the target software. Intelligent requirement analysis of the target software is realized, and multimodal input processing is introduced, so that the requirement analysis is no longer limited to text, but can process richer multimodal information, enhance the comprehensiveness and accuracy of the requirement analysis, and by applying a large language model to process multimodal data, and fusing and optimizing the information to obtain the software requirement document, automatic and intelligent processing is realized, manual intervention is reduced, and processing efficiency is improved.

[0048] Figure 1 The application scenario diagram of the software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0049] like Figure 1 As shown, the application scenario 100 according to this embodiment may include software requirement analysis. The network 104 is used to provide a medium for a communication link between the first terminal device 101, the second terminal device 102, the third terminal device 103 and the server 105. The network 104 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0050] The user can use the first terminal device 101, the second terminal device 102, and the third terminal device 103 to interact with the server 105 through the network 104 to receive or send messages, etc. Various communication client applications can be installed on the first terminal device 101, the second terminal device 102, and the third terminal device 103, such as shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0051] The first terminal device 101, the second terminal device 102, and the third terminal device 103 may be various electronic devices having display screens and supporting web browsing, including but not limited to smart phones, tablet computers, laptop computers, desktop computers, and the like.

[0052] The server 105 may be a server that provides various services, such as a background management server (only as an example) that provides support for websites browsed by users using the first terminal device 101, the second terminal device 102, and the third terminal device 103. The background management server may analyze and process the received data such as user requests, and feed back the processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0053] It should be noted that the software requirement analysis method provided in the embodiment of the present disclosure can generally be executed by the server 105. Accordingly, the software requirement analysis device provided in the embodiment of the present disclosure can generally be set in the server 105. The software requirement analysis method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105. Accordingly, the software requirement analysis device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 105 and can communicate with the first terminal device 101, the second terminal device 102, the third terminal device 103 and / or the server 105.

[0054] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to implementation requirements.

[0055] The following will be based on Figure 1 The scene described by Figure 2~Figure 11 The software requirement analysis method of the disclosed embodiment is described in detail.

[0056] Figure 2 The flowchart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown.

[0057] like Figure 2 As shown, the software requirement analysis method of this embodiment includes operations S210 to S240, and the software requirement analysis method can be automatically executed.

[0058] In operation S210, the requirement information of the target software is collected in a multimodal manner to obtain initial data of m modes; wherein m is a positive integer greater than 1.

[0059] Multimodal methods include but are not limited to text collection, voice collection, image collection and video collection. Through various collection methods, comprehensive information on user (customers, users, stakeholders, etc.) requirements for the target software can be collected. By communicating with users, the user's needs and expectations can be understood, and business requirements, user requirements and system requirements can be collected. The collected requirements can also be classified into different categories, such as functional requirements and non-functional requirements.

[0060] Before collecting the user's text, voice, image and video book data, the user's consent or authorization may be obtained. For example, before operation S210, a request for obtaining user information may be issued to the user, and operation S210 is performed when the user agrees or authorizes the user information to be obtained.

[0061] In the technical solution of the present disclosure, the user's demand information for the target software (including but not limited to user personal information, user software information, user voice information, user operation behavior information, user image information, user environment information, user device information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0062] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating an individual's usage needs, behavioral habits, interests, hobbies, operating habits, preferences, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge, and skills, and have reached a certain level of professionalism.

[0063] In operation S220, based on the initial data of the m modes, a multimodal demand analysis is performed using a first large language model to obtain information data of the m modes.

[0064] A large language model refers to an artificial intelligence language model with a large number of parameters and high learning ability, which can process and understand complex language information. It should be noted that the first large language model is an artificial intelligence language model used to implement multimodal demand analysis after being trained with a large amount of modal data. Multimodal demand analysis is a method of demand analysis using multiple input methods (such as text, voice, image, video and other multimodal data).

[0065] The method of using large language models for multimodal requirements analysis can provide more comprehensive and accurate requirements analysis results, improve the efficiency and quality of requirements analysis, and help development teams better understand and meet user needs.

[0066] In operation S230, information fusion is performed on the information data of the m modes to obtain software requirement data.

[0067] Information fusion can fuse information data of different modalities. It can adopt a simple direct splicing method, or it can adopt weighted splicing to assign different weights to the feature vectors of each modality according to the importance or relevance of different modalities to the expression of software requirements, and then splice them. It can also adopt simple voting method or probabilistic fusion method.

[0068] In operation S240, the software requirement data is subjected to requirement optimization processing to obtain a software requirement document corresponding to the target software.

[0069] In order to form a complete software requirement document, the analyzed software requirement data needs to be optimized. The requirement optimization process includes but is not limited to functional requirement combing, non-functional requirement classification, requirement priority sorting, document structure planning, and generation of detailed requirement documents.

[0070] In the embodiments of the present disclosure, multimodal input processing is introduced for intelligent demand analysis of target software, so that demand analysis is no longer limited to text, but can process richer multimodal information, enhance the comprehensiveness and accuracy of demand analysis, and process multimodal data by applying a large language model, and fuse and optimize the information to obtain software requirement documents, thereby realizing automated and intelligent processing, reducing manual intervention, and improving processing efficiency.

[0071] Figure 3 The multimodal data collection flow chart of the software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0072] like Figure 3As shown, according to an embodiment of the present disclosure, the modality includes a text modality, a voice modality and a visual modality, and in operation S210, the requirement information of the target software is collected in a multimodal manner to obtain initial data of m modalities, including operations S310-S330.

[0073] In operation S310, a text collection method is used to collect text information of user requirements for target software to obtain initial data in a textual mode.

[0074] Through text collection methods such as survey questionnaires, interviews, meeting minutes, user feedback, user community forums, etc., the software requirements in text state are collected through text input to obtain the initial data in text mode.

[0075] In operation S320, voice information of the user's demand for the target software is collected in a voice manner, and the demand voice information is processed using voice recognition technology to obtain initial data of the voice modality.

[0076] Through voice recording, voice surveys and interviews, voice assistants, user voice feedback and other voice methods, we can collect software requirements in the voice state through voice input, use voice recognition technology to convert the user's verbal description into text, and use the first language model to convert voice data from speech to text and obtain relevant initial voice information, that is, the initial data of the voice modality.

[0077] In operation S330 , images and videos are collected by capturing the user's behavior and environment to obtain initial data of the visual modality.

[0078] You can use embedded recording tools to capture images of the software interface at a certain frequency (such as one frame per second), or record user operation videos. You can also use external devices (such as mobile phone cameras, video cameras, and professional screen recording devices) to record the process of users using the software or the environment in which the software is located. Capture user behavior and environment through images and videos. Collect software requirements in the visual state through image and video input, and obtain initial data of the visual modality, which is convenient for analyzing user operation habits and preferences.

[0079] In the embodiments of the present disclosure, modal initial data is obtained through collection methods corresponding to different modalities, and multimodal collection is introduced, so that the basis of demand analysis is no longer limited to text, but can process richer information, improve information integrity and richness, flexibly respond to different scenarios, and enhance the comprehensiveness, accuracy and reliability of demand analysis.

[0080] Figure 4 A multimodal requirement analysis flow chart of a software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0081] like Figure 4As shown, according to an embodiment of the present disclosure, in operation S220, based on initial data of m modes, a multimodal demand analysis is performed using a first large language model to obtain information data of m modes, including operations S410-S430.

[0082] In operation S410 , initial data of m modes are preprocessed.

[0083] In operation S420, the preprocessed initial data of the m modes are subjected to multi-modal conversion processing to obtain feature data of the m modes.

[0084] In operation S430, feature data of the m modes are analyzed using the first large language model to obtain information data of the m modes.

[0085] In the embodiments of the present disclosure, multimodal data is processed, the modal data is analyzed using the first large prediction model, and the multimodal data content is fully understood through the first large language model, thereby enhancing the accuracy and reliability of data analysis, reducing human intervention, lowering the error rate, and improving the efficiency of demand analysis.

[0086] Figure 5 The data preprocessing flow chart of the software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0087] like Figure 5 As shown, according to an embodiment of the present disclosure, in the operation S410 of pre-processing the initial data of m modes, operations S510-S520 are included.

[0088] In operation S510 , modality cleaning is performed on initial data of m modalities.

[0089] Different cleaning methods are used for different modal data, as follows:

[0090] Clean the initial data of the text modality: Use natural language processing tools to check and correct grammatical and spelling errors in the text. You can set keyword filtering rules to filter out keywords that are not related to software requirements when they appear in the text; check whether the text is complete and whether there is any truncation; for incomplete text, if the missing part can be determined, it can be filled in through context speculation; for text data, you can use hash algorithms (such as MD5, SHA-1) or text similarity calculation methods (such as cosine similarity, edit distance) to detect duplicate text; if the duplicate data has no additional value for subsequent analysis, or will affect the accuracy of the analysis results, you can choose to delete the duplicate data.

[0091] Clean the initial data of the speech mode: For environmental noise in the audio (such as background noise, electric current sound), you can use audio noise reduction software or algorithms; you can use image file format verification, pixel value range check and other methods to find out whether there are audio fragments missing, channel imbalance, missing pixels or channels, etc.; if there is a short audio loss, you can fill it through interpolation and other methods; for audio, extract audio features (such as Mel-frequency cepstral coefficients) and detect repeated audio by comparing the similarity of features; if the repeated data has no additional value for subsequent analysis or will affect the accuracy of the analysis results, you can choose to delete the repeated data.

[0092] Clean the initial data of the visual modality: For the common salt and pepper noise and Gaussian noise in the image, filtering technology can be used to process them. There are some objects in the image that are irrelevant to the research purpose (such as watermarks of the shooting equipment, irrelevant background objects), which can be removed using image editing technology or target detection and segmentation algorithms. In the video, remove the irrelevant segments at the beginning and end, and determine the location of these irrelevant segments through manual marking or automatic scene detection algorithms. For the video, check whether the video frame is complete, whether the audio and video are synchronized, etc.; For small parts of the image that are damaged or missing pixels, you can use the image repair algorithm to fill them. For the video, if a small number of frames are lost, you can copy adjacent frames or use inter-frame interpolation algorithms to repair them; for images, you can calculate the feature vector of the image (such as using a convolutional neural network to extract image features) and then compare the similarity of the feature vectors to detect duplicate images. For videos, you can consider the image features and audio features of the video frame to detect duplication; if the duplicate data has no additional value for subsequent analysis or will affect the accuracy of the analysis results, you can choose to delete the duplicate data.

[0093] In operation S520 , the cleaned initial data of the m modes are formatted.

[0094] For text modalities, unify the text encoding format to ensure that all text data can be read correctly.

[0095] For speech modes, unify the audio sampling rate, number of channels, etc.

[0096] For visual modalities, unify the image size, color mode, etc., and unify the video frame rate to ensure consistency in data format.

[0097] In the embodiments of the present disclosure, the collected initial data of multiple modalities such as text, voice, image, video, etc. are cleaned and formatted to effectively remove invalid data to adapt to subsequent analysis and processing, improve data quality, facilitate accurate analysis of demand information in the later stage, and enhance demand analysis and mining effects.

[0098] Figure 6 The multimodal conversion processing flow chart of the software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0099] like Figure 6 As shown, according to an embodiment of the present disclosure, in operation S420, the pre-processed initial data of m modes are subjected to multi-modal conversion processing to obtain feature data of m modes, including operations S610-S620.

[0100] In operation S610, modal data features corresponding to the preprocessed initial data of the m modes are extracted to obtain feature information of the m modes.

[0101] Extract features from data of various modalities, such as keywords in text, intonation of speech, color and shape of images, motion information of videos, etc. Feature extraction methods for text modality include but are not limited to using bag-of-words models, word frequency-inverse document frequency models, word vector models, etc. Feature extraction methods for speech modality include but are not limited to using time domain feature extraction, frequency domain feature extraction, and using deep learning models such as convolutional neural networks (CNN), recurrent neural networks (RNN) and their variants (such as long short-term memory networks (LSTM), gated recurrent units (GRU), etc. Image feature extraction methods for visual modality include but are not limited to using color feature extraction, texture feature extraction, shape feature extraction, convolutional neural network feature extraction, etc. Video feature extraction methods for visual modality include but are not limited to image feature extraction, spatiotemporal feature extraction, etc.

[0102] In operation S620, feature information of the m modes is converted into a data structured form to obtain feature data of the m modes.

[0103] Convert data of different modalities into a unified representation, such as vectorized representation, matrix representation, three-dimensional tensor representation, graph structure representation, tree structure representation, etc., to facilitate subsequent fusion processing.

[0104] In the embodiments of the present disclosure, feature extraction is performed on multimodal data and converted into a structured data form. The key feature information extracted from different modal data is converted into a unified representation form to break down modal barriers and facilitate data storage and management. The unified structured data can make it easier for large language models to learn common features across modalities and enhance the model application effect.

[0105] Figure 7 The data analysis flow chart of the software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0106] like Figure 7As shown, according to an embodiment of the present disclosure, in operation S430, the feature data of m modes are analyzed using the first large language model to obtain information data of m modes, including operations S710-S730.

[0107] In operation S710, a semantic sentiment analysis is performed on feature data of the text modality using a first language model to obtain information data of the text modality.

[0108] When the feature data of the text modality is input into the No.1 language model, the No.1 language model provides a variety of analysis modes, such as sentence-based analysis, paragraph-based analysis, and overall text analysis. It is also necessary to select the appropriate analysis mode according to the characteristics of the text and the purpose of the analysis. And determine the emotional category to be analyzed, such as positive, negative, neutral, etc. After analysis by the No.1 language model, the No.1 prediction model will output the information data of the text modality to achieve semantic understanding and emotional analysis.

[0109] In operation S720, semantic understanding is performed on feature data of the speech modality using the first language model to obtain information data of the speech modality.

[0110] According to the characteristics of the speech-to-text data and the purpose of analysis, select the appropriate analysis mode of the first language model, and set the degree of deep semantic understanding of the data, such as only analyzing the surface meaning of the text or digging deeper into its implicit meaning. Use the first language model for semantic understanding. After the first language model analyzes the feature data of the speech modality, it outputs the information data of the speech modality to achieve semantic understanding and sentiment analysis.

[0111] In operation S730, a content scene analysis is performed on the feature data of the visual modality using the first language model to obtain information data of the visual modality.

[0112] According to the type of visual data (image or video) and the purpose of analysis, select the appropriate analysis mode, such as image-based analysis, video frame-based analysis, overall video analysis, etc., and determine the content scene classification category to be analyzed, such as indoor scenes, outdoor scenes, character scenes, landscape scenes, business scenes, process operation scenes, etc. Use the first language model to perform content scene analysis. After the first language model analyzes the feature data of the visual modality, the information data of the visual modality is output to achieve semantic understanding and sentiment analysis, and to achieve content recognition and scene understanding.

[0113] In the embodiments of the present disclosure, the first language model adopts different semantic analysis methods for feature data of different modalities, and can perform precise semantic extraction for different modal data to obtain more accurate demand information, enhance collaborative understanding between modalities, and improve data mining depth and efficiency.

[0114] Figure 8 The information fusion flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown.

[0115] like Figure 8 As shown, according to an embodiment of the present disclosure, in operation S230, information fusion is performed on information data of m modes to obtain software requirement data, including operations S810-S820.

[0116] In operation S810, information data of m modes are integrated to obtain fused data.

[0117] Integrate the information data extracted from different modal data to form a comprehensive understanding of user needs. This can be done by building a fusion model (such as a multi-layer perceptron, support vector machine, etc.), taking a large number of samples of information data of different modalities as input, training this fusion model to learn the best fusion strategy, and using the trained fusion model to integrate information data of different modalities to obtain fused data.

[0118] In operation S820, the fused data is knowledge-structured to obtain software requirement data.

[0119] Knowledge structuring realizes the abstraction and generalization of requirements to form demand data for high-level requirement description. Knowledge structuring includes requirement classification and organization, key element extraction and correlation analysis, and abstraction and generalization of high-level requirements. Requirements classification and organization: Classify requirements according to functional modules, and divide non-functional requirements into performance requirements, security requirements, user experience requirements, etc. Key element extraction and correlation analysis: Extract key elements in each functional category and find the correlation between different functional categories and non-functional requirements. Abstraction and generalization of high-level requirements: Based on the key elements and correlation analysis of functional modules, summarize high-level functional requirements, form high-level descriptions for non-functional requirements, and integrate high-level functional requirements and non-functional requirements to form a complete high-level requirement description framework.

[0120] In the embodiments of the present disclosure, multimodal information data is integrated and information knowledge is structured, so that information extracted from different modal data can be integrated to form a comprehensive understanding of user needs. Knowledge structured realizes the abstraction and generalization of needs and forms a high-level description of needs.

[0121] Fig. 9 The requirement optimization flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown.

[0122] like Fig. 9 As shown, according to an embodiment of the present disclosure, in operation S240, the software requirement data is subjected to requirement optimization processing to obtain a software requirement document corresponding to the target software, including operations S910-S930.

[0123] In operation S910, the second largest language model is used to perform user requirement verification on the software requirement data, and the verified software requirement data is refined.

[0124] Communicate with users using questions and suggestions generated by the second largest language model to further clarify requirements, and verify and refine the accuracy of understanding requirements through interaction with users. Adjust and refine the requirements description based on user feedback to ensure that it accurately reflects the development goals and expectations of the software, verify the correctness and completeness of the requirements through review, testing and other methods, and provide clear guidance for subsequent system design, development and testing. It should be noted that the second largest language model is an artificial intelligence language model used to implement user requirements verification after being trained with a large amount of user communication requirements data.

[0125] In operation S920, the refined software requirement data is normalized to obtain an initial requirement document.

[0126] Standardize the requirements that have been understood, verified, and refined, clarify the specific content of the requirements, including input, processing, and output, define detailed specifications of the requirements, such as performance requirements, safety requirements, etc., and form a formal initial requirements document. The initial requirements document should contain key information extracted from different modal data and expressed in a clear and consistent manner. Write a requirements specification that describes each requirement in detail, and the requirements specification should be clear, consistent, complete, and testable.

[0127] In operation S930, the initial requirement document is prioritized using the third language model to obtain a software requirement document corresponding to the target software.

[0128] In the requirement prioritization, the third language model is used to evaluate the importance and urgency of the requirements to determine the priority of the requirements, and the final software requirement document is determined based on the prioritization results. It should be noted that the third language model is an artificial intelligence language model used to implement criticality assessment after being trained with a large amount of requirement assessment data.

[0129] In the embodiments of the present disclosure, the software requirement data obtained by processing the large language model can be further verified, refined, and prioritized to obtain a final software requirement document, verify and refine the accuracy of the understanding of the requirements, further clarify the requirements, and extract the software requirement document that best meets the actual requirements, so as to improve the success rate of software design, accurately plan the direction of software development projects, and reasonably allocate project resources.

[0130] Fig.10 A priority sorting flowchart of a software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0131] like Fig.10 As shown, according to an embodiment of the present disclosure, in operation S930, the initial requirement document is prioritized using the third language model to obtain a software requirement document corresponding to the target software, including operations S1010-S1040.

[0132] In operation S1010 , key element features of the initial requirement document are extracted.

[0133] Extract key element features from the initial requirement document. Key element features include but are not limited to requirement description, expected impact, relevant stakeholders, functional and non-functional indicators of requirements, etc.

[0134] In operation S1020, based on the key element features, the third language model is used to evaluate the criticality of each requirement content in the initial requirement document to obtain n criticality evaluation levels corresponding to each requirement content; wherein n is a positive integer greater than 1.

[0135] The third language model is an evaluation model that can quantify the importance and urgency of each requirement based on the extracted features. The historical project data and expert evaluation results can be used to train the large language model, so that it can identify the relationship between different requirement features and requirement priorities to determine the criticality of each requirement, where criticality includes but is not limited to importance and urgency.

[0136] Requirement importance assessment: The third language model is used to analyze the scope, depth and long-term value of each requirement to assess its importance. For example, if a requirement significantly improves the user experience or has a significant impact on the company's market competitiveness, it may be rated as high importance.

[0137] Demand urgency assessment: Use the third language model to assess the time sensitivity of each requirement, taking into account factors such as changes in market demand, technological development, competitor dynamics, etc. The urgency of the requirement may be associated with the project deadline, seasonal factors, or specific events.

[0138] In operation S1030 , the n criticality evaluation levels corresponding to each requirement content are converted into numerical scores to obtain a priority score of each requirement content.

[0139] Importance and urgency are converted into a numerical score to facilitate comparison of the relative priorities of different requirements. After comprehensively analyzing the feasibility of the requirements and considering factors such as technology, time, and cost, the priority of the requirements, i.e., the priority score, is determined to determine which requirements are necessary and which are optional.

[0140] In operation S1040, the initial requirement document is sorted by requirement priority according to the priority score of each requirement content to obtain a software requirement document corresponding to the target software.

[0141] According to the results of the priority scoring, the requirements are sorted by priority to form a priority list and a software requirement document based on priority sorting is formed.

[0142] In the embodiments of the present disclosure, the third language model is used to evaluate and quantify priorities to achieve priority sorting of the initial demand documents. Through the large prediction model, intelligent demand priority sorting and automated demand priority sorting are achieved, which improves decision-making efficiency and improves the scientificity and rationality of demand management.

[0143] Fig.11 The iterative update flow chart of the software requirement analysis method according to the embodiment of the present disclosure is schematically shown.

[0144] like Fig.11 As shown, according to an embodiment of the present disclosure, after operation S240, a software requirement document corresponding to the target software is obtained. When the requirement information of the target software changes, a requirement iteration update process is also included, including operations S1110-S1120.

[0145] In operation S1110, the changed requirement information data is collected, and multimodal requirement processing is performed using the first language model to update the software requirement document.

[0146] During the software development process, requirements may change at any time, so it is also necessary to pay attention to changes in requirement information. It should be noted that the method of collecting changed requirement information data is the same as operation S210, and multimodal requirement processing includes multimodal requirement analysis, information fusion, and requirement optimization processing, which is the same as the processing process of operations S220-S240, which will not be repeated here. The specific method can refer to the above operations S220-S240.

[0147] Among them, the multimodal demand analysis of the first large language model is an ongoing process, which requires continuous collection of new data and updating of the first large language model to adapt to changes in demand. The results of multimodal demand analysis can be integrated into the software development process as the basis for design and implementation, and the effects of multimodal demand analysis can be regularly evaluated, and the relevant large language models and processes can be optimized based on feedback.

[0148] In operation S1120, based on the software requirement document and the updated software requirement document, the fourth language model is used to track and manage the requirement changes, and the tracking results are integrated into the development process of the target software.

[0149] The fourth language model is used to track and manage demand changes to ensure the consistency and traceability of demand. It should be noted that the fourth language model is an artificial intelligence language model used to implement demand tracking management after being trained with a large amount of software requirement document data before and after demand changes. The tracking management process is as follows:

[0150] (1) Input the software requirements document into the 4L model in a suitable format (such as plain text, markup language, etc.). Ensure that the requirements document is complete, including functional requirements, non-functional requirements (such as performance, security, user experience, etc.), business process descriptions, etc. The 4L model can be pre-trained or fine-tuned based on the requirements document of a specific project and related domain knowledge.

[0151] (2) When a new version of the requirements document is released, the new version is also input into the fourth language model. The fourth language model automatically generates a detailed change record based on the identified changes in requirements, that is, the changed requirements information data. The change record should include the time of the change, the requirements module involved, the specific change content (such as the description of the new function, the details of the modified function, the deleted function points, etc.), and the initiator of the change (such as the business department, customer, etc.).

[0152] (3) The fourth language model performs functional impact analysis on the identified requirement changes. It considers the impact of each requirement change on the existing system functions and determines whether it will cause abnormalities or failures in existing functions or require corresponding adjustments. For non-functional requirements, the large language model analyzes the impact of requirement changes on performance, security, user experience, etc. For example, if the requirement change involves adding a large amount of data storage requirements, the model will point out that this may have an impact on the system's storage performance, data backup and recovery strategies, etc.; if major changes are made to the user interface, the model will evaluate the impact on the user experience, such as whether it will complicate the operation process and whether the interface aesthetics will be affected.

[0153] (4) Integrate the fourth language model with project management tools to monitor the approval, development, testing and other process links of requirement changes. It can keep track of the stage of each requirement change (such as waiting for approval, under development, under testing) in real time, and remind relevant personnel to handle it in a timely manner according to the preset schedule and rules. Utilize the knowledge extraction ability of the fourth language model to build a requirement knowledge graph, integrating multi-source information such as requirement documents, change records, and related discussions (such as meeting minutes and email communications). Through this knowledge graph, it is easy to trace the source of requirements and query all historical records of a requirement from the initial proposal to subsequent changes.

[0154] The fourth language model can integrate the tracking results (demand changes) into the development process of the target software and automatically generate content for communication with different stakeholders (such as customers, development teams, test teams, etc.). In terms of team collaboration, the large language model can analyze the professional skills and workload of team members, and combine the task allocation of demand changes to make reasonable collaboration suggestions.

[0155] In the embodiments of the present disclosure, when the requirement information changes, the first language model is used to perform multimodal analysis, and the fourth language model is used to track and manage the requirement information. The analysis results are integrated into the software development process, and a dynamic tracking mechanism that can update the requirement document in real time is required to continuously collect new data and update the model to adapt to the rapid changes in requirements, improve the real-time and adaptability of the requirement document, and help the software development team and relevant stakeholders quickly understand the changes in requirements.

[0156] Based on the above software requirement analysis method, the present disclosure also provides a software requirement analysis device. Fig.12 The device is described in detail.

[0157] Fig.12 The structural block diagram of the software requirement analysis device according to an embodiment of the present disclosure is schematically shown.

[0158] like Fig.12 As shown, the software requirement analysis device 1200 of this embodiment includes a requirement collection module 1210 , a multimodal analysis module 1220 , an information fusion module 1230 and a requirement optimization module 1240 .

[0159] The requirement collection module 1210 is used to collect the requirement information of the target software in a multimodal manner to obtain initial data of m modes, wherein m is a positive integer greater than 1. In one embodiment, the requirement collection module 1210 can be used to perform the operation S210 described above, which will not be described in detail here.

[0160] The multimodal analysis module 1220 is used to perform multimodal demand analysis based on the initial data of the m modes using the first language model to obtain information data of the m modes. In one embodiment, the multimodal analysis module 1220 can be used to perform the operation S220 described above, which will not be repeated here.

[0161] The information fusion module 1230 is used to fuse the information data of the m modes to obtain the software requirement data. In one embodiment, the information fusion module 1230 can be used to perform the operation S230 described above, which will not be described in detail here.

[0162] The requirement optimization module 1240 is used to perform requirement optimization processing on the software requirement data to obtain a software requirement document corresponding to the target software. In one embodiment, the requirement optimization module 1240 can be used to perform the operation S240 described above, which will not be described in detail here.

[0163] According to an embodiment of the present disclosure, the modality includes a text modality, a voice modality and a visual modality, and the demand collection module 1210 includes: a first demand collection submodule, which is used to collect user demand text information for the target software by using a text collection method to obtain initial data of the text modality; a second demand collection submodule, which is used to collect user demand voice information for the target software by using a voice method, and use voice recognition technology to process the demand voice information to obtain initial data of the voice modality; and a third demand collection submodule, which is used to collect images and videos by capturing user behavior and environment to obtain initial data of the visual modality.

[0164] According to an embodiment of the present disclosure, the multimodal analysis module 1120 includes: a first multimodal analysis submodule, used to preprocess the initial data of m modalities; a second multimodal analysis submodule, used to perform multimodal conversion processing on the preprocessed initial data of m modalities to obtain feature data of m modalities; and a third multimodal analysis submodule, used to analyze the feature data of m modalities using the first large language model to obtain information data of m modalities.

[0165] According to an embodiment of the present disclosure, the third multimodal analysis submodule includes: a text modality analysis unit, which is used to perform semantic sentiment analysis on feature data of text modality using the first largest language model to obtain information data of text modality; a speech modality analysis unit, which is used to perform semantic understanding on feature data of speech modality using the first largest language model to obtain information data of speech modality; and a visual modality analysis unit, which is used to perform content scene analysis on feature data of visual modality using the first largest language model to obtain information data of visual modality.

[0166] According to an embodiment of the present disclosure, the requirement optimization module 1240 includes: a first requirement optimization sub-module, which is used to use the second largest language model to perform user requirement verification on the software requirement data, and to refine the verified software requirement data; a second requirement optimization sub-module, which is used to normalize the refined software requirement data to obtain an initial requirement document; and a third requirement optimization sub-module, which is used to use the third largest language model to prioritize the initial requirement document to obtain a software requirement document corresponding to the target software.

[0167] According to an embodiment of the present disclosure, the third requirement optimization submodule includes: an element feature extraction unit, which is used to extract key element features of the initial requirement document; a criticality assessment extraction unit, which is used to perform criticality assessment on each requirement content in the initial requirement document based on the key element features and using the third language model to obtain n criticality assessment levels corresponding to each requirement content; wherein n is a positive integer greater than 1; a priority scoring unit, which is used to convert the n criticality assessment levels corresponding to each requirement content into a numerical score to obtain a priority score for each requirement content; and a priority sorting unit, which is used to sort the initial requirement document according to the priority score of each requirement content to obtain a software requirement document corresponding to the target software.

[0168] According to an embodiment of the present disclosure, the second multimodal analysis submodule includes: a modal feature extraction unit, used to extract modal data features corresponding to the initial data of m modes after preprocessing, and obtain feature information of the m modes; and a feature structuring unit, used to convert the feature information of the m modes into a data structured form to obtain feature data of the m modes.

[0169] According to an embodiment of the present disclosure, the first multimodal analysis submodule includes: a cleaning unit for performing modal cleaning on initial data of m modalities; and a formatting unit for formatting the cleaned initial data of the m modalities.

[0170] According to an embodiment of the present disclosure, when the requirement information of the target software changes, the software requirement analysis device 1200 also includes: an iterative update module 1250, which is used to collect the changed requirement information data, use the first language model to perform multimodal requirement processing, and update the software requirement document; and based on the software requirement document and the updated software requirement document, use the fourth language model to track and manage the requirement changes, and integrate the tracking results into the development process of the target software. In one embodiment, the iterative update module 1250 can be used to perform the operations S1110-S1120 described above, which will not be repeated here.

[0171] According to an embodiment of the present disclosure, the information fusion module 1230 includes: a first information fusion submodule, used to integrate information data of m modes to obtain fused data; and a second information fusion submodule, used to knowledge-structure the fused data to obtain software requirement data.

[0172] According to an embodiment of the present disclosure, any multiple modules of the demand collection module 1210, the multimodal analysis module 1220, the information fusion module 1230, the demand optimization module 1240 and the iterative update module 1250 can be combined in one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the demand collection module 1210, the multimodal analysis module 1220, the information fusion module 1230, the demand optimization module 1240 and the iterative update module 1250 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in an appropriate combination of any of them. Alternatively, at least one of the demand collection module 1210, the multimodal analysis module 1220, the information fusion module 1230, the demand optimization module 1240 and the iterative update module 1250 can be at least partially implemented as a computer program module, which can perform corresponding functions when executed.

[0173] Fig.13 A block diagram of an electronic device suitable for implementing a software requirement analysis method according to an embodiment of the present disclosure is schematically shown.

[0174] like Fig.13 As shown, the electronic device 1300 according to an embodiment of the present disclosure includes a processor 1301, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1302 or a program loaded from a storage part 1308 to a random access memory (RAM) 1303. The processor 1301 may include, for example, a general-purpose microprocessor (such as a CPU), an instruction set processor and / or a related chipset and / or a special-purpose microprocessor (for example, an application-specific integrated circuit (ASIC)), etc. The processor 1301 may also include an onboard memory for caching purposes. The processor 1301 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0175] In RAM 1303, various programs and data required for the operation of electronic device 1300 are stored. Processor 1301, ROM 1302 and RAM 1303 are connected to each other via bus 1304. Processor 1301 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 1302 and / or RAM 1303. It should be noted that the program can also be stored in one or more memories other than ROM 1302 and RAM 1303. Processor 1301 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in one or more memories.

[0176] According to an embodiment of the present disclosure, the electronic device 1300 may further include an input / output (I / O) interface 1305, which is also connected to the bus 1304. The electronic device 1300 may further include one or more of the following components connected to the input / output (I / O) interface 1305: an input portion 1306 including a keyboard, a mouse, etc.; an output portion 1307 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 1308 including a hard disk, etc.; and a communication portion 1309 including a network interface card such as a LAN card, a modem, etc. The communication portion 1309 performs communication processing via a network such as the Internet. A drive 1310 is also connected to the input / output (I / O) interface 1305 as needed. A removable medium 1311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 1310 as needed, so that a computer program read therefrom is installed into the storage portion 1308 as needed.

[0177] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0178] According to an embodiment of the present disclosure, a computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus, or a device. For example, according to an embodiment of the present disclosure, a computer-readable storage medium may include the ROM 1302 and / or RAM 1303 described above and / or one or more memories other than ROM 1302 and RAM 1303.

[0179] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the software requirements analysis method provided by the embodiment of the present disclosure.

[0180] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 1301. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0181] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 1309, and / or installed from the removable medium 1311. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0182] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 1309, and / or installed from the removable medium 1311. When the computer program is executed by the processor 1301, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, apparatus, module, unit, etc. described above can be implemented by a computer program module.

[0183] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0184] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0185] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0186] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A software requirements analysis method, characterized in that: The method comprises: Collect the target software requirement information in a multimodal manner to obtain initial data of m modes; where m is a positive integer greater than 1; Based on the initial data of the m modes, a multimodal demand analysis is performed using the first language model to obtain information data of the m modes; Performing information fusion on the information data of the m modes to obtain software requirement data; and The software requirement data is subjected to requirement optimization processing to obtain a software requirement document corresponding to the target software.

2. The method according to claim 1, characterized in that The modalities include text modalities, voice modalities, and visual modalities. The method of collecting the target software requirement information in a multimodal manner to obtain initial data of m modalities includes: Using a text collection method, collecting user demand text information for the target software to obtain initial data of the text modality; Using voice, collecting the user's demand voice information for the target software, and using voice recognition technology to process the demand voice information to obtain initial data of the voice modality; and By capturing the user's behavior and environment, images and videos are collected to obtain initial data of the visual modality.

3. The method according to claim 2, characterized in that The multimodal demand analysis is performed based on the initial data of the m modes using the first language model to obtain the information data of the m modes, including: Preprocessing the initial data of the m modes; Performing multimodal conversion processing on the preprocessed initial data of the m modes to obtain feature data of the m modes; and The first largest language model is used to analyze the feature data of the m modes to obtain information data of the m modes.

4. The method according to claim 3, characterized in that The using the first language model to analyze the feature data of the m modes to obtain the information data of the m modes includes: Using the first language model to perform semantic sentiment analysis on the feature data of the text modality to obtain information data of the text modality; Using the first language model to perform semantic understanding on the feature data of the speech modality to obtain information data of the speech modality; and The first language model is used to perform content scene analysis on the feature data of the visual modality to obtain information data of the visual modality.

5. The method according to claim 1, characterized in that The step of performing demand optimization processing on the software demand data to obtain a software demand document corresponding to the target software includes: Using the second language model, verifying the software requirement data for user requirements, and refining the verified software requirement data; Normalize the refined software requirement data to obtain the initial requirement document; and The initial requirement document is prioritized by using the third language model to obtain a software requirement document corresponding to the target software.

6. The method according to claim 5, characterized in that The prioritizing the initial requirement documents by using the third language model to obtain software requirement documents corresponding to the target software includes: Extracting key element features of the initial requirements document; Based on the key element features, using the third language model, a criticality assessment is performed on each requirement content in the initial requirement document to obtain n criticality assessment levels corresponding to each requirement content; wherein n is a positive integer greater than 1; Converting the n criticality evaluation levels corresponding to each requirement content into numerical scores to obtain a priority score for each requirement content; and According to the priority score of each requirement content, the initial requirement document is sorted by requirement priority to obtain a software requirement document corresponding to the target software.

7. The method according to claim 3, characterized in that The preprocessed initial data of the m modes are subjected to multimodal conversion processing to obtain feature data of the m modes, including: Extracting modal data features corresponding to the preprocessed initial data of the m modalities to obtain feature information of the m modalities; and The feature information of the m modes is converted into a data structured form to obtain the feature data of the m modes.

8. The method according to claim 3, characterized in that The preprocessing of the initial data of the m modes includes: Performing modal cleaning on the initial data of the m modes; and Format the cleaned initial data of m modes.

9. The method according to any one of claims 1 to 8, characterized in that When the requirement information of the target software changes, the method further includes: Collecting the changed requirement information data, and performing multimodal requirement processing using the first language model to update the software requirement document; and Based on the software requirement document and the updated software requirement document, the fourth language model is used to track and manage the requirement changes, and the tracking results are integrated into the development process of the target software.

10. The method according to any one of claims 1 to 8, characterized in that The fusing the information data of the m modes to obtain software requirement data includes: Integrating the information data of the m modes to obtain fused data; and The fused data is knowledge-structured to obtain the software requirement data.

11. A software requirement analysis device, characterized in that: The device comprises: The requirement collection module is used to collect the requirement information of the target software in a multimodal manner and obtain initial data of m modes; wherein m is a positive integer greater than 1; A multimodal analysis module, configured to perform a multimodal demand analysis based on the initial data of the m modes using a first language model to obtain information data of the m modes; An information fusion module, used for fusing the information data of the m modes to obtain software requirement data; and The requirement optimization module is used to perform requirement optimization processing on the software requirement data to obtain a software requirement document corresponding to the target software.

12. An electronic device comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 10.

13. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 10 are implemented.

Images