CN116028098A - Software management system and method for nonstandard enterprises - Google Patents

Software management system and method for nonstandard enterprises Download PDF

Info

Publication number
CN116028098A
CN116028098A CN202310035303.6A CN202310035303A CN116028098A CN 116028098 A CN116028098 A CN 116028098A CN 202310035303 A CN202310035303 A CN 202310035303A CN 116028098 A CN116028098 A CN 116028098A
Authority
CN
China
Prior art keywords
function
personalized
feature vector
vector
software
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
CN202310035303.6A
Other languages
Chinese (zh)
Inventor
张华礼
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Hangzhou Xingzhi Ark Information Technology Co ltd
Original Assignee
Hangzhou Xingzhi Ark Information Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Hangzhou Xingzhi Ark Information Technology Co ltd filed Critical Hangzhou Xingzhi Ark Information Technology Co ltd
Priority to CN202310035303.6A priority Critical patent/CN116028098A/en
Publication of CN116028098A publication Critical patent/CN116028098A/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30—Computing systems specially adapted for manufacturing

Landscapes

  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

A software management system and method for nonstandard enterprises is disclosed, which accurately expresses the personalized function requirements of nonstandard enterprises; and searching for an adapted generic software model based on the personalized functional requirements of the nonstandard enterprise. Specifically, the multi-scale semantic understanding is performed on the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model, in this way, the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model are accurately understood and expressed so as to improve the matching precision and the matching degree, and further, the function of the general software model is ensured to be capable of maximally matching the personalized function requirements of the nonstandard enterprise.

Description

Software management system and method for nonstandard enterprises
Technical Field
The present application relates to the field of enterprise digital management, and more particularly, to a software management system and method for non-standard enterprises.
Background
A non-standard business refers to a business that produces and sells non-standard products (e.g., non-standard automation equipment). Because of the particularity of the product positioning attribute of non-standard enterprises, it is difficult to find universal software suitable for the non-standard enterprises to carry out enterprise digital management on the market. To save costs, it is common practice to: firstly, selecting more adaptive general software; and then, performing function personalized configuration or adjustment on the general software to obtain the software meeting the application requirements.
However, the process of selecting the general-purpose software is a very troublesome process, and the existing method is manual selection, but the number of the general-purpose software is large, and the screening and the testing are all at a great cost. Thus, an optimized software management scheme for non-standard enterprises is desired.
Disclosure of Invention
The present application has been made in order to solve the above technical problems. The embodiment of the application provides a software management system and a method for nonstandard enterprises, which accurately express the personalized function requirements of the nonstandard enterprises; and searching for an adapted generic software model based on the personalized functional requirements of the nonstandard enterprise. Specifically, the multi-scale semantic understanding is performed on the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model, in this way, the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model are accurately understood and expressed so as to improve the matching precision and the matching degree, and further, the function of the general software model is ensured to be capable of maximally matching the personalized function requirements of the nonstandard enterprise.
Accordingly, in accordance with one aspect of the present application, there is provided a software management system for a non-standard enterprise, comprising:
The data acquisition module to be managed is used for acquiring the personalized function requirements of nonstandard enterprises and the function description of the alternative software model;
the first semantic understanding module is used for carrying out word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative software model, and then respectively obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector through a first semantic encoder comprising an embedded layer, wherein the first semantic encoder is constructed based on a converter structure;
the second semantic understanding module is used for carrying out word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative application software model, and then respectively obtaining a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector through a second semantic encoder comprising an embedded layer, wherein the second semantic encoder is constructed based on a two-way long-short-term memory neural network model;
the first fusion module is used for fusing the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector;
The second fusion module is used for fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector;
the matching expression module is used for calculating a difference feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector;
the feature convergence optimization module is used for carrying out convergence optimization on the differential feature vector based on the software function semantic feature vector and the personalized demand semantic feature vector so as to obtain a classification feature vector; and
and the management result generation module is used for passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
In the above software management system for non-standard enterprises, the first semantic understanding module includes: the first word segmentation unit is used for respectively carrying out word segmentation on the personalized function requirements of the nonstandard enterprises and the function description of the alternative software model, processing the consultation text and carrying out word segmentation processing so as to obtain a plurality of personalized function requirement words and function description words of the alternative software model; a first word embedding unit, configured to pass the plurality of personalized function requirement words through an embedding layer to convert each personalized function requirement word in the plurality of personalized function requirement words into a personalized function requirement word embedding vector to obtain a sequence of personalized function requirement word embedding vectors, pass the function descriptors of the plurality of alternative software models through the embedding layer to convert each function descriptor of the alternative software model in the function descriptors of the plurality of alternative software models into a function descriptor embedding vector to obtain a sequence of function descriptor embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each personalized function requirement word and each function descriptor; a first context understanding unit, configured to input the sequence of the personalized function requirement word embedding vector and the sequence of the function description word embedding vector into the first semantic encoder including an embedding layer respectively to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors; and the cascading unit is used for cascading the personalized function demand feature vectors to obtain the first-scale personalized demand semantic feature vector, and cascading the function description feature vectors to obtain the first-scale software function semantic feature vector.
In the software management system for a non-standard enterprise, the first context understanding unit is further configured to: arranging the sequence of the personalized function requirement word embedded vectors into function requirement input vectors, and arranging the sequence of the function description word embedded vectors into function description input vectors; converting the function requirement input vector into a function requirement query vector and a function requirement key vector through a learning embedding matrix respectively, and converting the function description input vector into a function description query vector and a function description key vector through the learning embedding matrix respectively; calculating the product between the function requirement query vector and the transpose vector of the function requirement key vector to obtain a function requirement self-attention correlation matrix, and calculating the product between the function description query vector and the transpose vector of the function description key vector to obtain a function description self-attention correlation matrix; performing standardization processing on the function requirement self-attention correlation matrix to obtain a standardized function requirement self-attention correlation matrix, and performing standardization processing on the function description self-attention correlation matrix to obtain a standardized function description self-attention correlation matrix; activating the standardized function demand self-attention association matrix input Softmax activation function to obtain a function demand self-attention feature matrix, and activating the standardized function description self-attention association matrix input Softmax activation function to obtain a function description self-attention feature matrix; and multiplying the function demand self-attention feature matrix with each personalized function demand word embedded vector in the sequence of personalized function demand word embedded vectors as a value vector to obtain the plurality of personalized function demand feature vectors, and multiplying the function description self-attention feature matrix with each function description word embedded vector in the sequence of function description word embedded vectors as a value vector to obtain the plurality of function description feature vectors.
In the software management system for non-standard enterprises, the second semantic understanding module includes: the second word segmentation unit is used for respectively carrying out word segmentation on the personalized function requirements of the nonstandard enterprise and the function description of the alternative software model, processing the consultation text and carrying out word segmentation processing so as to obtain a plurality of personalized function requirement words and function description words of a plurality of alternative software models; a second word embedding unit, configured to pass the plurality of personalized function requirement words through an embedding layer to convert each personalized function requirement word in the plurality of personalized function requirement words into a personalized function requirement word embedding vector to obtain a sequence of personalized function requirement word embedding vectors, pass the function descriptors of the plurality of alternative software models through the embedding layer to convert each function descriptor of the plurality of alternative software models into a function descriptor embedding vector to obtain a sequence of function descriptor embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each personalized function requirement word and each function descriptor; and a second context understanding unit, configured to input the sequence of the personalized function requirement word embedding vector and the sequence of the function description word embedding vector into the second semantic encoder including the embedding layer, respectively, so as to obtain the second-scale personalized requirement semantic feature vector and the second-scale software function semantic feature vector.
In the software management system for non-standard enterprises, the first fusion module is further configured to: fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector by the following formula; wherein, the formula is:
Figure 489430DEST_PATH_IMAGE002
wherein ,
Figure DEST_PATH_IMAGE003
representing the first scale software functional semantic feature vector,/for>
Figure 278395DEST_PATH_IMAGE004
Representing the second scale software functional semantic feature vector,/for>
Figure DEST_PATH_IMAGE005
Representing the software functional semantic feature vector, < >>
Figure 770556DEST_PATH_IMAGE006
Representing a cascading function; the second fusion module is further configured to: the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are obtained according to the following formula; wherein, the formula is:
Figure 949864DEST_PATH_IMAGE008
wherein ,
Figure DEST_PATH_IMAGE009
representing the first scale personalized demand semantic feature vector,>
Figure 598889DEST_PATH_IMAGE010
representing the second scale personalized demand semantic feature vector,>
Figure DEST_PATH_IMAGE011
representing the personalized demand semantic feature vector, < >>
Figure 711202DEST_PATH_IMAGE006
Representing a cascading function.
In the software management system for non-standard enterprises, the matching expression module is further configured to: calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector in the following formula; wherein, the formula is:
Figure 374264DEST_PATH_IMAGE012
, wherein ,/>
Figure 368765DEST_PATH_IMAGE003
Representing the software functional semantic feature vector, < >>
Figure 916421DEST_PATH_IMAGE004
Representing personalized demand semantic feature vectors, +.>
Figure DEST_PATH_IMAGE013
Representing the differential eigenvector,>
Figure 758606DEST_PATH_IMAGE014
representing by positionAnd (3) difference.
In the software management system for non-standard enterprises, the feature convergence optimization module comprises: the correction feature vector generation unit is configured to perform vector-based hilbert space constraint on the software functional semantic feature vector and the personalized demand semantic feature vector according to the following formula to obtain the correction feature vector, where the formula is:
Figure 264674DEST_PATH_IMAGE016
wherein ,
Figure DEST_PATH_IMAGE017
representing the software functional semantic feature vector, < >>
Figure 277629DEST_PATH_IMAGE018
Representing the personalized demand semantic feature vector,
Figure DEST_PATH_IMAGE019
representing the correction feature vector,>
Figure 628976DEST_PATH_IMAGE020
representing one-dimensional convolution operations, i.e. with the convolution operator +.>
Figure DEST_PATH_IMAGE021
Vector->
Figure 824203DEST_PATH_IMAGE022
One-dimensional convolution is performed, < > and->
Figure DEST_PATH_IMAGE023
Representing the square of the two norms of the vector, i.e. the inner product of the vector itself, < >>
Figure 235593DEST_PATH_IMAGE024
Is a weighting parameter; and a correction unit configured to use the correction feature vector as a weighting vector and the differenceAnd multiplying the classified feature vectors by position points to obtain the classified feature vectors.
In the above software management system for non-standard enterprises, the management result generating module includes: the full-connection coding unit is used for carrying out full-connection coding on the classification characteristic vectors by using a full-connection layer of the classifier so as to obtain coded classification characteristic vectors; and the classification result generating unit is used for inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
According to another aspect of the present application, there is also provided a software management method for a non-standard enterprise, including:
acquiring the personalized function requirements of non-standard enterprises and the function description of alternative application software models;
after word segmentation processing is carried out on the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model, a first semantic encoder containing an embedded layer is respectively used for obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector, wherein the first semantic encoder is constructed based on a converter structure;
after word segmentation is carried out on the personalized function requirements of the nonstandard enterprise and the function description of the alternative application software model, a second semantic encoder containing an embedded layer is respectively used for obtaining a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector, wherein the second semantic encoder is constructed based on a two-way long-short-term memory neural network model;
fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector;
Fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector;
calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector;
based on the software functional semantic feature vector and the personalized demand semantic feature vector, performing convergence optimization on the differential feature vector to obtain a classification feature vector; and
and the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
In the above software management method for non-standard enterprises, after the personalized function requirements of the non-standard enterprises and the function descriptions of the alternative application software model are subjected to word segmentation, the first semantic encoder including an embedded layer is used to obtain a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector, wherein the first semantic encoder is constructed based on a converter structure and comprises: the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing, and the consultation text is subjected to word segmentation processing, so that a plurality of personalized function requirement words and function description words of a plurality of alternative application software models are obtained; converting the personalized function requirement words into personalized function requirement word embedded vectors by an embedding layer to obtain a sequence of personalized function requirement word embedded vectors, converting the function descriptors of the alternative software models into function descriptor embedded vectors by an embedding layer, and obtaining a sequence of function descriptor embedded vectors by using a learnable embedding matrix, wherein the embedding layer carries out embedded coding on the personalized function requirement words and the function descriptors; respectively inputting the sequence of the personalized function requirement word embedded vector and the sequence of the function description word embedded vector into the first semantic encoder containing an embedded layer to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors; and cascading the plurality of personalized function requirement feature vectors to obtain the first-scale personalized requirement semantic feature vector, and cascading the plurality of function description feature vectors to obtain the first-scale software function semantic feature vector.
In the above software management method for non-standard enterprises, the step of inputting the sequence of the personalized function requirement word embedded vectors and the sequence of the function description word embedded vectors into the first semantic encoder including the embedded layer to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors, respectively, includes: arranging the sequence of the personalized function requirement word embedded vectors into function requirement input vectors, and arranging the sequence of the function description word embedded vectors into function description input vectors; converting the function requirement input vector into a function requirement query vector and a function requirement key vector through a learning embedding matrix respectively, and converting the function description input vector into a function description query vector and a function description key vector through the learning embedding matrix respectively; calculating the product between the function requirement query vector and the transpose vector of the function requirement key vector to obtain a function requirement self-attention correlation matrix, and calculating the product between the function description query vector and the transpose vector of the function description key vector to obtain a function description self-attention correlation matrix; performing standardization processing on the function requirement self-attention correlation matrix to obtain a standardized function requirement self-attention correlation matrix, and performing standardization processing on the function description self-attention correlation matrix to obtain a standardized function description self-attention correlation matrix; activating the standardized function demand self-attention association matrix input Softmax activation function to obtain a function demand self-attention feature matrix, and activating the standardized function description self-attention association matrix input Softmax activation function to obtain a function description self-attention feature matrix; and multiplying the function demand self-attention feature matrix with each personalized function demand word embedded vector in the sequence of personalized function demand word embedded vectors as a value vector to obtain the plurality of personalized function demand feature vectors, and multiplying the function description self-attention feature matrix with each function description word embedded vector in the sequence of function description word embedded vectors as a value vector to obtain the plurality of function description feature vectors.
In the above software management method for non-standard enterprises, after the word segmentation processing is performed on the personalized function requirements of the non-standard enterprises and the function descriptions of the alternative application software model, a second semantic encoder including an embedded layer is used to obtain a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector, where the second semantic encoder is configured based on a two-way long-short term memory neural network model, and includes: the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing, and the consultation text is subjected to word segmentation processing, so that a plurality of personalized function requirement words and function description words of a plurality of alternative application software models are obtained; converting the personalized function requirement words into personalized function requirement word embedded vectors by an embedding layer to obtain a sequence of personalized function requirement word embedded vectors, converting the function descriptors of the alternative software models into function descriptor embedded vectors by an embedding layer, and obtaining a sequence of function descriptor embedded vectors by using a learnable embedding matrix, wherein the embedding layer carries out embedded coding on the personalized function requirement words and the function descriptors; and respectively inputting the sequence of the personalized function requirement word embedded vector and the sequence of the function description word embedded vector into the second semantic encoder containing the embedded layer to obtain the second-scale personalized requirement semantic feature vector and the second-scale software function semantic feature vector.
In the above software management method for non-standard enterprises, the fusing the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector includes: fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector by the following formula; wherein, the formula is:
Figure 939107DEST_PATH_IMAGE002
wherein ,
Figure 687620DEST_PATH_IMAGE003
representing the first scale software functional semantic feature vector,/for>
Figure 425769DEST_PATH_IMAGE004
Representing the second scale software functional semantic feature vector,/for>
Figure 476901DEST_PATH_IMAGE005
Representing the software functional semantic feature vector, < >>
Figure 746340DEST_PATH_IMAGE006
Representing a cascading function; the fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector comprises the following steps: the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are obtained according to the following formula; wherein, the formula is:
Figure 970648DEST_PATH_IMAGE008
/>
wherein ,
Figure 766566DEST_PATH_IMAGE009
representing the first scale personalized demand semantic feature vector, >
Figure 316496DEST_PATH_IMAGE010
Representing the second scale personalization needSolving semantic feature vectors, < >>
Figure 56918DEST_PATH_IMAGE011
Representing the personalized demand semantic feature vector, < >>
Figure 84917DEST_PATH_IMAGE006
Representing a cascading function.
In the above software management method for non-standard enterprises, the calculating the differential feature vector between the software function semantic feature vector and the personalized demand semantic feature vector includes: calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector in the following formula; wherein, the formula is:
Figure 469762DEST_PATH_IMAGE012
, wherein ,/>
Figure 190594DEST_PATH_IMAGE003
Representing the software functional semantic feature vector, < >>
Figure 667580DEST_PATH_IMAGE004
Representing the semantic feature vectors of the personalized demand,
Figure 499270DEST_PATH_IMAGE013
representing the differential eigenvector,>
Figure 738621DEST_PATH_IMAGE014
indicating difference by position.
In the above software management method for non-standard enterprises, the performing convergence optimization on the differential feature vector based on the software function semantic feature vector and the personalized demand semantic feature vector to obtain a classification feature vector includes: performing vector-based Hilbert space constraint on the software functional semantic feature vector and the personalized demand semantic feature vector by using the following formula to obtain the correction feature vector, wherein the formula is as follows:
Figure 630354DEST_PATH_IMAGE016
wherein ,
Figure 610948DEST_PATH_IMAGE017
representing the software functional semantic feature vector, < >>
Figure 918433DEST_PATH_IMAGE018
Representing the personalized demand semantic feature vector,
Figure 418816DEST_PATH_IMAGE019
representing the correction feature vector,>
Figure 481449DEST_PATH_IMAGE020
representing one-dimensional convolution operations, i.e. with the convolution operator +.>
Figure 293548DEST_PATH_IMAGE021
Vector->
Figure 732619DEST_PATH_IMAGE022
One-dimensional convolution is performed, < > and->
Figure 71197DEST_PATH_IMAGE023
Representing the square of the two norms of the vector, i.e. the inner product of the vector itself, < >>
Figure 304732DEST_PATH_IMAGE024
Is a weighting parameter; and multiplying the correction feature vector with the difference feature vector by position points to obtain the classification feature vector.
In the above software management method for non-standard enterprises, the step of passing the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether the alternative software model is suitable for the personalized function requirement of the non-standard enterprise, and the method includes: performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
According to still another aspect of the present application, there is provided an electronic apparatus including: a processor; and a memory having stored therein computer program instructions that, when executed by the processor, cause the processor to perform the software management method for a non-standard enterprise as described above.
According to yet another aspect of the present application, there is provided a computer readable medium having stored thereon computer program instructions which, when executed by a processor, cause the processor to perform a software management method for a non-standard enterprise as described above.
Compared with the prior art, the software management system and method for the nonstandard enterprises accurately express the personalized function requirements of the nonstandard enterprises; and searching for an adapted generic software model based on the personalized functional requirements of the nonstandard enterprise. Specifically, the multi-scale semantic understanding is performed on the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model, in this way, the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model are accurately understood and expressed so as to improve the matching precision and the matching degree, and further, the function of the general software model is ensured to be capable of maximally matching the personalized function requirements of the nonstandard enterprise.
Drawings
The foregoing and other objects, features and advantages of the present application will become more apparent from the following more particular description of embodiments of the present application, as illustrated in the accompanying drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.
FIG. 1 is a block diagram of a software management system for a non-standard enterprise according to an embodiment of the present application.
FIG. 2 is a schematic architecture diagram of a software management system for a non-standard enterprise according to an embodiment of the present application.
FIG. 3 is a block diagram of a first semantic understanding module in a software management system for non-standard enterprises according to an embodiment of the present application.
FIG. 4 is a block diagram of a second semantic understanding module in a software management system for non-standard enterprises according to an embodiment of the present application.
FIG. 5 is a flow chart of a method of software management for a non-standard enterprise according to an embodiment of the present application.
Fig. 6 is a block diagram of an electronic device according to an embodiment of the present application.
Detailed Description
Hereinafter, example embodiments according to the present application will be described in detail with reference to the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application and not all of the embodiments of the present application, and it should be understood that the present application is not limited by the example embodiments described herein.
Summary of the application
As described above, the process of selecting the general-purpose software is a very troublesome process, and the existing method is manual selection, but the number of general-purpose software is large, and the screening and testing are costly. Thus, an optimized software management scheme for non-standard enterprises is desired.
Accordingly, in the technical scheme of the application, the process observation of manually screening the general software finds that: the key points of the screening of the universal software are as follows: firstly, accurately expressing the personalized function requirements of nonstandard enterprises; further, based on the personalized function requirements of the nonstandard enterprise, searching for an adapted generic software model, wherein the generic software model function description is capable of maximally adapting to the personalized function requirements of the nonstandard enterprise. This may be achieved by semantic understanding of the search model based on artificial intelligence.
Specifically, the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are obtained. In order to accurately understand and express the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model so as to improve matching accuracy and adaptation, in the technical scheme of the application, multi-scale semantic understanding is carried out on the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model so as to obtain software function semantic feature vectors and personalized requirement semantic feature vectors.
Specifically, firstly, the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing, and then a first semantic encoder containing an embedded layer is used for obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector, wherein the first semantic encoder is constructed based on a converter structure. And simultaneously, performing word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative application software model, and then respectively obtaining a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector through a second semantic encoder comprising an embedded layer, wherein the second semantic encoder is constructed based on a bidirectional long-short-term memory neural network model.
After the first-scale software function semantic feature vector and the second-scale software function semantic feature vector are obtained, the first-scale software function semantic feature vector and the second-scale software function semantic feature vector are fused to obtain a software function semantic feature vector. Meanwhile, after the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are obtained, the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are fused to obtain the personalized demand semantic feature vector. Preferably, in the technical scheme of the application, a cascade mode is adopted to perform multi-scale semantic understanding feature fusion.
Further, a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector is calculated. The difference between the software functional semantic feature vector and the personalized demand semantic feature vector in the semantic feature space is represented by the difference feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector, if the difference between the software functional semantic feature vector and the personalized demand semantic feature vector is smaller, the adaptation degree of the software functional semantic feature vector and the personalized demand semantic feature vector is higher, and if the difference between the software functional semantic feature vector and the personalized demand semantic feature vector is larger, the adaptation degree of the software functional semantic feature vector and the personalized demand semantic feature vector is lower. And finally, the differential feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise. The class probability label to which the differential feature vector belongs is determined by the classifier, wherein the class probability label is an adaptation degree label.
Here, when the differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector is calculated as the classification feature vector, considering that the software functional semantic feature vector and the personalized demand semantic feature vector express the semantic feature of the functional description of the generic software model to be detected and the semantic feature of the personalized description of the currently required function of the automated system integrated device, respectively, there may be a problem that the convergence of the overall feature distribution of the differential feature vector is poor, which may result in poor fitting effect of the classifier. On the other hand, when classifying the classification feature vector, if the correlation between the feature values of the classification feature vector is high, the classification accuracy is lowered.
Thus, semantic feature vectors are generated for the software functions
Figure 604126DEST_PATH_IMAGE017
And said personalized demand semantic feature vector +.>
Figure 315730DEST_PATH_IMAGE018
Hilbert spatial constraint of vector modulus basis is performed to obtain correction feature vectors, expressed as:
Figure 23661DEST_PATH_IMAGE016
Figure 428098DEST_PATH_IMAGE020
representing one-dimensional convolution operations, i.e. with the convolution operator +.>
Figure 214788DEST_PATH_IMAGE021
Vector->
Figure 730083DEST_PATH_IMAGE022
One-dimensional convolution is performed.
That is, the feature distribution of the fused vector is defined in a finite closed domain in the hilbert space based on the modulus of the vector by restricting the fused feature vector with a convolution operator in the hilbert space defining the vector sum modulus and the vector inner product, and orthogonality between each base dimension of the high-dimensional manifold of the feature distribution of the fused vector is improved, so that sparse correlation between feature values is realized while the convergence of the feature distribution as a whole is maintained. Thus, by correcting the feature vector
Figure 43253DEST_PATH_IMAGE019
And performing point multiplication with the differential feature vector, and correcting the classified feature vector.
Based on this, the present application provides a software management system for a non-standard enterprise, comprising: the data acquisition module to be managed is used for acquiring the personalized function requirements of nonstandard enterprises and the function description of the alternative software model; the first semantic understanding module is used for carrying out word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative software model, and then respectively obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector through a first semantic encoder comprising an embedded layer, wherein the first semantic encoder is constructed based on a converter structure; the second semantic understanding module is used for carrying out word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative application software model, and then respectively obtaining a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector through a second semantic encoder comprising an embedded layer, wherein the second semantic encoder is constructed based on a two-way long-short-term memory neural network model; the first fusion module is used for fusing the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector; the second fusion module is used for fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector; the matching expression module is used for calculating a difference feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector; the feature convergence optimization module is used for carrying out convergence optimization on the differential feature vector based on the software function semantic feature vector and the personalized demand semantic feature vector so as to obtain a classification feature vector; and the management result generation module is used for passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
Having described the basic principles of the present application, various non-limiting embodiments of the present application will now be described in detail with reference to the accompanying drawings.
Exemplary System
FIG. 1 is a block diagram of a software management system for a non-standard enterprise according to an embodiment of the present application. As shown in fig. 1, a software management system 100 for a non-standard enterprise according to an embodiment of the present application includes: the data acquisition module to be managed 110 is configured to acquire a personalized function requirement of a nonstandard enterprise and a function description of an alternative software model; the first semantic understanding module 120 is configured to obtain a first-scale software functional semantic feature vector and a first-scale personalized demand semantic feature vector by respectively passing through a first semantic encoder including an embedded layer after performing word segmentation processing on the personalized functional requirements of the nonstandard enterprise and the functional description of the alternative software model, where the first semantic encoder is configured based on a converter structure; the second semantic understanding module 130 is configured to perform word segmentation processing on the personalized function requirements of the nonstandard enterprise and the functional description of the alternative application software model, and then obtain a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector through a second semantic encoder including an embedded layer, where the second semantic encoder is configured based on a bidirectional long-short-term memory neural network model; a first fusion module 140, configured to fuse the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector; a second fusion module 150, configured to fuse the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector; a matching expression module 160, configured to calculate a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector; the feature convergence optimization module 170 is configured to perform convergence optimization on the differential feature vector based on the software functional semantic feature vector and the personalized demand semantic feature vector to obtain a classification feature vector; and a management result generating module 180, configured to pass the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether the alternative software model is adapted to the personalized function requirement of the nonstandard enterprise.
FIG. 2 is a schematic architecture diagram of a software management system for a non-standard enterprise according to an embodiment of the present application. As shown in fig. 2, first, the personalized function requirement of the nonstandard enterprise and the function description of the alternative software model are obtained; then, the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing and then pass through a first semantic encoder comprising an embedded layer respectively to obtain a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector, wherein the first semantic encoder is constructed based on a converter structure; then, the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing and then respectively pass through a second semantic encoder comprising an embedded layer to obtain second-scale software function semantic feature vectors and second-scale personalized requirement semantic feature vectors, wherein the second semantic encoder is constructed based on a bidirectional long-short-term memory neural network model; then, fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector, and simultaneously fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector; then, calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector; then, based on the software functional semantic feature vector and the personalized demand semantic feature vector, performing convergence optimization on the differential feature vector to obtain a classification feature vector; and finally, the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
As described above, the process of selecting the general-purpose software is a very troublesome process, and the existing method is manual selection, but the number of general-purpose software is large, and the screening and testing are costly. Thus, an optimized software management scheme for non-standard enterprises is desired.
Accordingly, in the technical scheme of the application, the process observation of manually screening the general software finds that: the key points of the screening of the universal software are as follows: firstly, accurately expressing the personalized function requirements of nonstandard enterprises; further, based on the personalized function requirements of the nonstandard enterprise, searching for an adapted generic software model, wherein the generic software model function description is capable of maximally adapting to the personalized function requirements of the nonstandard enterprise. This may be achieved by semantic understanding of the search model based on artificial intelligence.
In the software management system 100 for non-standard enterprises, the data collection module 110 to be managed is configured to obtain the personalized function requirements of the non-standard enterprises and the function description of the alternative software model. That is, in the technical solution of the present application, the personalized function requirement of the nonstandard enterprise and the functional description of the alternative software model are used as the input of the convolutional neural network model.
In the software management system 100 for non-standard enterprises, the first semantic understanding module 120 is configured to perform word segmentation processing on the personalized function requirements of the non-standard enterprises and the function descriptions of the alternative application software model, and then obtain a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector through a first semantic encoder including an embedded layer, where the first semantic encoder is configured based on a converter structure.
Specifically, in one example of the present application, the first semantic encoder includes an embedded layer and a Transformer (Transformer) -based Bert model. The embedded layer is used for respectively carrying out vectorization conversion on a plurality of personalized function requirement words obtained by word segmentation processing on personalized function requirements of non-standard enterprises and function descriptions of the alternative software models and function description words of the alternative software models, namely converting one personalized function requirement word into one personalized function requirement word embedded vector and converting the function description words of the alternative software models into one function description word embedded vector. In particular embodiments, the embedded layer vector transformer may be constructed based on knowledge-graph. In addition, the plurality of personalized function requirement words and the plurality of function description words of the alternative software model can be converted into structural data which is more convenient for a computer to operate through vectorization.
The function of the converter-based Bert model is to perform global-based context semantic coding on the above-mentioned sequence of embedded vectors (i.e., global context semantic coding on each embedded vector in the sequence of embedded vectors based on the sequence of embedded vectors) to obtain a plurality of personalized functional requirement feature vectors and a plurality of functional description feature vectors corresponding to the sequence of embedded vectors.
FIG. 3 is a block diagram of a first semantic understanding module in a software management system for non-standard enterprises according to an embodiment of the present application. As shown in fig. 3, the first semantic understanding module 120 includes: a first word segmentation unit 121, configured to perform word segmentation processing on the consultation text by performing word segmentation processing on the personalized function requirement of the nonstandard enterprise and the function description of the alternative software model, so as to obtain a plurality of personalized function requirement words and function description words of the alternative software model; a first word embedding unit 122, configured to pass the plurality of personalized function requirement words through an embedding layer to convert each personalized function requirement word in the plurality of personalized function requirement words into a personalized function requirement word embedding vector to obtain a sequence of personalized function requirement word embedding vectors, pass the function descriptors of the plurality of candidate software models through the embedding layer to convert each function descriptor of the candidate software models in the function descriptors of the plurality of candidate software models into a function descriptor embedding vector to obtain a sequence of function descriptor embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each personalized function requirement word and each function descriptor; a first context understanding unit 123, configured to input the sequence of the personalized function requirement word embedding vectors and the sequence of the function description word embedding vectors into the first semantic encoder including an embedding layer respectively to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors; and a concatenation unit 124, configured to concatenate the plurality of personalized function requirement feature vectors to obtain the first-scale personalized requirement semantic feature vector, and concatenate the plurality of function description feature vectors to obtain the first-scale software function semantic feature vector.
More specifically, in the embodiment of the present application, in the first context understanding unit 123, first, the sequence of the personalized function requirement word embedded vectors is arranged as a function requirement input vector, and the sequence of the function description word embedded vectors is arranged as a function description input vector; then, the function requirement input vector is respectively converted into a function requirement query vector and a function requirement key vector through a learning embedding matrix, and the function description input vector is respectively converted into a function description query vector and a function description key vector through the learning embedding matrix; then, calculating the product between the function requirement query vector and the transpose vector of the function requirement key vector to obtain a function requirement self-attention correlation matrix, and calculating the product between the function description query vector and the transpose vector of the function description key vector to obtain a function description self-attention correlation matrix; then, carrying out standardization processing on the function requirement self-attention correlation matrix to obtain a standardized function requirement self-attention correlation matrix, and carrying out standardization processing on the function description self-attention correlation matrix to obtain a standardized function description self-attention correlation matrix; then, the standardized function demand self-attention association matrix is input into a Softmax activation function to be activated to obtain a function demand self-attention feature matrix, and the standardized function description self-attention association matrix is input into the Softmax activation function to be activated to obtain a function description self-attention feature matrix; and finally, respectively multiplying the function requirement self-attention feature matrix with each personalized function requirement word embedded vector in the sequence of personalized function requirement word embedded vectors as a value vector to obtain a plurality of personalized function requirement feature vectors, and respectively multiplying the function description self-attention feature matrix with each function description word embedded vector in the sequence of function description word embedded vectors as a value vector to obtain a plurality of function description feature vectors.
In the software management system 100 for non-standard enterprises, the second semantic understanding module 130 is configured to perform word segmentation processing on the personalized function requirements of the non-standard enterprises and the function descriptions of the alternative application software model, and then obtain second-scale software function semantic feature vectors and second-scale personalized requirement semantic feature vectors through a second semantic encoder including an embedded layer, where the second semantic encoder is configured based on a two-way long-short-term memory neural network model.
It should be understood that the two-way Long Short-Term Memory neural network model (LSTM) is a time-cycled neural network, which enables the weight of the neural network to be self-updated by adding an input gate, an output gate and a forgetting gate, and the weight scale at different moments can be dynamically changed under the condition of fixed parameters of the network model, so that the problems of gradient disappearance or gradient expansion can be avoided. The bidirectional long-short-term memory neural network model is formed by combining a forward LSTM and a backward LSTM, the forward LSTM can learn the personalized function requirements of the nonstandard enterprise and the semantic association feature distribution among words in the function description of the alternative use software model based on the current word, and the backward LSTM can learn the personalized function requirements of the nonstandard enterprise and the semantic association feature distribution among words in the function description of the alternative use software model based on the current word, so that the second-scale software function semantic feature vector and the second-scale personalized requirement semantic feature vector obtained through the bidirectional long-short-term memory neural network model learn the information of the global context of the middle distance.
FIG. 4 is a block diagram of a second semantic understanding module in a software management system for non-standard enterprises according to an embodiment of the present application. As shown in fig. 4, the second semantic understanding module 130 includes: the second word segmentation unit 131 is configured to perform word segmentation processing on the consultation text by performing word segmentation processing on the personalized function requirement of the nonstandard enterprise and the function description of the alternative software model, so as to obtain a plurality of personalized function requirement words and function description words of the alternative software model; a second word embedding unit 132, configured to pass the plurality of personalized function requirement words through an embedding layer to convert each personalized function requirement word in the plurality of personalized function requirement words into a personalized function requirement word embedding vector to obtain a sequence of personalized function requirement word embedding vectors, pass the function descriptors of the plurality of candidate software models through the embedding layer to convert the function descriptors of each candidate software model in the function descriptors of the plurality of candidate software models into function descriptor embedding vectors to obtain a sequence of function descriptor embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each personalized function requirement word and each function descriptor; and a second context understanding unit 133, configured to input the sequence of the personalized function requirement word embedding vector and the sequence of the function description word embedding vector into the second semantic encoder including an embedding layer, respectively, to obtain the second-scale personalized requirement semantic feature vector and the second-scale software function semantic feature vector.
In the software management system 100 for non-standard enterprises, the first fusion module 140 is configured to fuse the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector. That is, after the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector are obtained, the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector are fused to obtain a software functional semantic feature vector. Preferably, in the technical scheme of the application, a cascade mode is adopted to perform multi-scale semantic understanding feature fusion.
Specifically, in the embodiment of the application, the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector are fused by the following formula to obtain a software functional semantic feature vector; wherein, the formula is:
Figure 618590DEST_PATH_IMAGE002
wherein ,
Figure 892577DEST_PATH_IMAGE003
representing the first scale software functional semantic feature vector,/for>
Figure 945984DEST_PATH_IMAGE004
Representing the second scale software functional semantic feature vector,/for>
Figure 129971DEST_PATH_IMAGE005
Representing the software functional semantic feature vector, < > >
Figure 610631DEST_PATH_IMAGE006
Representing a cascading function.
In the software management system 100 for non-standard enterprises, the second fusion module 150 is configured to fuse the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector. Likewise, after the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are obtained, the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are fused to obtain the personalized demand semantic feature vector.
Specifically, in the embodiment of the application, the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are expressed as the following formula to obtain the personalized demand semantic feature vector; wherein, the formula is:
Figure 371914DEST_PATH_IMAGE008
wherein ,
Figure 229012DEST_PATH_IMAGE009
representing the first scale personalized demand semantic feature vector,>
Figure 188877DEST_PATH_IMAGE010
representing the second scale personalized demand semantic feature vector,>
Figure 433914DEST_PATH_IMAGE011
representing the personalized demand semantic feature vector, < >>
Figure 682493DEST_PATH_IMAGE006
Representing a cascading function.
In the software management system 100 for non-standard enterprises, the matching expression module 160 is configured to calculate a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector. The difference between the software functional semantic feature vector and the personalized demand semantic feature vector in the semantic feature space is represented by the difference feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector, if the difference between the software functional semantic feature vector and the personalized demand semantic feature vector is smaller, the adaptation degree of the software functional semantic feature vector and the personalized demand semantic feature vector is higher, and if the difference between the software functional semantic feature vector and the personalized demand semantic feature vector is larger, the adaptation degree of the software functional semantic feature vector and the personalized demand semantic feature vector is lower.
Specifically, in the embodiment of the present application, the matching expression module 160 is further configured to: calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector in the following formula; wherein, the formula is:
Figure 343281DEST_PATH_IMAGE012
, wherein ,/>
Figure 469238DEST_PATH_IMAGE003
Representing the software functional semantic feature vector, < >>
Figure 822859DEST_PATH_IMAGE004
Representing personalized demand semantic feature vectors, +.>
Figure 89892DEST_PATH_IMAGE013
Representing the differential eigenvector,>
Figure 226475DEST_PATH_IMAGE014
indicating difference by position.
In the software management system 100 for non-standard enterprises, the feature convergence optimization module 170 is configured to perform convergence optimization on the differential feature vector to obtain a classification feature vector based on the software function semantic feature vector and the personalized demand semantic feature vector. Here, when the differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector is calculated as the classification feature vector, considering that the software functional semantic feature vector and the personalized demand semantic feature vector express the semantic feature of the functional description of the generic software model to be detected and the semantic feature of the personalized description of the currently required function of the automated system integrated device, respectively, there may be a problem that the convergence of the overall feature distribution of the differential feature vector is poor, which may result in poor fitting effect of the classifier. On the other hand, when classifying the classification feature vector, if the correlation between the feature values of the classification feature vector is high, the classification accuracy is lowered.
Thus, semantic feature vectors are generated for the software functions
Figure 160933DEST_PATH_IMAGE017
And said personalized demand semantic feature vector +.>
Figure 747772DEST_PATH_IMAGE018
Hilbert spatial constraint of vector modulus basis is performed to obtain correction feature vectors, expressed as:
Figure 502102DEST_PATH_IMAGE016
wherein ,
Figure 707955DEST_PATH_IMAGE017
representing the software functional semantic feature vector, < >>
Figure 496920DEST_PATH_IMAGE018
Representing the personalized demand semantic feature vector,
Figure 739813DEST_PATH_IMAGE019
representing the correction feature vector,>
Figure 247018DEST_PATH_IMAGE020
representing one-dimensional convolution operations, i.e. with the convolution operator +.>
Figure 990983DEST_PATH_IMAGE021
Vector->
Figure 634454DEST_PATH_IMAGE022
One-dimensional convolution is performed, < > and->
Figure 563096DEST_PATH_IMAGE023
Representing the square of the two norms of the vector, i.e. the inner product of the vector itself, < >>
Figure 292017DEST_PATH_IMAGE024
Is a weighting parameter.
That is, the feature distribution of the fused vector is defined in a finite closed domain in the hilbert space based on the modulus of the vector by restricting the fused feature vector with a convolution operator in the hilbert space defining the vector sum modulus and the vector inner product, and orthogonality between each base dimension of the high-dimensional manifold of the feature distribution of the fused vector is improved, so that sparse correlation between feature values is realized while the convergence of the feature distribution as a whole is maintained.
Then, the correction feature vector is used as a weighting vector to be multiplied by the difference feature vector according to position points by a correction unit to obtain the classification feature vector. Thus, by correcting the feature vector
Figure 839673DEST_PATH_IMAGE019
And performing point multiplication with the differential feature vector, and correcting the classified feature vector.
In the software management system 100 for non-standard enterprises, the management result generating module 180 is configured to pass the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether the alternative software model is adapted to the personalized function requirement of the non-standard enterprise. The class probability label to which the differential feature vector belongs is determined by the classifier, wherein the class probability label is an adaptation degree label.
Specifically, in the embodiment of the present application, the management result generating module 180 includes: the full-connection coding unit is used for carrying out full-connection coding on the classification characteristic vectors by using a full-connection layer of the classifier so as to obtain coded classification characteristic vectors; and the classification result generating unit is used for inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
In summary, a software management system 100 for non-standard enterprises in accordance with embodiments of the present application is illustrated that accurately expresses the personalized functional needs of non-standard enterprises; and searching for an adapted generic software model based on the personalized functional requirements of the nonstandard enterprise. Specifically, the multi-scale semantic understanding is performed on the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model, in this way, the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model are accurately understood and expressed so as to improve the matching precision and the matching degree, and further, the function of the general software model is ensured to be capable of maximally matching the personalized function requirements of the nonstandard enterprise.
As described above, the software management system 100 for non-standard enterprises according to the embodiment of the present application may be implemented in various terminal devices, such as a server or the like for software management of non-standard enterprises. In one example, the software management system 100 for non-standard enterprises according to embodiments of the present application may be integrated into a terminal device as one software module and/or hardware module. For example, the software management system 100 for a nonstandard enterprise may be a software module in the operating system of the terminal device, or may be an application developed for the terminal device; of course, the software management system 100 for non-standard enterprises may also be one of a plurality of hardware modules of the terminal device.
Alternatively, in another example, the software management system 100 for non-standard enterprises and the terminal device may be separate devices, and the software management system 100 for non-standard enterprises may be connected to the terminal device through a wired and/or wireless network and transmit interactive information in a agreed data format.
Exemplary method
FIG. 5 is a flow chart of a method of software management for a non-standard enterprise according to an embodiment of the present application. As shown in fig. 5, a software management method for a non-standard enterprise according to an embodiment of the present application includes: s110, acquiring the personalized function requirement of a nonstandard enterprise and the function description of an alternative software model; s120, performing word segmentation processing on the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model, and then respectively obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector through a first semantic encoder comprising an embedded layer, wherein the first semantic encoder is constructed based on a converter structure; s130, performing word segmentation processing on the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model, and then respectively obtaining second-scale software function semantic feature vectors and second-scale personalized requirement semantic feature vectors through a second semantic encoder comprising an embedded layer, wherein the second semantic encoder is constructed based on a bidirectional long-short-term memory neural network model; s140, fusing the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector; s150, fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector; s160, calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector; s170, performing convergence optimization on the differential feature vector based on the software functional semantic feature vector and the personalized demand semantic feature vector to obtain a classification feature vector; and S180, the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
In one example, in the above software management method for a non-standard enterprise, after the word segmentation processing is performed on the personalized function requirement of the non-standard enterprise and the function description of the alternative application software model, a first semantic encoder including an embedded layer is used to obtain a first scale software function semantic feature vector and a first scale personalized requirement semantic feature vector, where the first semantic encoder is configured based on a converter structure, and includes:
the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing, and the consultation text is subjected to word segmentation processing, so that a plurality of personalized function requirement words and function description words of a plurality of alternative application software models are obtained;
converting the personalized function requirement words into personalized function requirement word embedded vectors by an embedding layer to obtain a sequence of personalized function requirement word embedded vectors, converting the function descriptors of the alternative software models into function descriptor embedded vectors by an embedding layer, and obtaining a sequence of function descriptor embedded vectors by using a learnable embedding matrix, wherein the embedding layer carries out embedded coding on the personalized function requirement words and the function descriptors;
Respectively inputting the sequence of the personalized function requirement word embedded vector and the sequence of the function description word embedded vector into the first semantic encoder containing an embedded layer to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors; and
cascading the plurality of personalized function requirement feature vectors to obtain the first-scale personalized requirement semantic feature vector, and cascading the plurality of function description feature vectors to obtain the first-scale software function semantic feature vector.
In one example, in the software management method for a nonstandard enterprise, the inputting the sequence of the personalized function requirement word embedding vectors and the sequence of the function description word embedding vectors into the first semantic encoder including an embedding layer to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors includes: arranging the sequence of the personalized function requirement word embedded vectors into function requirement input vectors, and arranging the sequence of the function description word embedded vectors into function description input vectors; converting the function requirement input vector into a function requirement query vector and a function requirement key vector through a learning embedding matrix respectively, and converting the function description input vector into a function description query vector and a function description key vector through the learning embedding matrix respectively; calculating the product between the function requirement query vector and the transpose vector of the function requirement key vector to obtain a function requirement self-attention correlation matrix, and calculating the product between the function description query vector and the transpose vector of the function description key vector to obtain a function description self-attention correlation matrix; performing standardization processing on the function requirement self-attention correlation matrix to obtain a standardized function requirement self-attention correlation matrix, and performing standardization processing on the function description self-attention correlation matrix to obtain a standardized function description self-attention correlation matrix; activating the standardized function demand self-attention association matrix input Softmax activation function to obtain a function demand self-attention feature matrix, and activating the standardized function description self-attention association matrix input Softmax activation function to obtain a function description self-attention feature matrix; and multiplying the function demand self-attention feature matrix with each personalized function demand word embedded vector in the sequence of personalized function demand word embedded vectors as a value vector to obtain the plurality of personalized function demand feature vectors, and multiplying the function description self-attention feature matrix with each function description word embedded vector in the sequence of function description word embedded vectors as a value vector to obtain the plurality of function description feature vectors.
In one example, in the software management method for a nonstandard enterprise, after the word segmentation processing is performed on the personalized function requirement of the nonstandard enterprise and the function description of the alternative application software model, a second semantic encoder including an embedded layer is used to obtain a second scale software function semantic feature vector and a second scale personalized requirement semantic feature vector, where the second semantic encoder is configured based on a two-way long-short term memory neural network model, and the method includes: the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing, and the consultation text is subjected to word segmentation processing, so that a plurality of personalized function requirement words and function description words of a plurality of alternative application software models are obtained; converting the personalized function requirement words into personalized function requirement word embedded vectors by an embedding layer to obtain a sequence of personalized function requirement word embedded vectors, converting the function descriptors of the alternative software models into function descriptor embedded vectors by an embedding layer, and obtaining a sequence of function descriptor embedded vectors by using a learnable embedding matrix, wherein the embedding layer carries out embedded coding on the personalized function requirement words and the function descriptors; and respectively inputting the sequence of the personalized function requirement word embedded vector and the sequence of the function description word embedded vector into the second semantic encoder containing the embedded layer to obtain the second-scale personalized requirement semantic feature vector and the second-scale software function semantic feature vector.
In one example, in the software management method for non-standard enterprises, the fusing the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector includes: fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector by the following formula; wherein, the formula is:
Figure 603230DEST_PATH_IMAGE002
wherein ,
Figure 686461DEST_PATH_IMAGE003
representing the first scale software functional semantic feature vector,/for>
Figure 168258DEST_PATH_IMAGE004
Representing the second scale software functional semantic feature vector,/for>
Figure 254026DEST_PATH_IMAGE005
Representing the software functional semantic feature vector, < >>
Figure 606510DEST_PATH_IMAGE006
Representing a cascading function; the fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector comprises the following steps: the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are obtained according to the following formula; wherein, the formula is:
Figure 549058DEST_PATH_IMAGE008
wherein ,
Figure 314889DEST_PATH_IMAGE009
representing the first scale personalized demand semantic feature vector, >
Figure 266664DEST_PATH_IMAGE010
Representing the second scale personalized demand semantic feature vector,>
Figure 676917DEST_PATH_IMAGE011
representing the personalized demand semantic feature vector, < >>
Figure 55946DEST_PATH_IMAGE006
Representing a cascading function.
In one example, in the software management method for non-standard enterprises, the calculating the differential feature vector between the software function semantic feature vector and the personalized demand semantic feature vector includes: calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector in the following formula; wherein, the formula is:
Figure 59805DEST_PATH_IMAGE012
, wherein ,/>
Figure 549692DEST_PATH_IMAGE003
Representing the software functional semantic feature vector, < >>
Figure 80031DEST_PATH_IMAGE004
Representing personalized demand semantic feature vectors, +.>
Figure 629961DEST_PATH_IMAGE013
Representing the differential eigenvector,>
Figure 104804DEST_PATH_IMAGE014
indicating difference by position.
In one example, in the software management method for non-standard enterprises, the performing convergence optimization on the differential feature vector to obtain a classification feature vector based on the software function semantic feature vector and the personalized demand semantic feature vector includes: performing vector-based Hilbert space constraint on the software functional semantic feature vector and the personalized demand semantic feature vector by using the following formula to obtain the correction feature vector, wherein the formula is as follows:
Figure 398382DEST_PATH_IMAGE016
wherein ,
Figure 783227DEST_PATH_IMAGE017
representing the software functional semantic feature vector, < >>
Figure 504059DEST_PATH_IMAGE018
Representing the personalized demand semantic feature vector,
Figure 981045DEST_PATH_IMAGE019
representing the correction feature vector,>
Figure 812735DEST_PATH_IMAGE020
representing one-dimensional convolution operations, i.e. with the convolution operator +.>
Figure 52087DEST_PATH_IMAGE021
Vector->
Figure 943819DEST_PATH_IMAGE022
One-dimensional convolution is performed, < > and->
Figure 393255DEST_PATH_IMAGE023
Representing the square of the two norms of the vector, i.e. the inner product of the vector itself, < >>
Figure 28636DEST_PATH_IMAGE024
Is a weighting parameter; and multiplying the correction feature vector with the difference feature vector by position points to obtain the classification feature vector.
In one example, in the software management method for non-standard enterprises, the step of passing the classification feature vector through a classifier to obtain a classification result, where the classification result is used to indicate whether the alternative software model is adapted to the personalized function requirement of the non-standard enterprise, and the method includes: performing full-connection coding on the classification feature vectors by using a full-connection layer of the classifier to obtain coded classification feature vectors; and inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
In summary, the software management method for non-standard enterprises of the embodiments of the present application is illustrated, which accurately expresses the personalized function requirements of the non-standard enterprises; and searching for an adapted generic software model based on the personalized functional requirements of the nonstandard enterprise. Specifically, the multi-scale semantic understanding is performed on the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model, in this way, the personalized function requirements of the nonstandard enterprise and the function description of the alternative use software model are accurately understood and expressed so as to improve the matching precision and the matching degree, and further, the function of the general software model is ensured to be capable of maximally matching the personalized function requirements of the nonstandard enterprise.
Exemplary electronic device
Next, an electronic device according to an embodiment of the present application is described with reference to fig. 6. Fig. 6 is a block diagram of an electronic device according to an embodiment of the present application. As shown in fig. 6, the electronic device 10 includes one or more processors 11 and a memory 12.
The processor 11 may be a Central Processing Unit (CPU) or other form of processing unit having data processing and/or instruction execution capabilities, and may control other components in the electronic device 10 to perform desired functions.
Memory 12 may include one or more computer program products that may include various forms of computer-readable storage media, such as volatile memory and/or non-volatile memory. The volatile memory may include, for example, random Access Memory (RAM) and/or cache memory (cache), and the like. The non-volatile memory may include, for example, read Only Memory (ROM), hard disk, flash memory, and the like. One or more computer program instructions may be stored on the computer readable storage medium that can be executed by the processor 11 to implement the functions in the software management method for non-standard enterprises and/or other desired functions of the various embodiments of the present application as described above. Various content such as personalized functionality requirements of non-standard enterprises and functional descriptions of alternative utility software models may also be stored in the computer readable storage medium.
In one example, the electronic device 10 may further include: an input device 13 and an output device 14, which are interconnected by a bus system and/or other forms of connection mechanisms (not shown).
The input means 13 may comprise, for example, a keyboard, a mouse, etc.
The output device 14 may output various information including the classification result and the like to the outside. The output means 14 may include, for example, a display, speakers, a printer, and a communication network and remote output devices connected thereto, etc.
Of course, only some of the components of the electronic device 10 that are relevant to the present application are shown in fig. 6 for simplicity, components such as buses, input/output interfaces, etc. are omitted. In addition, the electronic device 10 may include any other suitable components depending on the particular application.
Exemplary computer program product and computer readable storage Medium
In addition to the methods and apparatus described above, embodiments of the present application may also be a computer program product comprising computer program instructions which, when executed by a processor, cause the processor to perform steps in the functions in the software management methods for non-standard enterprises according to the various embodiments of the present application described in the "exemplary methods" section of the present specification.
The computer program product may write program code for performing the operations of embodiments of the present application in any combination of one or more programming languages, including an object oriented programming language such as Java, C++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device, partly on a remote computing device, or entirely on the remote computing device or server.
Furthermore, embodiments of the present application may also be a computer-readable storage medium, having stored thereon computer program instructions that, when executed by a processor, cause the processor to perform steps in the functions in the software management method for non-standard enterprises according to the various embodiments of the present application described in the "exemplary methods" section of the present specification.
The computer readable storage medium may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. The readable storage medium may include, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium would include the following: an electrical connection having one or more wires, a portable disk, a hard disk, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
The basic principles of the present application have been described above in connection with specific embodiments, however, it should be noted that the advantages, benefits, effects, etc. mentioned in the present application are merely examples and not limiting, and these advantages, benefits, effects, etc. are not to be considered as necessarily possessed by the various embodiments of the present application. Furthermore, the specific details disclosed herein are for purposes of illustration and understanding only, and are not intended to be limiting, as the application is not intended to be limited to the details disclosed herein as such.
The block diagrams of the devices, apparatuses, devices, systems referred to in this application are only illustrative examples and are not intended to require or imply that the connections, arrangements, configurations must be made in the manner shown in the block diagrams. As will be appreciated by one of skill in the art, the devices, apparatuses, devices, systems may be connected, arranged, configured in any manner. Words such as "including," "comprising," "having," and the like are words of openness and mean "including but not limited to," and are used interchangeably therewith. The terms "or" and "as used herein refer to and are used interchangeably with the term" and/or "unless the context clearly indicates otherwise. The term "such as" as used herein refers to, and is used interchangeably with, the phrase "such as, but not limited to.
It is also noted that in the apparatus, devices and methods of the present application, the components or steps may be disassembled and/or assembled. Such decomposition and/or recombination should be considered as equivalent to the present application.
The previous description of the disclosed aspects is provided to enable any person skilled in the art to make or use the present application. Various modifications to these aspects will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other aspects without departing from the scope of the application. Thus, the present application is not intended to be limited to the aspects shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.
The foregoing description has been presented for purposes of illustration and description. Furthermore, this description is not intended to limit the embodiments of the application to the form disclosed herein. Although a number of example aspects and embodiments have been discussed above, a person of ordinary skill in the art will recognize certain variations, modifications, alterations, additions, and subcombinations thereof.

Claims (10)

1. A software management system for a non-standard enterprise, comprising:
the data acquisition module to be managed is used for acquiring the personalized function requirements of nonstandard enterprises and the function description of the alternative software model;
The first semantic understanding module is used for carrying out word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative software model, and then respectively obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector through a first semantic encoder comprising an embedded layer, wherein the first semantic encoder is constructed based on a converter structure;
the second semantic understanding module is used for carrying out word segmentation processing on the personalized function requirements of the nonstandard enterprise and the function description of the alternative application software model, and then respectively obtaining a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector through a second semantic encoder comprising an embedded layer, wherein the second semantic encoder is constructed based on a two-way long-short-term memory neural network model;
the first fusion module is used for fusing the first-scale software function semantic feature vector and the second-scale software function semantic feature vector to obtain a software function semantic feature vector;
the second fusion module is used for fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector;
The matching expression module is used for calculating a difference feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector;
the feature convergence optimization module is used for carrying out convergence optimization on the differential feature vector based on the software function semantic feature vector and the personalized demand semantic feature vector so as to obtain a classification feature vector; and
and the management result generation module is used for passing the classification feature vector through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
2. The software management system for a non-standard enterprise of claim 1, wherein the first semantic understanding module comprises:
the first word segmentation unit is used for respectively carrying out word segmentation on the personalized function requirements of the nonstandard enterprises and the function description of the alternative software model, processing the consultation text and carrying out word segmentation processing so as to obtain a plurality of personalized function requirement words and function description words of the alternative software model;
a first word embedding unit, configured to pass the plurality of personalized function requirement words through an embedding layer to convert each personalized function requirement word in the plurality of personalized function requirement words into a personalized function requirement word embedding vector to obtain a sequence of personalized function requirement word embedding vectors, pass the function descriptors of the plurality of alternative software models through the embedding layer to convert each function descriptor of the alternative software model in the function descriptors of the plurality of alternative software models into a function descriptor embedding vector to obtain a sequence of function descriptor embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each personalized function requirement word and each function descriptor;
A first context understanding unit, configured to input the sequence of the personalized function requirement word embedding vector and the sequence of the function description word embedding vector into the first semantic encoder including an embedding layer respectively to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors; and
the cascade unit is used for cascading the personalized function demand feature vectors to obtain the first-scale personalized demand semantic feature vector, and cascading the function description feature vectors to obtain the first-scale software function semantic feature vector.
3. The software management system for a non-standard enterprise of claim 2, wherein the first context understanding unit is further configured to:
arranging the sequence of the personalized function requirement word embedded vectors into function requirement input vectors, and arranging the sequence of the function description word embedded vectors into function description input vectors;
converting the function requirement input vector into a function requirement query vector and a function requirement key vector through a learning embedding matrix respectively, and converting the function description input vector into a function description query vector and a function description key vector through the learning embedding matrix respectively;
Calculating the product between the function requirement query vector and the transpose vector of the function requirement key vector to obtain a function requirement self-attention correlation matrix, and calculating the product between the function description query vector and the transpose vector of the function description key vector to obtain a function description self-attention correlation matrix;
performing standardization processing on the function requirement self-attention correlation matrix to obtain a standardized function requirement self-attention correlation matrix, and performing standardization processing on the function description self-attention correlation matrix to obtain a standardized function description self-attention correlation matrix;
activating the standardized function demand self-attention association matrix input Softmax activation function to obtain a function demand self-attention feature matrix, and activating the standardized function description self-attention association matrix input Softmax activation function to obtain a function description self-attention feature matrix; and
and multiplying the function demand self-attention feature matrix with each personalized function demand word embedded vector in the sequence of personalized function demand word embedded vectors as a value vector to obtain a plurality of personalized function demand feature vectors, and multiplying the function description self-attention feature matrix with each function description word embedded vector in the sequence of function description word embedded vectors as a value vector to obtain a plurality of function description feature vectors.
4. The software management system for a non-standard enterprise of claim 3, wherein the second semantic understanding module comprises:
the second word segmentation unit is used for respectively carrying out word segmentation on the personalized function requirements of the nonstandard enterprise and the function description of the alternative software model, processing the consultation text and carrying out word segmentation processing so as to obtain a plurality of personalized function requirement words and function description words of a plurality of alternative software models;
a second word embedding unit, configured to pass the plurality of personalized function requirement words through an embedding layer to convert each personalized function requirement word in the plurality of personalized function requirement words into a personalized function requirement word embedding vector to obtain a sequence of personalized function requirement word embedding vectors, pass the function descriptors of the plurality of alternative software models through the embedding layer to convert each function descriptor of the plurality of alternative software models into a function descriptor embedding vector to obtain a sequence of function descriptor embedding vectors, where the embedding layer uses a learnable embedding matrix to perform embedded encoding on each personalized function requirement word and each function descriptor; and
And the second context understanding unit is used for respectively inputting the sequence of the personalized function requirement word embedded vector and the sequence of the function description word embedded vector into the second semantic encoder containing the embedded layer to obtain the second-scale personalized requirement semantic feature vector and the second-scale software function semantic feature vector.
5. The software management system for a non-standard enterprise of claim 4,
the first fusion module is further configured to: fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector by the following formula;
wherein, the formula is:
Figure DEST_PATH_IMAGE001
wherein ,
Figure DEST_PATH_IMAGE002
representing the first scale software functional semantic feature vector,/for>
Figure 643200DEST_PATH_IMAGE003
Representing the second scale software functional semantic feature vector,/for>
Figure DEST_PATH_IMAGE004
Representing the software functional semantic feature vector, < >>
Figure 29182DEST_PATH_IMAGE005
Representing a cascading function;
the second fusion module is further configured to: the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector are obtained according to the following formula;
Wherein, the formula is:
Figure DEST_PATH_IMAGE006
wherein ,
Figure DEST_PATH_IMAGE007
representing the first scale personalized demand semantic feature vector,>
Figure DEST_PATH_IMAGE008
representing the second scale personalized demand semantic feature vector,>
Figure 667974DEST_PATH_IMAGE009
representing the personalized demand semantic feature vector, < >>
Figure 266445DEST_PATH_IMAGE005
Representing a cascading function.
6. The software management system for a non-standard enterprise of claim 5, wherein the matching expression module is further configured to:
calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector in the following formula;
wherein,the formula is:
Figure DEST_PATH_IMAGE010
, wherein ,/>
Figure 923561DEST_PATH_IMAGE002
Representing the software functional semantic feature vector, < >>
Figure 11602DEST_PATH_IMAGE003
Representing personalized demand semantic feature vectors, +.>
Figure 481898DEST_PATH_IMAGE011
Representing the differential eigenvector,>
Figure DEST_PATH_IMAGE012
indicating difference by position.
7. The software management system for a non-standard enterprise of claim 6, wherein the feature convergence optimization module comprises:
the correction feature vector generation unit is configured to perform vector-based hilbert space constraint on the software functional semantic feature vector and the personalized demand semantic feature vector according to the following formula to obtain the correction feature vector, where the formula is:
Figure 274273DEST_PATH_IMAGE013
wherein ,
Figure DEST_PATH_IMAGE014
representing the software functional semantic feature vector, < >>
Figure DEST_PATH_IMAGE015
Representing the personalized demand semantic feature vector, < >>
Figure DEST_PATH_IMAGE016
Representing the correction feature vector,>
Figure 677573DEST_PATH_IMAGE017
representing one-dimensional convolution operations, i.e. with the convolution operator +.>
Figure DEST_PATH_IMAGE018
Vector pair
Figure 15144DEST_PATH_IMAGE019
One-dimensional convolution is performed, < > and->
Figure DEST_PATH_IMAGE020
Representing the square of the two norms of the vector, i.e. the inner product of the vector itself, < >>
Figure 972736DEST_PATH_IMAGE021
Is a weighting parameter; and
and the correction unit is used for multiplying the correction feature vector serving as a weighting vector and the difference feature vector by position points to obtain the classification feature vector.
8. The software management system for a non-standard enterprise of claim 7, wherein the management result generation module comprises:
the full-connection coding unit is used for carrying out full-connection coding on the classification characteristic vectors by using a full-connection layer of the classifier so as to obtain coded classification characteristic vectors; and
and the classification result generating unit is used for inputting the coding classification feature vector into a Softmax classification function of the classifier to obtain the classification result.
9. A method of software management for a non-standard enterprise, comprising:
acquiring the personalized function requirements of non-standard enterprises and the function description of alternative application software models;
After word segmentation processing is carried out on the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model, a first semantic encoder containing an embedded layer is respectively used for obtaining a first-scale software function semantic feature vector and a first-scale personalized requirement semantic feature vector, wherein the first semantic encoder is constructed based on a converter structure;
after word segmentation is carried out on the personalized function requirements of the nonstandard enterprise and the function description of the alternative application software model, a second semantic encoder containing an embedded layer is respectively used for obtaining a second-scale software function semantic feature vector and a second-scale personalized requirement semantic feature vector, wherein the second semantic encoder is constructed based on a two-way long-short-term memory neural network model;
fusing the first-scale software functional semantic feature vector and the second-scale software functional semantic feature vector to obtain a software functional semantic feature vector;
fusing the first-scale personalized demand semantic feature vector and the second-scale personalized demand semantic feature vector to obtain a personalized demand semantic feature vector;
calculating a differential feature vector between the software functional semantic feature vector and the personalized demand semantic feature vector;
Based on the software functional semantic feature vector and the personalized demand semantic feature vector, performing convergence optimization on the differential feature vector to obtain a classification feature vector; and
and the classification feature vector passes through a classifier to obtain a classification result, wherein the classification result is used for indicating whether the alternative software model is suitable for the personalized function requirement of a nonstandard enterprise.
10. The method for software management of a non-standard enterprise according to claim 9, wherein the word segmentation processing is performed on the personalized function requirement of the non-standard enterprise and the function description of the alternative software model, and then the first semantic encoder is respectively configured based on a converter structure, and the method further comprises the steps of:
the personalized function requirements of the nonstandard enterprises and the function description of the alternative application software model are subjected to word segmentation processing, and the consultation text is subjected to word segmentation processing, so that a plurality of personalized function requirement words and function description words of a plurality of alternative application software models are obtained;
Converting the personalized function requirement words into personalized function requirement word embedded vectors by an embedding layer to obtain a sequence of personalized function requirement word embedded vectors, converting the function descriptors of the alternative software models into function descriptor embedded vectors by an embedding layer, and obtaining a sequence of function descriptor embedded vectors by using a learnable embedding matrix, wherein the embedding layer carries out embedded coding on the personalized function requirement words and the function descriptors;
respectively inputting the sequence of the personalized function requirement word embedded vector and the sequence of the function description word embedded vector into the first semantic encoder containing an embedded layer to obtain a plurality of personalized function requirement feature vectors and a plurality of function description feature vectors; and
cascading the plurality of personalized function requirement feature vectors to obtain the first-scale personalized requirement semantic feature vector, and cascading the plurality of function description feature vectors to obtain the first-scale software function semantic feature vector.
CN202310035303.6A 2023-01-10 2023-01-10 Software management system and method for nonstandard enterprises Withdrawn CN116028098A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310035303.6A CN116028098A (en) 2023-01-10 2023-01-10 Software management system and method for nonstandard enterprises

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310035303.6A CN116028098A (en) 2023-01-10 2023-01-10 Software management system and method for nonstandard enterprises

Publications (1)

Publication Number Publication Date
CN116028098A true CN116028098A (en) 2023-04-28

Family

ID=86079167

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310035303.6A Withdrawn CN116028098A (en) 2023-01-10 2023-01-10 Software management system and method for nonstandard enterprises

Country Status (1)

Country Link
CN (1) CN116028098A (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116610803A (en) * 2023-07-19 2023-08-18 北京每日信动科技有限公司 Industrial chain excellent enterprise information management method and system based on big data
CN116821195A (en) * 2023-05-31 2023-09-29 郑州富铭科技股份有限公司 Method for automatically generating application based on database
CN117172256A (en) * 2023-07-31 2023-12-05 台州道致科技股份有限公司 Laboratory management method and system based on modularized setting

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN116821195A (en) * 2023-05-31 2023-09-29 郑州富铭科技股份有限公司 Method for automatically generating application based on database
CN116610803A (en) * 2023-07-19 2023-08-18 北京每日信动科技有限公司 Industrial chain excellent enterprise information management method and system based on big data
CN116610803B (en) * 2023-07-19 2023-12-08 橙客时代(北京)网络科技有限公司 Industrial chain excellent enterprise information management method and system based on big data
CN117172256A (en) * 2023-07-31 2023-12-05 台州道致科技股份有限公司 Laboratory management method and system based on modularized setting
CN117172256B (en) * 2023-07-31 2024-03-12 台州道致科技股份有限公司 Laboratory management method and system based on modularized setting

Similar Documents

Publication Publication Date Title
CN116028098A (en) Software management system and method for nonstandard enterprises
CN115203380B (en) Text processing system and method based on multi-mode data fusion
CN114647198B (en) Intelligent home control method and system based on Internet of things and electronic equipment
CN115796173A (en) Data processing method and system for supervision submission requirements
WO2021196954A1 (en) Serialized data processing method and device, and text processing method and device
CN116308754B (en) Bank credit risk early warning system and method thereof
CN116245513B (en) Automatic operation and maintenance system and method based on rule base
CN111144124A (en) Training method of machine learning model, intention recognition method, related device and equipment
CN116257406A (en) Gateway data management method and system for smart city
CN115222566A (en) Learning method and system for international finance and finance metrology teaching
CN112800774A (en) Entity relationship extraction method, device, medium and equipment based on attention mechanism
CN110990596A (en) Multi-mode hash retrieval method and system based on self-adaptive quantization
CN116307624A (en) Resource scheduling method and system of ERP system
CN116776872A (en) Medical data structured archiving system
CN116151845A (en) Product full life cycle management system and method based on industrial Internet of things technology
CN116992304A (en) Policy matching analysis system and method based on artificial intelligence
Ye et al. Automated conversion of engineering rules: Towards flexible manufacturing collaboration
CN116340703B (en) Soil sample parameter measurement method and system based on data
CN116069953B (en) MDATA knowledge representation method based on knowledge graph superposition space-time attribute
CN116151604A (en) Office system flow analysis system and method under web environment
CN114693107A (en) Enterprise financial risk prediction method and system and electronic equipment
KR20230060320A (en) Knowledge graph integration method and machine learning device using the same
CN113536797A (en) Slice document key information single model extraction method and system
CN116595976B (en) Scientific research innovation platform control method and system
CN116842131A (en) Library book management method and system

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
WW01 Invention patent application withdrawn after publication

Application publication date: 20230428

WW01 Invention patent application withdrawn after publication