Product Stage Certification Method and Device Based on Mobile Collaborative Electronic Signature

Abstract

The present invention discloses a method and device for product stage authentication based on mobile collaborative electronic signatures, which relates to the technical field of security authentication. The method includes: generating a first feature vector representing the core content of product authentication application materials; obtaining a second feature vector representing the index text of the target product, where the index text is generated based on the first feature vectors of the core content of each product authentication application material completed for the target product and multiple R & D stage labels corresponding to the target product, and the R & D stage labels corresponding to the target product are determined according to the category of the target product; inputting the first feature vector and the second feature vector into a trained neural network model to obtain an output result representing whether to add the core content of the product authentication application materials to the index text; if the output result is yes, generating a stage authentication electronic seal for the target product. The present invention ensures the security, timeliness, legality and convenience of the submission of authentication application materials.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a method and device for product stage certification based on mobile collaborative electronic signatures. Background Art

[0002] With the popularization of the mobile Internet and the development of the informatization of product certification services, various online product certification application materials, as the basis and core of product certification services, are increasingly used in various product certification activities.

[0003] In the process of implementing the present invention, the inventor found that due to the particularity of some products, such as long R & D cycles or belonging to a product series, stage product certification must be carried out. In this certification process, it is necessary to deeply and fully understand the current certification application materials, and also necessary to retrieve the previous certification application materials to determine whether the product meets the stage progress and thus give certification. Since it is necessary to carefully analyze the current materials and review the past materials, a large amount of time is consumed, and the entire certification process is significantly untimely. In addition, the entire process has extremely high requirements for the professionalism of the certifiers to determine whether the stage progress of the product meets the certification standards. Such high requirements result in relatively few eligible certifiers, further affecting the certification efficiency. In addition, the security of traditional electronic signature methods is not good. Although it ensures the legality and non-repudiation of the application materials to a certain extent, it does not guarantee the legal identity of the senders of the certification application materials. Summary of the Invention

[0004] In order to solve the above technical problems or at least partially solve the above technical problems, embodiments of the present invention provide a method and device for product stage certification based on mobile collaborative electronic signatures, which can intelligently determine whether a product meets the stage progress and thus give certification according to the certification application materials, improving the certification efficiency and ensuring the security, timeliness, legality and convenience of the submission of the certification application materials.

[0005] Embodiments of the present invention provide a method for product stage certification based on mobile collaborative electronic signatures, including:

[0006] Receive the product certification application materials of the target product, and generate a first feature vector representing the core content of the product certification application materials; obtain a second feature vector representing the index text of the target product; wherein, the index text is generated based on the first feature vectors of the core contents of the product certification application materials that have been completed for the target product and multiple R & D stage labels corresponding to the target product; the R & D stage labels corresponding to the target product are determined according to the category of the target product; input the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to represent whether to add the core content of the product certification application materials to the index text; if the output result is yes, generate a mobile collaborative electronic signature for the phased certification of the target product.

[0007] An embodiment of the present invention further provides a product phased certification device based on a mobile collaborative electronic signature, including:

[0008] A first generation module, configured to receive the product certification application materials of the target product and generate a first feature vector representing the core content of the product certification application materials; an acquisition module, configured to obtain a second feature vector representing the index text of the target product; wherein, the index text is generated based on the first feature vectors of the core contents of the product certification application materials that have been completed for the target product and multiple R & D stage labels corresponding to the target product; the R & D stage labels corresponding to the target product are determined according to the category of the target product; an output module, configured to input the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to represent whether to add the core content of the product certification application materials to the index text; a second generation module, configured to generate a mobile collaborative electronic signature for the phased certification of the target product if the output result is yes.

[0009] An embodiment of the present invention further provides an electronic device, and the electronic device includes:

[0010] One or more processors; a storage device, configured to store one or more programs; when the one or more programs are executed by the one or more processors, the one or more processors implement the product phased certification method based on a mobile collaborative electronic signature as described above.

[0011] An embodiment of the present invention further provides a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the product phased certification method based on a mobile collaborative electronic signature as described above is implemented.

[0012] An embodiment of the present invention also provides a computer program product, which includes a computer program or instruction. When the computer program or instruction is executed by a processor, it implements the product stage authentication method based on mobile collaborative electronic signatures as described above.

[0013] The product stage authentication method and device based on mobile collaborative electronic signatures provided by the embodiments of the present invention have at least the following advantages compared with the prior art: The present invention can intelligently determine whether a product meets the stage progress according to the authentication application materials, thereby giving authentication, improving the authentication efficiency, and ensuring the security, timeliness, legality, and convenience of the submission of authentication application materials. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Combined with the drawings and referring to the following specific embodiments, the above and other features, advantages, and aspects of the embodiments of the present invention will become more obvious. Throughout the drawings, the same or similar reference numerals represent the same or similar elements. It should be understood that the drawings are schematic, and the original components and elements are not necessarily drawn to scale.

[0015] Figure 1 It is a flowchart of a product stage authentication method based on mobile collaborative electronic signatures provided by an embodiment of the present application;

[0016] Figure 2 It is a flowchart of a method for generating a mobile collaborative electronic signature for product stage authentication provided by an embodiment of the present application;

[0017] Figure 3 It is a schematic structural diagram of an electronic device provided by an embodiment of the present application;

[0018] Figure 4 It is a schematic structural diagram of a product stage authentication device based on mobile collaborative electronic signatures provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] The embodiments of the present invention will be described in more detail below with reference to the drawings. Although some embodiments of the present invention are shown in the drawings, it should be understood that the present invention can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. On the contrary, these embodiments are provided to more thoroughly and completely understand the present invention. It should be understood that the drawings and embodiments of the present invention are only for exemplary purposes and are not used to limit the protection scope of the present invention.

[0020] It should be understood that the steps recorded in the method embodiments of the present invention can be executed in different orders and / or in parallel. In addition, the method embodiments may include additional steps and / or omit the steps shown. The scope of the present invention is not limited in this regard.

[0021] As used herein, the term "including" and its variations are open-ended, that is, "including but not limited to". The term "based on" means "at least in part based on". The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments". The relevant definitions of other terms will be given in the following description.

[0022] It should be noted that the concepts such as "first", "second", etc. mentioned in the present invention are only used to distinguish different devices, modules or units, and are not used to limit the order or interdependence of the functions performed by these devices, modules or units.

[0023] It should be noted that the modifications of "one" and "multiple" mentioned in the present invention are illustrative rather than restrictive. Those skilled in the art should understand that unless otherwise clearly specified in the context, it should be understood as "one or more".

[0024] Reference Figure 1 As shown, an embodiment of the present invention provides a product stage certification method based on mobile collaborative electronic signatures. This method is executed on the server of the product certification application data submission system in the form of a computer program, and includes the following steps:

[0025] Step S101, receiving the product certification application data of the target product, and generating a first feature vector representing the core content of the product certification application data.

[0026] In this step, the product certification application data may include text or images.

[0027] For the text, first, a text segmentation algorithm in natural language processing technology is adopted to segment the text of the application materials according to paragraphs, sentences or chapters. For each segmented unit, it is labeled according to the key information category it belongs to, such as labeled as "basic product information", "technical parameters", "experimental results", etc. Secondly, for the key information of each segmented unit, word frequency statistics and keyword extraction are carried out. Specifically, the term frequency-inverse document frequency (TF-IDF) algorithm can be used to calculate the importance weight of each word. TF-IDF can highlight the words that frequently appear in the product certification application materials of the target product but relatively rarely appear in the product certification application materials of similar products. These words often have high distinctiveness and representativeness. Based on the TF-IDF results, the top several words with higher weights are selected as the keywords of this key information. For example, for the basic product information part, the key materials related to the product composition, the key numbers in the model, etc. may be extracted as keywords. Finally, an initial vector is created for each key information. The dimension of the initial vector is equal to the number of keywords extracted from this key information. Each dimension corresponds to a keyword, and its value can be initialized according to the importance weight of the keyword (such as the TF-IDF value). The vectors of all key information are concatenated in a pre-determined order (such as in the order of basic product information, technical parameters, production process flow, experimental results, etc.) to form a long vector, and this long vector is the first feature vector representing the core content of the product certification application materials.

[0028] As an alternative implementation manner of the embodiment of the present invention, generating the first feature vector representing the core content of the product certification application materials includes: performing a segmentation process on the product certification application materials to obtain at least one text segment; for each text segment, obtaining the previous text segment corresponding to this text segment, and the summary vector of the previous text segment, where the previous text segment is all the text segments before this text segment in the product certification application materials, concatenating the summary vector of the previous text segment and the feature vector of this text segment to obtain the summary vector of this text segment; generating the first feature vector representing the core content of the product certification application materials according to the summary vectors corresponding to each text segment.

[0029] In this solution, for each text segment, both the feature vector of the text segment and the summary vector of the text segment are generated. Among them, the summary vector of the text segment is obtained by understanding and summarizing this text segment and its previous text. That is to say, each word in the text segment is understood and defined through the context, so as to be able to more accurately extract the key information of the text segment.

[0030] Specifically, use the sentence segmentation algorithm in natural language processing technology to segment the product certification application materials, ensuring that each text segment is relatively independent and complete semantically. Preliminary segmentation can be performed according to rules such as punctuation marks and paragraph separators. After determining the previous text segment for each text segment, sequentially input the previous text segment into a pre-trained generative model (such as GPT series, BERT, etc.) to obtain the vector representation of a specific layer output by the generative model as the initial encoding result. Then, use a text summarization algorithm (such as TextRank, Transformer-based summarization model, etc.) to further process the initial encoding result, extract key information, and generate a summary text. Then, input the summary text into the generative model again to obtain its corresponding vector representation as the summary vector of the previous text segment.

[0031] Furthermore, concatenate the summary vector of the previous text segment and the feature vector of this text segment. Before concatenation, the two vectors can be normalized to ensure their similarity in the numerical range. For example, use the min-max normalization method to map the values of the vector to the interval [0, 1]. For the concatenated vector, further fusion processing can be performed through a fully connected neural network to obtain the summary vector of this text segment.

[0032] Finally, based on the summary vectors corresponding to each text segment, clustering algorithms (such as K-Means clustering, hierarchical clustering, etc.) can be used to group these vectors. Each cluster represents a theme or feature type. By analyzing the distribution of the cluster center and the vectors within the cluster, key feature information can be extracted. Perform a weighted sum of the center vectors of each cluster to obtain the first feature vector representing the core content of the product certification application materials. The weights can be determined according to factors such as the size and importance of the cluster. For example, a higher weight can be given to a cluster containing more text segments.

[0033] For pictures, first perform preprocessing operations such as denoising and enhancement on the pictures to improve the image quality. And adjust the image size so that all pictures have a unified size for subsequent processing. Secondly, based on traditional image processing methods, such as color histograms and texture features, feature extraction of the pictures can be carried out, and the extracted features are combined into a first feature vector to represent the image. Among them, the color histogram can describe the color distribution of the picture, and the texture feature can reflect the texture structure of the picture. In addition, deep learning models, such as convolutional neural networks (CNNs), can also be used for feature extraction of the pictures, and the output of the last fully connected layer is used as the first feature vector of the picture.

[0034] Furthermore, if the product certification application materials can include text and pictures, the first feature vectors of the text and the pictures need to be concatenated in a certain order to obtain a combined first feature vector.

[0035] In the process of implementing the present invention, the inventors found that pictures generally appear to assist in understanding the text. Therefore, as an optional implementation manner of the embodiments of the present invention, the product certification application materials include pictures, and the method further includes: for each picture, inputting the picture into a generative model to obtain a description text of the picture; in the product certification application materials, determining a target text segment related to the picture, and integrating the description text with the target text segment.

[0036] Specifically, the picture types of the pictures in the product certification application materials include data types representing experimental effects, product form types representing the basic information of the target product, structure types representing the production process flow, etc. Different picture types correspond to different prompt messages, and the prompt messages are used to represent the key information dimensions that need to be described for this type of picture. The dimensions of the key information match the picture types (for example, the dimension of the data type is data analysis, the dimension of the product form type is form description, and the dimension of the structure type is structure and connection description, etc.). Inputting the picture and the prompt message corresponding to the picture type of the picture into the generative model to obtain a description text, which can be a sentence or a short paragraph, describing the main content and features of the key information dimensions that the picture needs to describe. The generative model can select a pre-trained vision-language model, such as CLIP (Contrastive Language–Image Pretraining) or DALLE 2, etc. These models are trained on a large amount of image and text pair data and can well understand the image content and generate corresponding description texts in combination with the prompt messages. It is also possible to select a multi-modal model based on the Transformer architecture to realize the generation of description texts from images through the joint learning of image features and text features.

[0037] Further, based on the position of the picture in the product certification application materials, several text segments close to the picture position are determined. Methods such as the bag-of-words model and TF-IDF are used to extract the keywords and feature vectors of each text segment, and the correlation between the description text and each text segment is calculated by calculating indicators such as the cosine similarity and Jaccard similarity between the text vectors. The text segments with a correlation exceeding a preset threshold are determined as the target text segments related to the picture.

[0038] A simple splicing method can be adopted to directly connect the picture description text and the target text segment. For example, the picture description text can be inserted at the beginning or end of the target text segment and distinguished by specific delimiters. More complex fusion methods can also be used, such as fusion based on the attention mechanism. By training an attention model, the model can automatically learn the importance weights between the picture description text and the target text segment, and then perform weighted fusion.

[0039] Optionally, this solution can also calculate the similarity between pictures first. For pictures with high similarity, picture groups representing the same key information can be identified (such as pictures of different magnification ratios of product forms, data analysis of different dimensions under the same index, etc.), and the description texts of the picture groups can be aggregated into a description text group to uniformly determine and integrate the target text segment.

[0040] As an optional implementation of the embodiment of the present invention, if the picture includes text, inputting the picture into the generative model to obtain the description text of the picture includes: recognizing the picture to obtain a text recognition result; determining the image quality of the picture based on the image feature vector of the picture, where the image quality is used to characterize the clarity of the picture; inputting the text recognition result, the image feature vector, and the image quality score into a pre-trained generative model to obtain the description text of the picture; the image quality score is used to determine the participation weight of the text recognition result when the generative model understands the picture.

[0041] Among them, the higher the image quality score, the higher the clarity of the text in the picture, and the lower the image quality score, the lower the clarity of the text in the picture. The biggest feature of the generative model is its content understanding ability and generation ability, that is, it can predict the text in the picture according to the image feature vector. That is, on the basis of obtaining the text recognition result by recognizing the picture according to the optical character recognition method, the generative model can correct the text recognition result by combining the understanding of the picture to ensure the accuracy of the text recognition result, and then obtain the description text of the picture according to the prompt information, the image feature vector, and the corrected text recognition result.

[0042] In this solution, a mapping relationship table between image clarity and image quality score can be pre-set. After obtaining the image quality score corresponding to the image clarity, the model participation weight of the generative model in the process of text content recognition of pictures is adjusted according to the image quality score, so that the degree of relying on the generative model for preliminary correction of text recognition results in text content recognition varies with different image quality scores. For example, if the clarity of the text in the picture is high, the text in the image can be recognized only by the optical character recognition method, that is, the generative model does not need to correct the text recognition result, and the description text of the picture can be directly obtained according to the prompt information, the image feature vector, and the text recognition result.

[0043] Before implementing this solution, a batch of pictures with known clarity are collected as training data, their image feature vectors are extracted, and the corresponding clarity levels (such as high clarity, medium clarity, low clarity, etc.) are labeled. Machine learning algorithms, such as support vector machine (SVM), random forest, neural network, etc., are used to train the clarity evaluation model. The image feature vectors of the pictures are used as the input, and the clarity level is used as the output for supervised learning. Optionally, the image feature vectors of the pictures can be obtained by deep learning methods (using pre-trained convolutional neural networks such as VGG, ResNet, etc.).

[0044] Step S102, obtain a second feature vector representing the index text of the target product; wherein, the index text is generated based on the first feature vector of the core content of each product certification application material completed for the target product and multiple R & D stage labels corresponding to the target product; the R & D stage labels corresponding to the target product are determined according to the category of the target product.

[0045] The R & D of the target product usually goes through multiple different stages. For example, the R & D stages of chemical products include the formula research stage, the reaction condition research stage, the equipment process research stage, etc.; the R & D stages of medical device products include the material R & D stage, the equipment structure R & D stage, the safety research stage, the effectiveness test stage, etc. The R & D information of different categories of products is pre-statistically analyzed, and the R & D stages included in each category of products are statistically clustered to determine the R & D stage labels of each category of products.

[0046] Here, the index text of the target product includes the first R & D stage labels of multiple completed certification R & D stages of the target product and the second R & D stage labels of the uncompleted certification R & D stages; the index file also includes the first feature vector of the core content of the product certification application material corresponding to each first R & D stage label.

[0047] In this step, the first feature vector is converted into a text description and then concatenated with the R & D stage label. For example, the key feature words in the vector can be extracted to form a descriptive text, and then combined with the R & D stage label and its corresponding R & D stage label. A specific format or template is used to structurally represent the first feature vector and the R & D stage label for subsequent processing. For example, the JSON format can be used to store the feature vector and the label as different fields respectively. Accordingly, the concatenated text of each R & D stage is obtained, and the concatenated text of each R & D stage is respectively input into a deep learning model, such as a language model (BERT, GPT, etc.) with a Transformer architecture, to encode the indexed text and obtain a second feature vector.

[0048] As an alternative implementation of the embodiment of the present invention, the indexed text is generated based on the first feature vector of the core content of each product certification application material completed for the target product and multiple R & D stage labels corresponding to the target product, specifically including:

[0049] Step S1021, setting multiple nodes of the indexed text according to each of the R & D stage labels.

[0050] Here, the R & D stage label includes multiple levels. For example, the first-level R & D stage label includes the formula research stage, the reaction condition research stage, and the equipment process research stage. The second-level R & D stage label is a sub-label of the first-level R & D stage label. For example, the formula research stage label includes sub-labels such as new ingredient research label and new ratio research label. Specifically, the level of the label can be determined according to the granularity of the phased certification. For example, if the granularity of the certification is relatively coarse, nodes of the indexed text are set only for the first-level R & D stage label, that is, one node is set for each R & D stage label. For example, if the granularity of the certification is relatively fine, two layers of nodes can be set for the first-level R & D stage label and the second-level R & D stage label.

[0051] Step S1022, determining the explanatory text of each node and generating a third feature vector of the explanatory text, where the explanatory text is used to characterize the key features of the target product required for each node.

[0052] In this step, the method for generating the third feature vector is as described above and will not be elaborated herein. The explanatory text is used to characterize the key feature points of each dimension that the target product needs to achieve at the corresponding R & D stage. For example, the dimensions corresponding to the formulation research stage include the innovation, feasibility, and safety of the formulation. The key feature points of the innovation dimension include new chemical structures, reaction pathways, or performance characteristics. The key feature points of the feasibility dimension include the feasibility of chemical reactions, the stability of products, and the preliminary tests of expected performance. The key feature points of the safety dimension include the toxicity of raw materials, potential hazards during the reaction process, and specific values of the safety of the final product.

[0053] As some alternative embodiments of the present invention, determining the explanatory text for each of the nodes includes:

[0054] Determining search keywords and at least one genre search intention based on the R & D stage labels corresponding to each node; searching for a plurality of multimedia information corresponding to the genre search intention according to the search keywords; performing multimodal understanding on the plurality of multimedia information to determine the explanatory text for each of the nodes.

[0055] Here, it is necessary to search for product information of the same type as the target product corresponding to each R & D stage to summarize and determine the required feature points to achieve the requirements of this R & D stage. Specifically, the R & D stage labels corresponding to each node, as well as synonyms and near-synonyms of the R & D stage labels, are used as search keywords, and it is determined according to the R & D stage labels which type or types of product information of genres are needed to summarize and determine the key feature points of this R & D stage. For example, in the safety dimension of the formulation research stage, product information of the picture genre is needed to determine the reaction process state. Therefore, after determining the search keywords and at least one genre search intention, a plurality of multimedia information is searched. Multimodal fusion algorithms, such as fusion based on attention mechanisms, fusion based on feature splicing, etc., can be used to fuse information of different modalities to generate the explanatory text.

[0056] Step S1023: Calculate the similarity between the third feature vector of each explanatory text and the first feature vector of the core content of each of the completed product certification application materials to obtain the target product certification application materials under each of the nodes.

[0057] Specifically, for each explanatory text, the bag-of-words model, TF-IDF, or more advanced text vector representation methods in natural language processing, such as Word2Vec, GloVe, or deep learning-based language models (such as BERT, GPT, etc.), are used to extract feature vectors as the third feature vectors. And through methods such as cosine similarity, Euclidean distance, Pearson correlation coefficient, etc., the similarity between the third feature vector of each explanatory text and the first feature vector of each completed product certification application document is calculated respectively. For each node, sorting is performed according to the calculated similarity values. The product certification application document with the highest similarity can be selected as the product certification application document for the target product under this node.

[0058] Step S1024, for each of the said nodes, generate the index text according to the core content of the target product certification application document under this node.

[0059] Specifically, the first feature vector of the core content of the target product certification document has been generated according to the above steps. Therefore, based on the R & D stage label of each node, the third feature vector of the explanatory text, and the first feature vector of the core content of the target product certification document, the index text is generated.

[0060] Step S103, input the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to represent whether to add the core content of the product certification application document to the index text.

[0061] In this step, a neural network model suitable for processing vector inputs can be selected according to the type of the target product, such as a multi-layer perceptron (MLP), a model combining a convolutional neural network (CNN) and a fully connected layer, or a recurrent neural network (RNN) variant such as a long short-term memory network (LSTM), a gated recurrent unit (GRU), etc. If the input first feature vector and second feature vector contain sequence information or complex relationships, the RNN and its variants are more suitable as the neural network model. Specifically, the samples in the training set include the first feature vector of the core content of the product certification application document, the second feature vector of the index text, and the probability that the product certification application document corresponds to the target node in the nodes included in the index text. This probability is obtained by calculating the similarity between the first feature vector of the core content of the product certification application document and the third feature vector of the explanatory text corresponding to each node of the index text.

[0062] The input layer of the neural network model receives the concatenation of the first feature vector and the second feature vector, and the two vectors can be joined together through a simple concatenate operation. The intermediate hidden layer can use activation functions such as ReLU and tanh to enhance the non-linear expression ability of the model. The output layer uses the sigmoid function to map the output value to the interval [0, 1], representing the probability of adding the core content of the product certification application materials to the index text. Whether to add the core content of the application materials to the index text is determined according to the output result. That is, if the output result is close to 1, it means it should be added; if it is close to 0, it means it should not be added.

[0063] As an alternative implementation of the embodiment of the present invention, the output result is used to characterize whether to add the core content of the product certification application materials to the index text, including:

[0064] The output result is used to characterize that the core content of the product certification application materials corresponds to the first target node of the index text, and there is no corresponding target product certification application materials under the first target node; or,

[0065] The output result is used to characterize that the core content of the product certification application materials corresponds to the second target node of the index text, and there is corresponding target product certification application materials under the second target node, and the core content of the product certification application materials is used to update the second target node.

[0066] If the output result is yes, execute step S104.

[0067] Here, the output result being close to 1 includes two possibilities. The first possibility is that the similarity between the first feature vector of the core content of the product certification application materials and the third feature vector of the explanatory text of the first target node is greater than the preset threshold, and there is no corresponding target product certification application materials under the first target node; the second possibility is that the similarity between the first feature vector of the core content of the product certification application materials and the third feature vector of the explanatory text of the second target node is greater than the preset threshold, but there is corresponding target product certification application materials under the second target node. By calculating the similarity between the first feature vector of the core content of the target product certification application materials and the first feature vector of the core content of the product certification application materials of the target product, it is determined that the similarity between the two is less than the preset threshold, indicating that the product certification application materials of the target product under the second target node have undergone a substantial update (this situation may occur when the second target node needs to undergo multiple phased certifications). The above two situations both characterize that the core content of the product certification application materials needs to be added to the index text.

[0068] Step S104: Generate a mobile collaborative electronic signature for the phased certification of the target product.

[0069] Specifically, it can be determined that the target product meets the phased progress through the certification application materials of the target product, and thus certification is given. The certification method is to generate a mobile collaborative electronic signature for the phased certification of the target product.

[0070] The commonly used method for submitting traditional product certification application materials is as follows: The operator of the certification applicant first determines the content of the final certification application materials through methods such as electronic data interchange or real-time communication tools, and then uses physical means or traditional electronic signature methods to complete the signature. Finally, the materials are submitted to the certification agency by mailing paper materials or through electronic data interchange. This method is rather cumbersome to operate. Since there is no identity real-name authentication for the operator of the certification applicant, enterprise authorization, and the product certification application is not encrypted during the confirmation and submission process, it is easily tampered with, posing a threat to the business secrets and transaction security of the certification applicant. Moreover, the management of the submitted certification application materials lacks security control. Although the traditional electronic signature method can ensure the legality and non-repudiation of the certification application materials, the operator of the certification applicant has a problem of inefficiency in the process of operating the submission of the certification application materials.

[0071] To solve the technical problems of the legal identity authentication of the operator of the product certification applicant in the product certification application material submission system and the inefficiency of the operator of the certification applicant in the process of submitting the product certification application materials, as some optional implementation manners of the embodiments of the present invention, as Figure 2 shown, a method for generating a mobile collaborative electronic signature for the phased certification of the target product is disclosed. This method is realized through the cooperation of a mobile terminal, a CA center, and a product certification application material submission system server, and includes the following steps:

[0072] Step S201: The operator of the certification applicant registers an account on the product certification application material submission system server and uses a mobile phone number certified by the operator as the account username.

[0073] Step S202: After the operator of the certification applicant logs in to the product certification application material submission system on the mobile intelligent terminal, personal identity authentication and enterprise real-name authentication are respectively performed.

[0074] Step S203: After the operator of the certification applicant completes the identity authentication of the corresponding role, a pair of public and private key factors are respectively generated on the mobile intelligent terminal and the CA center. After the mobile intelligent terminal sends the generated public key factor P1 to the CA center, P1 and the public key P2 owned by the CA center are synthesized into a public key P through operations such as dot multiplication, and a mobile terminal signature digital certificate and an encryption digital certificate are generated using the synthesized public key P and the above account.

[0075] Step S204: After the CA center verifies the identity of the operator of the authentication applicant, it issues the mobile signature digital certificate, encrypted digital certificate, and encrypted private key of the operator of the authentication applicant to the operator of the authentication applicant and records them on the mobile intelligent terminal side.

[0076] Step S205: After the operator of the authentication applicant passes the identity authentication and successfully obtains the mobile signature digital certificate and encrypted digital certificate, the product certification application data submission system creates an electronic seal according to the main information of the digital certificate user.

[0077] Step S206: After the operator of the authentication applicant passes the identity authentication and logs in to the product certification application data submission system using the mobile intelligent terminal, the operator implants the electronic seal, mobile signature digital certificate, encrypted digital certificate, and encrypted private key of the authentication applicant on the submitted product certification application data, and calculates and generates a mobile collaborative electronic signature seal.

[0078] Furthermore, the above method can be implemented in the form of a computer program on the server of the product certification application data submission system, including the following steps:

[0079] Step 1, obtain the signature digital certificate, encrypted digital certificate, and encrypted private key of the applicant of the product certification application data from the mobile terminal; wherein, the signature digital certificate and the encrypted digital certificate are generated by the CA center according to the synthesized public key and the account of the product certification application data submission system; the synthesized public key is generated according to the first public key factor of the mobile terminal and the second public key factor of the CA center;

[0080] Step 2, create an electronic seal according to the main information of the applicant, and calculate and generate a mobile collaborative electronic signature seal according to the electronic seal, signature digital certificate, encrypted digital certificate, and encrypted private key;

[0081] Step 3, implant the mobile collaborative electronic signature seal on the product certification application data.

[0082] After generating the mobile collaborative electronic signature seal, after the operator of the authentication applicant passes the identity authentication and logs in to the product certification application data submission system using the mobile intelligent terminal, preview the submitted product certification application data. If the identity authentication is unauthorized, the preview of the submitted product certification application data cannot be performed. After the operator of the authentication applicant passes the identity authentication and logs in to the product certification application data submission system, download the submitted product certification application data. If the identity authentication is unauthorized, the download of the submitted product certification application data cannot be performed. The product certification application data submission system provides the verification function of the product certification application data by verifying the mobile collaborative electronic signature seal. The failure of the verification of the mobile collaborative electronic signature seal indicates that the submitted product certification application data has been tampered with or is illegal. At this time, it is prompted that the submitted product certification application data is invalid.

[0083] The above method of mobile collaborative electronic signature for generating stage certifications of target products enables the operator of the certification applicant to achieve role-compliant identity authentication on the basis of submitting certification application materials, thereby guaranteeing the legitimate identity of the operator of the certification applicant. The operator of the certification applicant can quickly and conveniently complete the process of submitting the certification application materials, ensuring the timeliness of the submission of the certification application materials. Through the identity authentication of the operator of the certification applicant, security control over the access to and downloading of the certification application materials is guaranteed. A service for verifying the legality of the certification application materials is provided to ensure the ex post facto verification of the legality of the certification application materials.

[0084] The method for product stage certification based on mobile collaborative electronic signature provided in this embodiment can intelligently determine whether a product meets the stage progress based on the certification application materials and thus give a certification, improving the certification efficiency and ensuring the security, timeliness, legality, and convenience of the submission of the certification application materials.

[0085] Figure 3 It is a schematic structural diagram of an electronic device in an embodiment of the present invention. Specifically refer to Figure 3 hereinafter, which shows a schematic structural diagram of the electronic device 300 suitable for implementing the present invention. The electronic device 300 in the embodiment of the present invention may include, but is not limited to, a product certification application material submission system server composed of a laptop computer, a PAD (tablet computer), a desktop computer, a dedicated server, etc. Figure 3 The electronic device shown is only an example and should not impose any limitation on the functions and usage scope of the embodiments of the present invention.

[0086] As Figure 3 shown, the electronic device 300 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage device 308 into the random access memory (RAM) 303 to implement the method of the embodiments as described in the present invention. In the RAM 303, various programs and data required for the operation of the electronic device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. The input / output (I / O) interface 305 is also connected to the bus 304.

[0087] Typically, the following devices can be connected to the I / O interface 305: input devices 306 including, for example, a touch screen, a touchpad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; output devices 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; storage devices 308 including, for example, magnetic tapes, hard disks, etc.; and a communication device 309. The communication device 309 can allow the electronic device 300 to communicate with other devices wirelessly or wiredly to exchange data. Although Figure 3 the electronic device 300 with various devices is shown, it should be understood that it is not required to implement or have all the shown devices. More or fewer devices can be alternatively implemented or had.

[0088] Specifically, according to an embodiment of the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present invention includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium. The computer program contains program codes for executing the method shown in the flowchart, so as to implement the method as described above. In such an embodiment, the computer program can be downloaded and installed from the network through the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above functions defined in the method of the embodiment of the present invention are executed.

[0089] In one embodiment, referring to Figure 4 as shown, a product stage certification device 400 based on mobile collaborative electronic signature is provided. The device 400 includes: a first generation module 410, an acquisition module 420, an output module 430, and a second generation module 440; wherein,

[0090] The first generation module 410 is configured to receive the product certification application materials of the target product and generate a first feature vector representing the core content of the product certification application materials;

[0091] The acquisition module 420 is configured to acquire a second feature vector representing the index text of the target product; wherein, the index text is generated based on the first feature vectors of the core contents of the product certification application materials that the target product has completed and a plurality of R & D stage labels corresponding to the target product; the R & D stage labels corresponding to the target product are determined according to the category of the target product;

[0092] The output module 430 is configured to input the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to represent whether to add the core content of the product certification application materials to the index text;

[0093] The second generation module 440 is configured to generate a mobile collaborative electronic signature for the phased certification of the target product if the output result is yes.

[0094] Optionally, the first generation module 410 is further configured to perform segmentation processing on the product certification application materials to obtain at least one text segment; for each text segment, obtain the previous text segment corresponding to the text segment and the summary vector of the previous text segment, where the previous text segment is all text segments before the text segment in the product certification application materials, splice the summary vector of the previous text segment and the feature vector of the text segment to obtain the summary vector of the text segment; and generate a first feature vector representing the core content of the product certification application materials according to the summary vectors corresponding to the text segments.

[0095] Optionally, the first generation module 410 is further configured to input each picture into a generative model to obtain a description text of the picture; in the product certification application materials, determine a target text segment related to the picture, and integrate the description text with the target text segment.

[0096] Optionally, the first generation module 410 is further configured to perform recognition on the picture to obtain a text recognition result; determine the image quality of the picture based on the image feature vector of the picture, where the image quality is used to characterize the clarity of the picture; input the text recognition result, the image feature vector, and the image quality score into a pre-trained generative model to obtain a description text of the picture; and the image quality score is used to determine the participation weight of the text recognition result when the generative model understands the picture.

[0097] Optionally, the indexing module 420 is further configured to set a plurality of nodes of the index text according to each R & D stage label; determine the explanatory text of each node and generate a third feature vector of the explanatory text, where the explanatory text is used to characterize the key features of the target product required by each node; calculate the similarity between the third feature vectors of the explanatory texts and the first feature vectors of the core contents of the completed product certification application materials to obtain the product certification application materials of the target product under each node; and generate the index text according to the core content of the product certification application materials of the target product under each node.

[0098] Optionally, the indexing module 420 is further configured to determine a search keyword and at least one genre search intention based on the R & D stage label corresponding to each node; search for a plurality of multimedia information corresponding to the genre search intention according to the search keyword; and perform multi-modal understanding on the plurality of multimedia information to determine the explanatory text of each node.

[0099] Optionally, if the output result is yes, it includes: the output result is used to characterize whether the core content of the product certification application materials corresponds to the first target node of the index text, and there is no corresponding target product certification application materials under the first target node; or, the output result is used to characterize whether the core content of the product certification application materials corresponds to the second target node of the index text, and there is corresponding target product certification application materials under the second target node, and the core content of the product certification application materials is used to update the second target node.

[0100] Optionally, the second generation module 440 is further configured to: obtain the signature digital certificate, encryption digital certificate and encryption private key of the applicant of the product certification application materials from the mobile terminal; wherein, the signature digital certificate and the encryption digital certificate are generated by the CA center according to the synthesized public key and the product certification application materials submission system account; the synthesized public key is generated according to the first public key factor of the mobile terminal and the second public key factor of the CA center; create an electronic seal according to the subject information of the applicant, and calculate and generate a mobile collaborative electronic signature according to the electronic seal, signature digital certificate, encryption digital certificate and encryption private key; implant the mobile collaborative electronic signature on the product certification application materials.

[0101] It should be noted that the product stage certification device 400 based on the mobile collaborative electronic signature provided by the embodiments of the present invention can be used to execute the technical solutions of the above method embodiments, and its implementation principle and technical effects are similar, which will not be elaborated here.

[0102] In one embodiment, a computer-readable medium is further provided, which carries one or more programs. When the above one or more programs are executed by an electronic device, the electronic device can implement the following functions:

[0103] Receive the product certification application materials of the target product, and generate a first feature vector representing the core content of the product certification application materials; obtain a second feature vector representing the index text of the target product; wherein, the index text is generated based on the first feature vectors of the core contents of the product certification application materials that have been completed for the target product and multiple R & D stage labels corresponding to the target product; the R & D stage labels corresponding to the target product are determined according to the category of the target product; input the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to characterize whether to add the core content of the product certification application materials to the index text; if the output result is yes, generate a mobile collaborative electronic signature for the stage certification of the target product.

[0104] The above description is only a preferred embodiment of the present invention and an explanation of the technical principles applied. Those skilled in the art should understand that the scope of disclosure involved in the present invention is not limited to the technical solution formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the above disclosure concept. For example, the technical solution formed by mutually replacing the above features with the technical features (but not limited to) having similar functions disclosed in the present invention.

Claims

1. A product stage certification method based on mobile collaborative electronic signatures, characterized in that, The method includes: Receiving product certification application materials of a target product, where the product certification application materials include text, performing segmentation processing on the product certification application materials to obtain at least one text segment; for each text segment, obtaining the previous text segment corresponding to this text segment and the summary vector of the previous text segment, where the previous text segment is all text segments before this text segment in the product certification application materials, splicing the summary vector of the previous text segment and the feature vector of this text segment to obtain the summary vector of this text segment; generating a first feature vector representing the core content of the product certification application materials according to the summary vectors corresponding to the respective text segments; Obtaining a second feature vector representing an index text of the target product; where the index text is generated based on the first feature vectors of the core contents of the respective product certification application materials completed for the target product and multiple R & D stage labels corresponding to the target product, and includes: setting multiple nodes of the index text according to the respective R & D stage labels; determining search keywords and at least one genre search intention based on the R & D stage label corresponding to each node; searching for multiple multimedia information corresponding to the genre search intention according to the search keywords; performing multimodal understanding on the multiple multimedia information to determine the explanatory text for each node and generating a third feature vector of the explanatory text, where the explanatory text is used to represent the key features of the target product required for each node; calculating the similarity between the third feature vectors of the respective explanatory texts and the first feature vectors of the core contents of the respective completed product certification application materials to obtain the target product certification application materials under each node; generating the index text according to the core content of the target product certification application materials under each node; the R & D stage labels corresponding to the target product are determined according to the category of the target product; Inputting the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to represent whether to add the core content of the product certification application materials to the index text; If the output result is used to represent that the core content of the product certification application materials corresponds to a first target node of the index text, and there is no corresponding target product certification application materials under the first target node; or the output result is used to represent that the core content of the product certification application materials corresponds to a second target node of the index text, and there are corresponding target product certification application materials under the second target node, and the core content of the product certification application materials is used to update the second target node, then generate a mobile collaborative electronic signature for the target product's phased certification; The product certification application materials include pictures. Each picture is recognized to obtain a text recognition result. The image quality of the picture is determined based on the image feature vector of the picture, and the image quality is used to characterize the clarity of the picture. The text recognition result, the image feature vector, and the image quality score are input into a pre-trained generative model to obtain a description text of the picture. The image quality score is used to determine the participation weight of the text recognition result when the generative model understands the picture. In the product certification application materials, a target text segment related to the picture is determined, and the description text is integrated with the target text segment.

2. The product stage authentication method based on mobile collaborative electronic signature according to claim 1, characterized in that: The steps of generating the mobile collaborative electronic seal for the stage certification of the target product specifically include: Obtain the signature digital certificate, encryption digital certificate, and encryption private key of the applicant of the product certification application materials from the mobile terminal. Among them, the signature digital certificate and the encryption digital certificate are generated by the CA center according to the synthesized public key and the product certification application materials submission system account. The synthesized public key is generated according to the first public key factor of the mobile terminal and the second public key factor of the CA center. Create an electronic seal according to the applicant's subject information, and calculate and generate a mobile collaborative electronic seal according to the electronic seal, signature digital certificate, encryption digital certificate, and encryption private key. Embed the mobile collaborative electronic seal in the product certification application materials.

3. A product stage certification device based on mobile collaborative electronic signatures, characterized in that, Including: The first generation module is used to receive the product certification application materials of the target product. The product certification application materials include text. The product certification application materials are segmented to obtain at least one text segment. For each text segment, the corresponding previous text segment and the summary vector of the previous text segment are obtained. The previous text segment is all the text segments before this text segment in the product certification application materials. The summary vector of the previous text segment and the feature vector of this text segment are spliced to obtain the summary vector of this text segment. According to the summary vectors corresponding to each text segment, a first feature vector representing the core content of the product certification application materials is generated. An acquisition module, configured to acquire a second feature vector representing an index text of the target product; wherein, the index text is generated based on a first feature vector of the core content of each product certification application document completed for the target product and a plurality of R & D stage labels corresponding to the target product, including: setting a plurality of nodes of the index text according to each of the R & D stage labels; determining a search keyword and at least one genre search intention based on the R & D stage label corresponding to each node; searching for a plurality of multimedia information corresponding to the genre search intention according to the search keyword; performing multimodal understanding on the plurality of multimedia information to determine an explanatory text for each node and generating a third feature vector of the explanatory text, the explanatory text being used to represent the key features of the target product required for each node; calculating the similarity between the third feature vector of each explanatory text and the first feature vector of the core content of each of the completed product certification application documents to obtain the target product certification application document under each node; for each node, generating the index text according to the core content of the target product certification application document under the node; the R & D stage label corresponding to the target product is determined according to the category of the target product; An output module, configured to input the first feature vector and the second feature vector into a pre-trained neural network model to obtain an output result; the output result is used to represent whether to add the core content of the product certification application document to the index text; A second generation module, configured to generate a mobile collaborative electronic signature for the phased certification of the target product if the output result is used to represent that the core content of the product certification application document corresponds to a first target node of the index text and there is no corresponding target product certification application document under the first target node; or, the output result is used to represent that the core content of the product certification application document corresponds to a second target node of the index text, there is a corresponding target product certification application document under the second target node, and the core content of the product certification application document is used to update the second target node; A picture processing module, configured to, when the product certification application document includes pictures, identify each picture to obtain a text recognition result; determine the image quality of the picture based on the image feature vector of the picture, the image quality being used to represent the clarity of the picture; input the text recognition result, the image feature vector and the image quality score into a pre-trained generative model to obtain a description text of the picture; the image quality score is used to determine the participation weight of the text recognition result when the generative model understands the picture; in the product certification application document, determine a target text segment related to the picture, and integrate the description text with the target text segment.

4. The product stage authentication device based on mobile collaborative electronic signature according to claim 3, wherein, The second generation module is further configured to: The signature digital certificate, encrypted digital certificate, and encrypted private key of the applicant who obtains the product certification application materials from the mobile terminal; wherein, the signature digital certificate and the encrypted digital certificate are generated by the CA center according to the synthesized public key and the product certification application materials submission system account; the synthesized public key is generated according to the first public key factor of the mobile terminal and the second public key factor of the CA center; Create an electronic seal according to the applicant's entity information, and calculate and generate a mobile collaborative electronic signature according to the electronic seal, signature digital certificate, encrypted digital certificate, and encrypted private key; Embed the mobile collaborative electronic signature on the product certification application materials.

Images