Server structure data pushing method, system, equipment and medium

Abstract

The invention provides a server structure data pushing method, system and device and a medium, and the method comprises the steps: responding to detected retrieval problem data input by a user, and analyzing the retrieval problem data to obtain retrieval key features, the retrieval problem data comprising image data and / or text data; searching server structure data to be processed matched with the retrieval key features in a pre-established server structure knowledge base on the basis of the obtained retrieval key features; and obtaining a user portrait matched with the user, and carrying out personalized processing on the to-be-processed server structure data based on the user portrait to generate target server structure data and pushing the target server structure data. The server structure data with high matching degree with the user is pushed, and the user experience is improved; and furthermore, based on a pre-established comprehensive knowledge base, the accuracy and comprehensiveness of the pushed data are improved.

Description

Technical Field

[0001] The present application relates to the field of server technology, and in particular to a method, system, device and medium for pushing server structure data. Background Art

[0002] The server structure is a core component of a computer system, and its design and configuration directly affect the performance, reliability, and scalability of the server. In order to better understand the server structure, the existing technology usually provides users with queryable server structure data by building a server structure knowledge base.

[0003] However, the existing data push methods based on server structure knowledge base still have problems such as incomplete push content and inconsistency with user search content.

[0004] Therefore, a highly accurate server structure data push method is urgently needed to solve the above technical problems. Summary of the invention

[0005] Based on this, it is necessary to provide a server structure data push method, system,

[0006] Equipment and media to solve the above technical problems.

[0007] In a first aspect, the present application provides a method for pushing server structure data, the method comprising:

[0008] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0009] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0010] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0011] In some embodiments, the method further includes building the server structure knowledge base:

[0012] Acquire existing server related data and pre-process the existing server related data to generate server structure block data guided by server structure keywords;

[0013] Based on the server structure block data, generating a server component graph model including a server component structure and the server component text vector;

[0014] Generate a server structure text vector based on a preset embedding model and server component text vectors;

[0015] Generate a server structure graph vector based on the regional convolutional neural network, the graph convolutional network and the server component graph model;

[0016] A server structure knowledge base is constructed based on the server structure text vector and the server structure graph vector.

[0017] In some embodiments, generating a server component graph model including a server component structure and the server component text vector based on the server structure block data includes:

[0018] Parsing the server structure relationship contained in the server structure block data;

[0019] Building a server component graph model based on the parsed server structure relationship and the graph database;

[0020] Based on the server structure relationship and text embedding technology, obtaining a server component text vector;

[0021] A server structure vector database is constructed based on the server component graph model and the server component text vector.

[0022] In some embodiments, before parsing the search question data to obtain the search key features, the method further includes:

[0023] Parsing the search question data to divide the search question data into search levels, wherein the search levels include a first level and a second level;

[0024] In response to detecting that the retrieval question data is at the first level, calling a preset first database to obtain to-be-processed server structure data matching the retrieval question data;

[0025] In response to detecting that the retrieval question data is at the second level, parsing the retrieval question data to obtain a retrieval key feature;

[0026] Wherein, the first database pre-stores a plurality of first search questions and associated server structure data;

[0027] The step of parsing the search question data to classify the search question data into search levels includes:

[0028] comparing the search question data with a plurality of the first search questions contained in the first database;

[0029] If there is a first search question matching the search question data in the first database, determining that the search question data is at the first level;

[0030] If there is no first search question matching the search question data in the first database, it is determined that the search question data is at the second level.

[0031] In some embodiments, the key retrieval feature includes a text retrieval feature and / or an image retrieval feature, and the parsing of the retrieval question data to obtain the key retrieval feature includes:

[0032] Based on a preset large language model, parsing text data in the search question data to obtain text search features;

[0033] Based on a preset image processing module, the image data in the retrieval question data is parsed to obtain image retrieval features.

[0034] In some embodiments, based on the acquired search key feature, searching a pre-built server structure knowledge base for to-be-processed server structure data that matches the search key feature includes:

[0035] In response to the retrieval key feature being an image retrieval feature, generating an image retrieval vector matching the image retrieval feature;

[0036] Searching for a target server structure graph vector matching the image retrieval vector in the server structure knowledge base based on the image retrieval vector;

[0037] Calculating the image-text similarity between the target server structure graph vector and a plurality of server structure text vectors included in the server structure knowledge base, and determining the target server structure text vector based on the image-text similarity;

[0038] Generate server structure data to be processed based on the target server structure text vector and the target server structure graph vector;

[0039] Based on the acquired search key feature, searching a pre-built server structure knowledge base for to-be-processed server structure data matching the search key feature, further comprising:

[0040] In response to the retrieval key feature being a text retrieval feature, generating a text retrieval vector matching the image retrieval feature;

[0041] Searching for a target server structure text vector matching the text retrieval vector in the server structure knowledge base based on the text retrieval vector;

[0042] Calculating the image-text similarity between the target server structure text vector and a plurality of server structure graph vectors included in the server structure knowledge base, and determining the target server structure graph vector based on the image-text similarity;

[0043] The server structure data to be processed is generated based on the target server structure text vector and the target server structure graph vector.

[0044] In some embodiments, the to-be-processed server structure data is personalized based on the user portrait to generate target server structure data and push it, including:

[0045] Determine a server push restriction condition based on the user portrait and retrieve a target prompt word template matching the user portrait;

[0046] Screening the to-be-processed server structure data based on the server push restriction condition to generate transition server structure data;

[0047] Based on the target prompt word template, the transition server structure data is adjusted to generate the target server structure data and push it.

[0048] In a second aspect, the present application provides a server structure data push system, the system comprising:

[0049] A processing module, configured to parse the retrieval question data input by a user to obtain a retrieval key feature in response to detecting the retrieval question data, wherein the retrieval question data includes image data and / or text data;

[0050] A retrieval module, used to search for server structure data to be processed that matches the retrieval key feature in a pre-built server structure knowledge base based on the acquired retrieval key feature;

[0051] The push module is used to obtain a user portrait matched by the user, and based on the user portrait, personalize the server structure data to be processed to generate target server structure data and push it.

[0052] In a third aspect, the present application provides an electronic device, the electronic device comprising:

[0053] one or more processors;

[0054] and a memory associated with one or more processors, the memory being used to store program instructions, which, when read and executed by the one or more processors, perform the following operations:

[0055] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0056] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0057] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0058] In a fourth aspect, the present application further provides a computer-readable storage medium, on which a computer program is stored, and the computer program enables a computer to perform the following operations:

[0059] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0060] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0061] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0062] In a fifth aspect, the present application provides a computer program product, including a computer program, which implements the steps of the following method when executed by a processor:

[0063] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0064] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0065] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0066] The beneficial effects achieved by this application are:

[0067] The present application provides a method for pushing server structure data, including: in response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, the retrieval question data including image data and / or text data; based on the obtained retrieval key features, searching for server structure data to be processed that matches the retrieval key features in a pre-built server structure knowledge base; obtaining a user portrait matched by the user, and based on the user portrait, personalizing the server structure data to be processed to generate target server structure data and push it. Pushing server structure data with a high degree of matching with the user is achieved, improving the user experience; further based on the pre-established comprehensive knowledge base, the accuracy and comprehensiveness of the pushed data are improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments are briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative work, among which:

[0069] Figure 1 It is a schematic diagram of a server structure data push method provided in an embodiment of the present application;

[0070] Figure 2 This is a schematic diagram of raw data collection provided by an embodiment of the present application;

[0071] Figure 3 It is a text segmentation schematic diagram provided in an embodiment of the present application;

[0072] Figure 4 This is a schematic diagram of generating a vector database provided in an embodiment of the present application;

[0073] Figure 5 It is a schematic diagram of generating a server structure knowledge base provided in an embodiment of the present application;

[0074] Figure 6 This is a server structure data push system architecture diagram provided by an embodiment of the present application;

[0075] Figure 7 It is a structural diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0076] In order to make the purpose, technical solutions and advantages of this application clearer, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0077] It should be understood that in the description of the present application, unless the context clearly requires otherwise, words such as "include", "comprises", and the like in the entire specification and claims should be interpreted as inclusive rather than exclusive or exhaustive; that is, the meaning of "including but not limited to".

[0078] It should also be understood that the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of this application, unless otherwise specified, "plurality" means two or more.

[0079] It should be noted that the terms "S1", "S2", etc. are only used for the purpose of describing the steps, and do not specifically refer to the order or sequence, nor are they used to limit the present application. They are only for the convenience of describing the method of the present application, and cannot be understood as indicating the order of the steps. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in this field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection required by the present application.

[0080] Embodiment 1

[0081] The embodiment of the present application provides a server structure data push method, which is applied to an information push system, wherein the information push system is integrated with a server structure knowledge base. Specifically, the function of server architecture data push is realized based on the server structure knowledge base, such as Figure 1 As shown, including the following:

[0082] S1. In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain key retrieval features.

[0083] In a specific implementation scenario, before parsing the retrieval question data to obtain the key retrieval features, the embodiment of the present application further proposes: parsing the retrieval question data to divide the retrieval question data into retrieval levels, the retrieval levels including the first level and the second level; in response to detecting that the retrieval question data is the first level, calling the preset first database to obtain the to-be-processed server structure data matching the retrieval question data; in response to detecting that the retrieval question data is the second level, parsing the retrieval question data to obtain the key retrieval features; wherein the first database pre-stores a plurality of first retrieval questions and associated server structure data; wherein parsing the retrieval question data to divide the retrieval question data into retrieval levels includes: comparing the retrieval question data with the plurality of first retrieval questions contained in the first database; if there is a first retrieval question matching the retrieval question data in the first database, determining that the retrieval question data is the first level; if there is no first retrieval question matching the retrieval question data in the first database, determining that the retrieval question data is the second level. After the retrieval question data input by the user is divided into the first level, the first database is called for data feedback; after being divided into the second level, the key retrieval features contained therein are continuously extracted for subsequent retrieval. By pre-classifying the retrieval question data, when the retrieval question is simple and there are corresponding retrieval questions in the historical data, the first database can be directly called to provide the server structure data to be processed corresponding to the retrieval question data; that is, after the server structure data in the first database with a high degree of match with the retrieval question data is returned, it is further personalized to improve the accuracy and relevance of the server structure data finally fed back to the user, and also speed up the feedback speed of the retrieval question input by the user, while maximizing the computing resources.

[0084] Specifically, the first database pre-stores a first search question and server structure data associated with the search question; wherein the first search question can be a simple question about the server structure customized by a person skilled in the art, or a historical question that has been searched and has uniquely corresponding server structure data. The specific method of comparing the search question data with a plurality of first search questions contained in the first database and determining whether there is a first search question that matches the search question data includes converting the search question data and the first search question into vectors and then calculating the cosine similarity between the two vectors using cosine similarity, wherein the cosine similarity reaches a first preset threshold value to determine that the search question data is similar to the first search question, wherein the first preset threshold value can be set by a person skilled in the art based on actual scenarios and experience, and the present application does not impose any restrictions on this. The above-mentioned specific method for determining whether there is a first search question matching the search question data can also be obtained by converting the search question data and the first search question into sets containing vocabulary, and calculating the Jaccard similarity between the two sets, that is, calculating the ratio of the size of the intersection of the two sets to the size of the ice machine. When the Jaccard similarity reaches the second preset threshold, it is determined that the search question data is similar to the first search question, wherein the second preset threshold is set by a technician in this field according to actual scenarios and experience, and this application does not impose any restrictions on this. It can be understood that the above-mentioned specific method for determining whether there is a first search question matching the search question data can also be determined by converting the search question data and the first search question into sequences and calculating the edit distance (Levenshtein Distance) between the two sequences; it can also be determined by calculating the similarity between the first search question and the search question data based on a word embedding method; it can also be determined by comparing the hash values ​​between the first search question and the search question data based on a hash method, and the specific content will not be expanded; it can be understood that this application can select a suitable judgment method according to the actual scenario, and this application does not impose any restrictions on the specific judgment method.

[0085] It can be understood that the present application supports image and text retrieval, so the above-mentioned retrieval question data includes retrieval question data including image data and / or text data; specifically, the above-mentioned retrieval question data to obtain retrieval key features includes: based on a preset large language model, parsing the text data in the retrieval question data to obtain text retrieval features; based on a preset image processing module, parsing the image data in the retrieval question data to obtain image retrieval features. It can be understood that the present application parses the retrieval question data input by the user through a pre-trained large language model to obtain keywords with retrieval intent and keywords related to the server structure, and generates text retrieval features based on the keywords obtained by the above analysis, wherein the training process of the large language model is a conventional technical means in this field, and the present application will not be expanded here. The present application also extracts features of the input image through a preset image processing module, wherein the image processing module automatically extracts image features based on SIFT (scale invariant feature transform), SURF (accelerated robust features), HOG (histogram of oriented gradients) and convolutional neural network (CNN) and other technologies to generate image retrieval features after pre-processing the input image data such as image cropping and resizing, graying and normalization.

[0086] In addition, the present application also proposes to further process the above-mentioned image retrieval features: obtain the above-mentioned generated image retrieval features, and perform dimensionality reduction processing on the above-mentioned image retrieval features; calculate the matching degree between the multiple image retrieval features after dimensionality reduction processing and the server structure, and screen the image retrieval features corresponding to the above-mentioned multiple image retrieval features whose matching degree is greater than or equal to the third preset threshold as the target image retrieval features; use the screened target image retrieval features as the features for subsequent retrieval in the knowledge base. Among them, the dimensionality reduction processing of the image retrieval features specifically includes using principal component analysis and singular value decomposition and other methods to perform dimensionality reduction processing on the above-mentioned image retrieval features, thereby reducing the dimension of the feature vector and reducing the complexity of subsequent calculations; the present application further screens the characteristic retrieval features with high matching degree with the server structure in the image retrieval features according to the correlation between the image retrieval features and the server structure, in order to further improve the accuracy and efficiency of using image data to retrieve in the knowledge base.

[0087] S2. Based on the acquired retrieval key features, search for server structure data to be processed that matches the retrieval key features in a pre-built server structure knowledge base.

[0088] In a specific implementation scenario, based on the acquired retrieval key features, searching for server structure data to be processed that matches the retrieval key features in a pre-built server structure knowledge base includes:

[0089] In response to the retrieval key feature being an image retrieval feature, an image retrieval vector matching the image retrieval feature is generated; based on the image retrieval vector, a target server structure graph vector matching the image retrieval vector is searched in a server structure knowledge base; the image-text similarity between the target server structure graph vector and a plurality of server structure text vectors included in the server structure knowledge base is calculated, and the target server structure text vector is determined based on the image-text similarity; and server structure data to be processed is generated based on the target server structure text vector and the target server structure graph vector.

[0090] Based on the acquired retrieval key features, searching for server structure data to be processed that matches the retrieval key features in a pre-built server structure knowledge base also includes: in response to the retrieval key features being text retrieval features, generating a text retrieval vector that matches the image retrieval features; searching for a target server structure text vector that matches the text retrieval vector in the server structure knowledge base based on the text retrieval vector; calculating the graphic and text similarity between the target server structure text vector and multiple server structure graph vectors contained in the server structure knowledge base, and determining the target server structure graph vector based on the graphic and text similarity; and generating the server structure data to be processed based on the target server structure text vector and the target server structure graph vector.

[0091] Among them, the calculation formula of image-text similarity is specifically expressed as: Where T represents the server structure text vector, I is the image feature vector, and Sim(T,I) represents the cosine similarity between text and image. By using the image-text similarity calculation technology and comparing the vector representations of text and image, the matching and search between images and texts are realized. Users can input the text description of the server structure data to realize the relevant search of the image, or provide the server structure image to find the relevant text description; realize the push of multi-modal server structure data, and improve the user experience.

[0092] Among them, the construction process of the above-mentioned pre-built server structure knowledge base includes the following contents: obtaining existing server-related data and pre-processing the existing server-related data to generate server structure block data guided by server structure keywords; based on the server structure block data, generating a server component graph model containing the server component structure and a server component text vector; based on the preset embedding model and the server component text vector, generating a server structure text vector; based on the regional convolutional neural network, the graph convolutional network and the server component graph model, generating a server structure graph vector; based on the server structure text vector and the server structure graph vector, constructing a server structure knowledge base. The present application realizes the automated extraction of server structure data contained in the existing server-related data, including server structure graph vectors and server structure text vectors, and constructs a knowledge base based on the extracted server structure data, thereby reducing manual input, expanding the coverage of the knowledge base, and further improving the accuracy and comprehensiveness of the server structure queried by later users.

[0093] Specifically, the above method of obtaining existing server related data and preprocessing the existing server related data to generate server structure block data guided by server structure keywords includes two parts: data cleaning and text segmentation.

[0094] like Figure 2 As shown, the above data cleaning process includes: collecting all raw data related to any existing server, the source of which includes but is not limited to existing data in the databases of major server-related companies, offline collected data, downloading public website data, etc.; the raw data includes any documents and materials related to the server structure, such as document data in the format of technical documents, papers, blogs, patents, etc. of server manufacturers, as well as multimodal data such as tables, pictures, audio, and video; the above documents and materials are sufficient to cover all server configurations, network topologies, storage architectures, operating systems, and performance indicators in the current scenario. Data cleaning operations are performed on the collected raw data related to the existing servers, including server standardization data, server text cleaning, and server field analysis operations to generate existing server-related data. The above operations reduce noise and redundant information in the raw data, and extract the core content of the raw data to facilitate subsequent processing, and ensure the consistency and availability of the existing server-related data generated after cleaning.

[0095] like Figure 3As shown, the above text segmentation process includes: predetermining a keyword list related to the server structure, including names of the basic structures that constitute the server, such as "chassis", "CPU", "memory", "hard disk", "power supply", "cable", "guide rail", "label", "external card", "motherboard", "backplane", "fan", and "radiator". Based on the above keyword list, search for paragraphs containing these keywords in the above existing server-related data. Specifically, it can be implemented through a programming language or through a keyword extraction algorithm, such as the TF-IDF (Term Frequency-Inverse Document Frequency) algorithm. The present application does not impose constraints on the specific keyword extraction algorithm. Then, for these paragraphs containing the above keywords, segmentation is performed to generate and store server structure block data. Exemplarily, the present application discloses a code that uses the Python programming language to segment existing server structure data into paragraphs and stores them after segmentation:

[0096] import re

[0097] keywords=["chassis","CPU","memory","hard disk","power supply","cable","guide rail","label","external card","motherboard","backplane","fan","radiator"]

[0098] paragraph_pattern=r'(?:(?<!\n)\n{2,})(.*?)(?=\n{2,}|$)'

[0099] text_data="""

[0100] This is the first paragraph of text and does not contain keywords.

[0101] This is the second paragraph of text, which contains the key words CPU and memory. CPU is the core component of a computer, and memory is used to store data.

[0102] The third paragraph of text also does not contain keywords.

[0103] The fourth paragraph contains the keywords hard disk and chassis. The hard disk is used to store data, and the chassis is the skeleton of the computer.

[0104] This is the last paragraph and does not contain keywords.

[0105] """

[0106] paragraphs=re.findall(paragraph_pattern,text_data,re.DOTALL)

[0107] filtered_paragraphs=[]

[0108] forparagraph inparagraphs:

[0109] cleaned_paragraph=paragraph.strip()

[0110] ifany(keyword in cleaned_paragraph forkeyword inkeywords):

[0111] filtered_paragraphs.append(cleaned_paragraph)

[0112] for index,paragraph in enumerate(filtered_paragraphs,start=1):

[0113] print(f"paragraph{index}:\n{paragraph}\n")

[0114] The above-written code searches for paragraphs containing these keywords in the existing server-related data through regular expressions or string matching functions. If the existing server-related data is already organized by paragraphs, the paragraphs can be directly segmented by identifying paragraph markers such as line breaks, blank lines, or specific markers; if the existing server-related data is not organized by paragraphs, natural language processing tools are used to identify sentence boundaries and combine these sentences into logically related paragraphs. Compared with segmentation by the number of characters and segmentation by semantics, this paragraph segmentation guided by server structure keywords not only avoids the shortcomings of semantic rupture, information loss, and poor flexibility in segmentation by the number of characters, but also avoids the problems of complex calculation, context dependence, and difficulty in precise control in segmentation by semantics, so that the server structure block data generated after segmentation has higher semantic coherence, strong adaptability, and easy processing. At the same time, since the paragraph segmentation is based on the relevant keyword list unique to the server structure, the directionality and accuracy of the text segmentation results are greatly improved; it will help to improve the retrieval efficiency in the future and make up for the lack of retrieval accuracy to a certain extent.

[0115] In specific implementation scenarios, such as Figure 4As shown, the above-mentioned server component graph model and server component text vector containing the server component structure are generated based on the server structure block data, including: parsing the server structure relationship contained in the server structure block data; building a server component graph model based on the parsed server structure relationship and the graph database; obtaining the server component text vector based on the server structure relationship and text embedding technology; building a server structure vector database based on the server component graph model and the server component text vector.

[0116] Specifically, after the existing server-related data is divided into blocks to generate server structure block data, the relationship extraction technology is used to analyze the sentences in the above server structure block data to extract the server structure relationship contained therein, such as "CPU is installed in the chassis", "memory is connected to the motherboard", etc. Then, based on the graph database, the server component graph model is constructed using the above acquired server structure relationship. In the graph, each server component can be a node, and the relationship between components is an edge. The attributes of the edge can contain detailed information of the relationship, such as "installation", "connection", etc. Then, using the text embedding technology, a text vector is associated with each node or variable in the graph to realize the conversion of the server structure block data into a server component text vector represented by a text vector. Among them, the above text embedding technology includes but is not limited to the bag of words model, Word2Vec (Word to Vector, a group of related models used to generate word vectors), GloVe (Global Vectors for Word Representation, a word representation tool based on global word frequency statistics), FastText (a model for learning word embedding and text classification) and BERT (Bidirectional Encoder Representations from Transformers). This application reduces storage pressure by vectorizing server block data and provides a basis for realizing image and text fusion push in subsequent retrieval.

[0117] Among them, based on the preset embedding model and server component text vector, the server structure text vector is generated, specifically including using the BERT model through the formula Embedding(X)=BERT(X) to calculate the vector representation of the text data contained in the above server structure block data, where X represents the server structure text in the server block data, and BERT(X) represents the text embedding vector generated by the BERT model. In BERT, the Tokenembedding layer converts each word into a vector of fixed dimension; Segmentembeddings handles the classification task of the input sentence pair; PositionEmbeddings introduces the position information in the sequence to facilitate the knowledge base to understand the order relationship of words or characters in the sequence. It can be understood that in addition to the BERT model, other text embeddings can also be used to achieve the above technical effects, and this application does not limit this.

[0118] Among them, based on the regional convolutional neural network, graph convolutional network and server component graph model, the server structure graph vector is generated; based on the server structure text vector and server structure graph vector, the server structure knowledge base is constructed, which specifically includes the following contents: First, the server component graph model obtained above is further processed using FastR-CNN to generate the bounding boxes and category labels of these models, which are server components. The output of this step is the component region in the image and its preliminary feature representation. FastR-CNN generates candidate regions through the region proposal network (RPN) and uses the convolutional neural network to extract the features of these regions. V R = R-CNN(Image); where V R It is the feature vector of each detected area in the image. FastR-CNN feature extraction formula: i =RoIAlign(x,r i ). Among them, f i is the feature vector of the ith region, x is the input image, r i is the i-th detected area. These features are further input into the graph convolutional neural network (GCN), and the components detected by FastR-CNN and their relationships are constructed into a graph structure. GCN is used to learn the relationship between the nodes (components) in the graph. Through GCN, the mutual correlation between the components in the image is modeled, the image is represented as a graph structure, and the global feature vector of the image is generated, so as to generate the global graph vector corresponding to the entire server based on the server component graph model generated above. GCN generates a graph structure from the feature graph extracted by FastR-CNN, where nodes represent server components and edges represent the relationship between components. Then, a graph convolution operation is performed through GCN to obtain the overall vector representation V of the image G ,V G =GCN(VR ,E). Among them, E is the edge in the graph structure, which represents the relationship between components. GCN graph convolution formula: in, is the adjacency matrix of the graph, is the node degree matrix, H (l) is the node feature matrix of the lth layer, W (l) is a trainable weight matrix and σ is an activation function.

[0119] In addition, in order to improve the user's experience of the server structure data pushed by the server structure knowledge base, a prompt template (prompt word template) can be pre-written to guide the server structure knowledge base to generate output that meets specific requirements. Among them, the prompt template is written by a technician in this field according to actual needs. The specific writing method is a conventional technical means in this field, and this application will not be expanded here. The following is a specific prompt template, which is only for illustration and not for restriction. The specific prompt template does not need to contain all the content in the following template:

[0120] Introduction: Please briefly introduce the basic concepts of server structure, including its main components and their roles in the server. Server components: Describe the core components of the server, such as CPU, memory, motherboard, storage devices (such as hard disk, SSD), power supply, cooling system, expansion card (such as graphics card, network card), etc. For each component, explain its function, importance and role in the operation of the server. Component relationship and connection: Describe the connection mode and relationship between server components, such as how the motherboard connects to the CPU, memory and expansion card, how the power supply supplies power to each component, etc. Emphasize the compatibility and coordination between components, and their impact on the overall performance of the server. Server type and architecture: Introduce different types of servers (such as tower servers, rack servers, blade servers, etc.) and their applicable scenarios and advantages and disadvantages. Describe the server architecture (such as symmetric multi-processing architecture SMP, asymmetric multi-processing architecture AMP, distributed architecture, etc.), and explain the characteristics and applicability of each architecture. Server configuration and optimization: Explain how to choose the appropriate server configuration according to needs, including CPU model, memory capacity, storage capacity, etc. Discuss how to optimize the server to improve performance, such as adjusting BIOS settings, updating drivers, optimizing system configuration, etc. Troubleshooting and maintenance, describe common server failure types (such as hardware failure, software failure, network failure, etc.) and corresponding troubleshooting methods. Explain how to perform daily maintenance and management of the server to ensure its stable operation and extend its service life. Security and backup, discuss the importance of server security and how to protect the server from attacks and data leaks. Introduce the importance of data backup and common backup methods and strategies. Summary and outlook, summarize the main contents of the server structure knowledge base and emphasize the importance of servers in modern information technology. Look forward to the future development trends and possible new technologies of server technology. User guide and interaction, provide guidelines on how to use this knowledge base, including how to query information, how to understand the relationship between components, etc. Encourage users to provide feedback and suggestions to continuously improve and optimize the content and functions of the knowledge base. It is understandable that the data finally pushed by the knowledge base not only contains server structure data, but also includes server-related content in the above templates.

[0121] like Figure 5As shown, this application uses existing server-related data, vector databases, and prompt templates to build and train a server structure knowledge base. The above knowledge base can be understood as a model for retrieving server structure data based on the retrieval question data input by the user. Due to the rapid iteration of relevant content under the server structure, this application also proposes setting a regular update schedule, such as monthly or quarterly updates and establishing a feedback system so that users or internal teams can report any errors, omissions, or content that needs to be updated. After collecting the updated content, repeat the above content and use the new data set to update the server structure knowledge base.

[0122] S3. Obtain the user portrait matched by the user, and perform personalized processing on the server structure data to be processed based on the user portrait to generate the target server structure data and push it.

[0123] Specifically, based on the user portrait, the server push restriction conditions are determined and the target prompt word template matching the user portrait is retrieved; based on the server push restriction conditions, the server structure data to be processed is screened to generate transition server structure data; based on the target prompt word template, the transition server structure data is adjusted to generate the target server structure data and push it. The server push restriction conditions determined according to the user portrait include user needs, user industry, and service structure application scenarios, etc. The target prompt word template is a personalized output template pre-written based on the user portrait. It can be understood that the server structure knowledge base constructed above faces different groups of people, such as sales and R&D. Through the server push restriction conditions, the server structure data to be processed is further screened to obtain server structure data with high adaptability to the user, and then personalized push is performed by retrieving the target prompt word template matching the user portrait according to the user's user portrait, which can significantly improve the quality of the knowledge base push content and further improve the user experience.

[0124] Embodiment 2

[0125] Corresponding to the above-mentioned embodiment 1, as Figure 6 As shown, the embodiment of the present application also provides a server structure data push system, including:

[0126] A processing module 610 is used for parsing the search question data to obtain a search key feature in response to detecting the search question data input by the user, wherein the search question data includes image data and / or text data;

[0127] A retrieval module 620 is used to search for server structure data to be processed that matches the retrieval key feature in a pre-built server structure knowledge base based on the acquired retrieval key feature;

[0128] The push module 630 is used to obtain a user portrait matched with the user, and based on the user portrait, perform personalized processing on the server structure data to be processed to generate target server structure data and push it.

[0129] In some embodiments, the system also includes a preparation module (not shown in the figure) for building the server structure knowledge base: obtaining existing server-related data and preprocessing the existing server-related data to generate server structure block data guided by server structure keywords; based on the server structure block data, generating a server component graph model containing a server component structure and the server component text vector; based on a preset embedding model and the server component text vector, generating a server structure text vector; based on a regional convolutional neural network, a graph convolutional network and the server component graph model, generating a server structure graph vector; based on the server structure text vector and the server structure graph vector, building a server structure knowledge base.

[0130] In some embodiments, the preparation module is also used to parse the server structure relationship contained in the server structure block data; build a server component graph model based on the parsed server structure relationship and the graph database; obtain the server component text vector based on the server structure relationship and text embedding technology; and build a server structure vector database based on the server component graph model and the server component text vector.

[0131] In some embodiments, before parsing the retrieval question data to obtain retrieval key features, the processing module 610 is also used to parse the retrieval question data to divide the retrieval question data into retrieval levels, and the retrieval levels include a first level and a second level; in response to detecting that the retrieval question data is at the first level, calling a preset first database to obtain to-be-processed server structure data matching the retrieval question data; in response to detecting that the retrieval question data is at the second level, parsing the retrieval question data to obtain retrieval key features; wherein the first database pre-stores a plurality of first retrieval questions and associated server structure data; parsing the retrieval question data to divide the retrieval question data into retrieval levels includes: comparing the retrieval question data with the plurality of first retrieval questions contained in the first database; if there is a first retrieval question matching the retrieval question data in the first database, determining that the retrieval question data is at the first level; if there is no first retrieval question matching the retrieval question data in the first database, determining that the retrieval question data is at the second level.

[0132] In some embodiments, the processing module 610 is also used to parse the text data in the retrieval question data to obtain text retrieval features based on a preset large language model; and to parse the image data in the retrieval question data to obtain image retrieval features based on a preset image processing module.

[0133] In some embodiments, the retrieval module 620 is also used to generate an image retrieval vector matching the image retrieval feature in response to the retrieval key feature being an image retrieval feature; search for a target server structure graph vector matching the image retrieval vector in the server structure knowledge base based on the image retrieval vector; calculate the image-text similarity between the target server structure graph vector and multiple server structure text vectors contained in the server structure knowledge base, and determine the target server structure text vector based on the image-text similarity; generate the server structure data to be processed based on the target server structure text vector and the target server structure graph vector; the retrieval module 620 is also used to generate a text retrieval vector matching the image retrieval feature in response to the retrieval key feature being a text retrieval feature; search for a target server structure text vector matching the text retrieval vector in the server structure knowledge base based on the text retrieval vector; calculate the image-text similarity between the target server structure text vector and multiple server structure graph vectors contained in the server structure knowledge base, and determine the target server structure graph vector based on the image-text similarity; generate the server structure data to be processed based on the target server structure text vector and the target server structure graph vector.

[0134] In some embodiments, the push module 630 is also used to determine the server push restriction conditions based on the user portrait and retrieve the target prompt word template that matches the user portrait; based on the server push restriction conditions, filter the server structure data to be processed to generate transition server structure data; based on the target prompt word template, adjust the transition server structure data to generate target server structure data and push it.

[0135] Embodiment 3

[0136] Corresponding to all the above embodiments, an embodiment of the present application provides an electronic device, including: one or more processors; and a memory associated with the one or more processors, the memory is used to store program instructions, and when the program instructions are read and executed by the one or more processors, the following operations are performed:

[0137] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0138] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0139] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0140] in, Figure 7 The architecture of the electronic device is shown as an example, which may include a processor 710, a video display adapter 711, a disk drive 712, an input / output interface 713, a network interface 714, and a memory 720. The processor 710, the video display adapter 711, the disk drive 712, the input / output interface 713, the network interface 714, and the memory 720 may be communicatively connected via a bus 730.

[0141] Among them, the processor 710 can be implemented by a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (Application Specific Integrated Circuit, ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solution provided in this application.

[0142] The memory 720 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 720 can store an operating system 721 for controlling the execution of the electronic device 700, and a basic input and output system (BIOS) 722 for controlling the low-level operations of the electronic device 700. In addition, a web browser 723, a data storage management system 724, and an icon font processing system 725, etc. can also be stored. The above-mentioned icon font processing system 725 can be an application program that specifically implements the operations of the aforementioned steps in the embodiment of the present application. In short, when the technical solution provided in the present application is implemented by software or firmware, the relevant program code is stored in the memory 720 and is called and executed by the processor 710.

[0143] The input / output interface 713 is used to connect the input / output module to realize information input and output. The input / output module can be configured in the device as a component (not shown in the figure), or it can be externally connected to the device to provide corresponding functions. The input device may include a keyboard, a mouse, a touch screen, a microphone, various sensors, etc., and the output device may include a display, a speaker, a vibrator, an indicator light, etc.

[0144] The network interface 714 is used to connect to a communication module (not shown) to realize communication interaction between the device and other devices. The communication module can realize communication through a wired mode (such as USB, network cable, etc.) or a wireless mode (such as mobile network, WIFI, Bluetooth, etc.).

[0145] The bus 730 comprises a pathway for transmitting information between the various components of the device (eg, the processor 710, the video display adapter 711, the disk drive 712, the input / output interface 713, the network interface 714, and the memory 720).

[0146] In addition, the electronic device 700 can also obtain information on specific collection conditions from the virtual resource object collection condition information database for use in condition judgment, etc.

[0147] It should be noted that, although the above device only shows a processor 710, a video display adapter 711, a disk drive 712, an input / output interface 713, a network interface 714, a memory 720, a bus 730, etc., in the specific implementation process, the device may also include other components necessary for normal execution. In addition, it can be understood by those skilled in the art that the above device may also only include components necessary for implementing the solution of the present application, and does not necessarily include all the components shown in the figure.

[0148] It can be seen from the description of the above implementation methods that those skilled in the art can clearly understand that the present application can be implemented by means of software plus a necessary general hardware platform. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the prior art can be embodied in the form of a software product, which can be stored in a storage medium such as ROM / RAM, a disk, an optical disk, etc., including several instructions for a computer device (which can be a personal computer, a cloud service end, or a network device, etc.) to execute the methods of various embodiments of the present application or certain parts of the embodiments.

[0149] Embodiment 4

[0150] Corresponding to all the above embodiments, the embodiments of the present application further provide a computer-readable storage medium, characterized in that it stores a computer program, and the computer program enables a computer to perform the following operations:

[0151] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0152] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0153] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0154] Embodiment 5

[0155] Corresponding to all the above embodiments, the embodiments of the present application further provide a computer program product, which, when executed by a processor, implements the steps of the method described below:

[0156] In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data;

[0157] Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features;

[0158] A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

[0159] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system or system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment. The system and system embodiments described above are merely schematic, in which the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0160] The above are only preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application should be included in the protection scope of the present application.

Claims

1. A method for pushing server structure data, characterized in that: The method comprises: In response to detecting retrieval question data input by a user, parsing the retrieval question data to obtain retrieval key features, wherein the retrieval question data includes image data and / or text data; Based on the acquired search key features, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key features; A user portrait matched with the user is obtained, and based on the user portrait, the server structure data to be processed is personalized to generate target server structure data and push it.

2. The method according to claim 1, characterized in that The method also includes building the server structure knowledge base: Acquire existing server related data and pre-process the existing server related data to generate server structure block data guided by server structure keywords; Based on the server structure block data, generating a server component graph model including a server component structure and the server component text vector; Generate a server structure text vector based on a preset embedding model and server component text vectors; Generate a server structure graph vector based on the regional convolutional neural network, the graph convolutional network and the server component graph model; A server structure knowledge base is constructed based on the server structure text vector and the server structure graph vector.

3. The method according to claim 2, characterized in that The generating, based on the server structure block data, a server component graph model including a server component structure and the server component text vector comprises: Parsing the server structure relationship contained in the server structure block data; Building a server component graph model based on the parsed server structure relationship and the graph database; Based on the server structure relationship and text embedding technology, obtaining a server component text vector; A server structure vector database is constructed based on the server component graph model and the server component text vector.

4. The method according to claim 1, characterized in that: Before parsing the search question data to obtain the search key features, the method further includes: Parsing the search question data to divide the search question data into search levels, wherein the search levels include a first level and a second level; In response to detecting that the retrieval question data is at the first level, calling a preset first database to obtain to-be-processed server structure data matching the retrieval question data; In response to detecting that the retrieval question data is at the second level, parsing the retrieval question data to obtain a retrieval key feature; Wherein, the first database pre-stores a plurality of first search questions and associated server structure data; The step of parsing the search question data to classify the search question data into search levels includes: comparing the search question data with a plurality of the first search questions contained in the first database; If there is a first search question matching the search question data in the first database, determining that the search question data is at the first level; If there is no first search question matching the search question data in the first database, it is determined that the search question data is at the second level.

5. The method according to claim 2, characterized in that: The key retrieval features include text retrieval features and / or image retrieval features, and the step of parsing the retrieval question data to obtain the key retrieval features includes: Based on a preset large language model, parsing text data in the search question data to obtain text search features; Based on a preset image processing module, the image data in the retrieval question data is parsed to obtain image retrieval features.

6. The method according to claim 5, characterized in that Based on the acquired search key feature, searching a pre-built server structure knowledge base for server structure data to be processed that matches the search key feature includes: In response to the retrieval key feature being an image retrieval feature, generating an image retrieval vector matching the image retrieval feature; Searching for a target server structure graph vector matching the image retrieval vector in the server structure knowledge base based on the image retrieval vector; Calculating the image-text similarity between the target server structure graph vector and a plurality of server structure text vectors included in the server structure knowledge base, and determining the target server structure text vector based on the image-text similarity; Generate server structure data to be processed based on the target server structure text vector and the target server structure graph vector; Based on the acquired search key feature, searching a pre-built server structure knowledge base for to-be-processed server structure data matching the search key feature, further comprising: In response to the retrieval key feature being a text retrieval feature, generating a text retrieval vector matching the image retrieval feature; Searching for a target server structure text vector matching the text retrieval vector in the server structure knowledge base based on the text retrieval vector; Calculating the image-text similarity between the target server structure text vector and a plurality of server structure graph vectors included in the server structure knowledge base, and determining the target server structure graph vector based on the image-text similarity; The server structure data to be processed is generated based on the target server structure text vector and the target server structure graph vector.

7. The method according to claim 1, characterized in that The to-be-processed server structure data is personalized based on the user portrait to generate target server structure data and push it, including: Determine a server push restriction condition based on the user portrait and retrieve a target prompt word template matching the user portrait; Screening the to-be-processed server structure data based on the server push restriction condition to generate transition server structure data; Based on the target prompt word template, the transition server structure data is adjusted to generate the target server structure data and push it.

8. A server structure data push system, characterized in that: The system comprises: A processing module, configured to parse the retrieval question data input by a user to obtain a retrieval key feature in response to detecting the retrieval question data, wherein the retrieval question data includes image data and / or text data; A retrieval module, used to search for server structure data to be processed that matches the retrieval key feature in a pre-built server structure knowledge base based on the acquired retrieval key feature; The push module is used to obtain a user portrait matched by the user, and based on the user portrait, personalize the server structure data to be processed to generate target server structure data and push it.

9. An electronic device, characterized in that: The electronic device comprises: one or more processors; And a memory associated with the one or more processors, the memory is used to store program instructions, and when the program instructions are read and executed by the one or more processors, the method according to any one of claims 1-7 is executed.

10. A computer-readable storage medium, characterized in that: The computer program stores a computer program, wherein the computer program enables a computer to execute the method according to any one of claims 1 to 7.

Images