Method and device for determining network quality optimization scheme, computer device, readable storage medium and program product

Abstract

The application relates to a network quality difference optimization scheme determination method and device, computer equipment, a readable storage medium and a program product. The method comprises the following steps: obtaining a network quality difference problem of a target area; querying a pre-constructed vector database according to the network quality difference problem; the vector database stores information of multiple quality difference cases; in the case that a first matching result of the network quality difference problem is not queried from the vector database, querying a pre-constructed graph database according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship among a quality difference phenomenon, a quality difference root cause, an optimization scheme and an occurrence frequency; a second reasoning model is called to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target area. The method can improve the determination efficiency of the quality difference optimization scheme.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of big data, and in particular to a network quality difference optimization scheme determination method and device, computer equipment, computer readable storage medium and computer program product. BACKGROUND

[0002] With the development of network technology, base stations are becoming more and more important. At present, the number of base stations is increasing, and the operation and maintenance difficulty is increasing. When there are some progress performance problems in the network, which affect the use of users, base station cell optimization, i.e. quality difference optimization, is needed.

[0003] In the related art, the wireless network quality difference is generally determined by the experience of network optimization experts to determine the quality difference root cause and propose an optimization scheme. The personal knowledge of the expert is highly dependent, and the optimization scheme needs to be determined first before the optimization scheme is determined. The operation and maintenance efficiency of this step-by-step method is low. SUMMARY

[0004] Therefore, it is necessary to provide a network quality difference optimization scheme determination method, device, computer equipment, computer readable storage medium and computer program product to solve the technical problem of low operation and maintenance efficiency of the method of determining the quality difference root cause first and then determining the optimization scheme.

[0005] In a first aspect, the present application provides a network quality difference optimization scheme determination method. The method comprises:

[0006] Obtaining a network quality difference problem of a target area;

[0007] According to the network quality difference problem, querying a pre-constructed vector database; the vector database stores information of multiple quality difference cases;

[0008] In the case where no first matching result of the network quality difference problem is queried from the vector database, according to the network quality difference problem, querying a pre-constructed graph database to obtain a second matching result; the graph database stores a mapping relationship between quality difference phenomena, quality difference root causes, optimization schemes and occurrence times;

[0009] Calling a second reasoning model to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target area.

[0010] In one of the embodiments, the graph database is constructed by the following method:

[0011] Obtaining network data of a sample area;

[0012] Performing knowledge extraction on the network data to form multiple quality difference data combinations; each quality difference data combination includes a quality difference phenomenon, a quality difference root cause and an optimization scheme;

[0013] Count the occurrence times of each quality difference data combination to obtain a mapping relationship of quality difference phenomenon, quality difference root cause, optimization scheme and occurrence times, and form the graph database.

[0014] In one of the embodiments, the second matching result is obtained by querying the pre-constructed graph database according to the network quality difference problem, including:

[0015] The network quality difference problem is vectorized, and / or the keywords of the network quality difference problem are extracted;

[0016] The second matching result is obtained by querying the graph database based on the network quality difference problem vector and / or the keywords.

[0017] In one of the embodiments, the vector database is constructed by the following method:

[0018] Obtain a plurality of quality difference case documents;

[0019] Each quality difference case document is sliced and each slice obtained is vectorized;

[0020] The vector database is constructed according to the vectorized slices of each quality difference case document.

[0021] In one of the embodiments, the second matching result is obtained by querying the pre-constructed vector database according to the network quality difference problem, including:

[0022] The network quality difference problem is vectorized;

[0023] The network quality difference problem vector is matched with each slice in the vector database;

[0024] The target slice with a matching degree greater than a threshold value with the network quality difference problem vector is filtered from the slices as the first matching result.

[0025] In one of the embodiments, the method further includes:

[0026] In the case that the first matching result of the network quality difference problem is queried from the vector database, a first inference model is called to process the first matching result to obtain a target optimization scheme for the network quality difference problem of the target area.

[0027] In a second aspect, the application further provides a network quality difference optimization scheme determination device. The device includes:

[0028] An obtaining module is configured to obtain a network quality difference problem of a target area;

[0029] The first query module is configured to query a vector database pre-constructed according to the network quality difference problem; the vector database stores information of multiple quality difference cases;

[0030] The second query module is configured to, in a case where the first matching result of the network quality difference problem is not queried from the vector database, query a graph database pre-constructed according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship between a quality difference phenomenon, a quality difference root cause, an optimization scheme and an occurrence frequency;

[0031] The reasoning module is configured to call a second reasoning model to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target region.

[0032] In a third aspect, the present application further provides a computer device. The computer device comprises a memory and a processor, the memory stores a computer program, and the processor realizes the following steps when executing the computer program:

[0033] obtaining a network quality difference problem of a target region;

[0034] querying a vector database pre-constructed according to the network quality difference problem; the vector database stores information of multiple quality difference cases;

[0035] in a case where a first matching result of the network quality difference problem is not queried from the vector database, querying a graph database pre-constructed according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship between a quality difference phenomenon, a quality difference root cause, an optimization scheme and an occurrence frequency;

[0036] calling a second reasoning model to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target region.

[0037] In a fourth aspect, the present application further provides a computer readable storage medium. The computer readable storage medium stores a computer program, and the computer program is executed by a processor to realize the following steps:

[0038] obtaining a network quality difference problem of a target region;

[0039] querying a vector database pre-constructed according to the network quality difference problem; the vector database stores information of multiple quality difference cases;

[0040] In a case where the first matching result of the network quality problem is not queried from the vector database, a second matching result is obtained by querying a pre-constructed graph database according to the network quality problem; the graph database stores a mapping relationship between a quality problem phenomenon, a quality problem root cause, an optimization scheme and an occurrence frequency;

[0041] The second reasoning model is called to process the second matching result, to obtain a target optimization scheme for the network quality problem of the target region.

[0042] In a fifth aspect, the present application further provides a computer program product. The computer program product comprises a computer program, which, when executed by a processor, implements the following steps:

[0043] A network quality problem of a target region is obtained.

[0044] A pre-constructed vector database is queried according to the network quality problem; the vector database stores information of multiple quality problem cases.

[0045] In a case where the first matching result of the network quality problem is not queried from the vector database, a second matching result is obtained by querying a pre-constructed graph database according to the network quality problem; the graph database stores a mapping relationship between a quality problem phenomenon, a quality problem root cause, an optimization scheme and an occurrence frequency.

[0046] The second reasoning model is called to process the second matching result, to obtain a target optimization scheme for the network quality problem of the target region.

[0047] The above network quality optimization scheme determination method, device, computer device, storage medium and computer program product determine the network quality optimization scheme according to a network quality problem of a target region, first query a pre-constructed vector database, in a case where the first matching result of the network quality problem is not queried from the vector database, then query a pre-constructed graph database according to the network quality problem, to obtain a second matching result, and then call a second reasoning model to process the second matching result, to obtain a target optimization scheme for the network quality problem of the target region. This method constructs a vector database, a graph database and a reasoning model in advance, and combines the use, so that the optimization scheme can be obtained only by inputting the network quality problem, without determining the quality problem root cause in advance, thereby improving the determination efficiency of the quality problem optimization scheme. Further, this method takes into account the efficient retrieval capability of the vector database and the deep reasoning capability of the graph database, queries the vector database first, and then queries the graph database in a case where no corresponding matching result is found, and through the vector and knowledge graph double recall mode, the balance between the determination efficiency and accuracy of the quality problem optimization scheme can be effectively guaranteed. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 A flowchart of a method for determining a network quality difference optimization scheme in an embodiment;

[0049] Figure 2 A flowchart of a graph database construction step in an embodiment;

[0050] Figure 3 A flowchart of a vector database construction step in an embodiment;

[0051] Figure 4 A flowchart of a method for determining a network quality difference optimization scheme in another embodiment;

[0052] Figure 5 An architecture diagram of a method for determining a network quality difference optimization scheme in an embodiment;

[0053] Figure 6 A structural block diagram of a device for determining a network quality difference optimization scheme in an embodiment;

[0054] Figure 7 An internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0055] In order to make the purposes, technical solutions and advantages of the present application clearer, further detailed description will be made to the present application in combination with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application, and are not used to limit the present application.

[0056] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and do not necessarily describe a specific order or sequence. It should be understood that the data thus used can be interchanged under appropriate circumstances, so that the embodiments of the present application described herein can be implemented in an order other than that illustrated or described herein.

[0057] In an embodiment, as shown in Figure 1 A method for determining a network quality difference optimization scheme is provided, and the embodiment takes the method applied to a terminal as an example. It should be understood that the method can also be applied to a server, and can also be applied to a system including a terminal and a server, and can be realized through the interaction of the terminal and the server. The terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, Internet of Things devices and portable wearable devices. The Internet of Things device can be a smart speaker, a smart television, a smart air conditioner, a smart vehicle device, etc. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The server can be realized by an independent server or a server cluster composed of multiple servers. In the embodiment, the method includes the following steps:

[0058] Step S110, obtaining a network quality difference problem of a target area.

[0059] The target area represents an area for which a network quality difference optimization scheme needs to be determined. In actual applications, the target area can be a base station cell.

[0060] The network quality difference problem can be a description of the quality difference situation of the target area, such as unstable connection, low data rate, etc. In some embodiments, it can also be the original network quality difference data of the target area, such as signal strength, data rate, delay, etc.

[0061] In specific implementations, the content input by the user can be received, and the network quality difference problem can be obtained according to the content input by the user.

[0062] Step S120, querying a pre-constructed vector database according to the network quality difference problem; the vector database stores information of multiple quality difference cases.

[0063] In specific implementations, after obtaining the network quality difference problem, the vector database can be queried first to find similar quality difference cases, so as to obtain an optimization scheme. Specifically, before querying, the obtained network quality difference problem needs to be vectorized and converted into a vector form. For example, a pre-trained vectorization model can be used to convert the network quality difference problem from a text form into a numerical vector suitable for fast similarity comparison. After the conversion is completed, a similarity search is performed in the vector database to find cases similar to the current network quality difference problem vector. If a quality difference case highly related to the network quality difference problem is found, it is recorded as the first matching result.

[0064] Step S130, in the case where the first matching result of the network quality difference problem is not found from the vector database, querying a pre-constructed graph database according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship between quality difference phenomena, quality difference root causes, optimization schemes, and occurrence times.

[0065] In specific implementations, if the first matching result satisfying the condition cannot be found in the vector database, for example, a threshold can be set, and if the similarity between the content retrieved from the vector database and the network quality difference problem is lower than the threshold, it is considered that the first matching result cannot be found, and the graph database will be queried. The graph database pre-stores the mapping relationship between the quality difference phenomena, the quality difference root causes, the optimization schemes, and the occurrence times, and the information is organized as a graph structure, so that the relationship query and reasoning can be efficiently performed.

[0066] When querying the graph database, according to the network quality problem, a query request is sent to the graph database to find the quality difference phenomenon related to the problem, and the corresponding quality difference root cause, optimization scheme and occurrence frequency are obtained as the second matching result according to the determined quality difference phenomenon.

[0067] It can be understood that there may be multiple quality difference phenomena related to the network quality problem obtained by querying the graph database, and therefore multiple matching results may be obtained: <quality difference phenomenon, quality difference root cause, optimization scheme, occurrence frequency>.

[0068] In step S140, a second reasoning model is called to process the second matching result to obtain a target optimization scheme for the network quality problem of the target region.

[0069] In specific implementation, the second reasoning model is a reasoning model constructed for the matching result of the graph database. The second reasoning model can be a large model, for example, a general large model or a large model fine-tuned by network optimization corpus. The model can deduce a specific optimization scheme for the target region by analyzing the association between the quality difference root cause and the optimization scheme. The second matching result obtained usually includes the corresponding quality difference phenomenon, root cause, optimization scheme and occurrence frequency. The second reasoning model is used to analyze and process these data to generate a target optimization scheme for the target region.

[0070] It can be understood that the second matching result can include multiple scenarios. Therefore, in the reasoning process of the second reasoning model, the model will also consider the occurrence frequency as a reasoning basis for the target optimization scheme. For example, the more the occurrence frequency of a quality difference phenomenon is, the greater the probability that the optimization scheme corresponding to the quality difference phenomenon is selected into the target optimization scheme is.

[0071] In some embodiments, considering that the quality difference optimization scheme is usually in a unified format and has low humanization, a wireless expert quality difference optimization scheme output template can be defined. The second reasoning model is preset as a network optimization expert role and is given certain information summarization ability. The second reasoning model is required to output results in a specified format, and a generation example is provided for few-shot learning to make the generated results more stable.

[0072] In this step, a multi-thought chain reasoning method is used to match different thought chain abilities according to the types of knowledge recall. For example, for a graph database, reasoning is performed based on the large model knowledge graph reasoning in combination with the occurrence frequency of the quality difference optimization scheme, so that the returned knowledge is closer to the real network world.

[0073] In the method for determining the network quality difference optimization scheme, according to the network quality difference problem of the target area, a vector database constructed in advance is queried; in the case that no first matching result of the network quality difference problem is queried from the vector database, a graph database constructed in advance is queried according to the network quality difference problem to obtain a second matching result; and then a second reasoning model is called to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target area. The method, by constructing the vector database, the graph database and the reasoning model in advance and in combination, can obtain the optimization scheme only by inputting the network quality problem, without determining the quality difference root cause in advance, thereby improving the determination efficiency of the quality difference optimization scheme. Further, the method considers the efficient retrieval capability of the vector database and the deep reasoning capability of the graph database, queries the vector database first, and then queries the graph database in the case that no corresponding matching result is found, and through the vector and knowledge graph double recall mode, the balance between the determination efficiency and the accuracy of the quality difference optimization scheme can be effectively guaranteed.

[0074] In an example embodiment, as shown in FIG. 13, the graph database in step S130 is constructed by the following method: Figure 2

[0075] In step S210, network data of a sample area is obtained.

[0076] Specifically, through a big data platform, user measurement report (MR) data, base station cell performance report (KPI) data, alarm data, and work parameter data in the sample area are obtained as network data, as well as expert cases and work order return information. Specifically, the network data includes:

[0077] (1) The MR data, the KPI data, the alarm data, and the work parameter data are summarized in the dimension of a cell.

[0078] (2) All quality difference cells in history are summarized in the dimension of a cell, and work order data (including return data) are extracted from a work order system.

[0079] (3) The MR data, the KPI data, the alarm data, and the work parameter data are extracted in the dimension of a cell when a quality difference occurs, and are associated with a quality difference cell list to form quality difference phenomenon data mainly in the form of quality difference types and quality difference data.

[0080] (4) The quality difference phenomenon data and the work order return data are associated in the dimension of a cell.

[0081] In step S220, knowledge extraction is performed on the network data to form a plurality of quality difference data combinations; each quality difference data combination includes a quality difference phenomenon, a quality difference root cause, and an optimization scheme. ​

[0082] In a specific implementation, knowledge can be extracted through a large model knowledge graph prompt project. Specifically, the large model can be given certain roles, such as a knowledge graph project expert and a wireless network optimization expert, which are good at accurately extracting entities (subjects, objects) and relationships of a wireless network knowledge graph from text and can make appropriate summary descriptions of the meanings of the entities and relationships. The large model can also be given certain abilities, such as wireless network entity extraction ability, wireless network relationship extraction ability, and wireless network context association ability. The large model can also be given certain constraints, such as using as much information as possible in the associated context to enrich the content of entities and relationships and not fabricating arbitrarily. The data format of the large model can also be agreed upon, such as <quality defect phenomenon-quality defect root cause-quality defect optimization scheme>. Thus, multiple quality defect data combinations including quality defect phenomena, quality defect root causes, and optimization schemes are obtained.

[0083] In step S230, the number of occurrences of each quality defect data combination is counted to obtain a mapping relationship of quality defect phenomena, quality defect root causes, optimization schemes, and occurrence numbers, which constitutes a graph database.

[0084] In a specific implementation, after obtaining the quality defect data combinations, the number of occurrences of each quality defect data combination can be counted. For example, each time the knowledge extraction is performed, the obtained quality defect data combination is compared with the existing quality defect data combinations, and the similarity is judged. If the similarity meets the condition (e.g., greater than a threshold), it is considered as the same quality defect data combination, and the number of occurrences of the quality defect data combination is incremented by 1. If the similarity does not meet the condition, it is considered as a new quality defect data combination, and its number of occurrences is recorded as 1. In this way, after all data processing is completed, a <quality defect phenomenon-quality defect root cause-quality defect optimization scheme-occurrence number> four-tuple can be formed and stored in the graph database.

[0085] In this embodiment, semi-structured data such as work orders and cases are collected, wireless entity extraction and wireless relationship extraction are performed through a large model, similar knowledge is merged and counted, a <quality defect phenomenon-quality defect root cause-quality defect optimization scheme-occurrence number> four-tuple is formed, and stored in a graph database to realize the construction of a wireless quality defect knowledge graph. Subsequently, the database can be directly searched to determine the optimization scheme of the network quality defect problem and improve the efficiency of determining the network quality defect optimization scheme.

[0086] In an example embodiment, the above step S130 queries the pre-constructed graph database according to the network quality defect problem to obtain a second matching result, including: vectorizing the network quality defect problem, and / or extracting keywords of the network quality defect problem; querying the graph database based on the network quality defect problem vector and / or the keywords to obtain the second matching result.

[0087] In a specific implementation, the pre-constructed graph database according to the network quality problem query can be queried by vector retrieval, keyword or mixed keyword and vector retrieval, etc. Specifically, after the user inputs the network quality problem, a retrieval formula can be constructed according to the network quality problem, and the cosine similarity, Euclidean distance, etc. are used to calculate the similarity between the retrieval formula and each record in the database. According to the similarity score, a number of records most matched with the query are returned, which can be arranged in descending order of similarity.

[0088] More specifically, the text of the network quality problem can be directly converted into a vector as a first retrieval formula; based on the similarity between the first retrieval formula and each record in the graph database, a second matching result is obtained. The keywords in the network quality problem can also be extracted by natural language technology to construct a second retrieval formula; based on the similarity between the second retrieval formula and each record in the graph database, a second matching result is obtained. The matching result obtained by the first retrieval formula and the matching result obtained by the second retrieval formula can also be fused as the final second matching result.

[0089] For example, the result list and similarity obtained by the first retrieval formula retrieval are obtained, and the result list obtained by the second retrieval formula retrieval is obtained, and the relevance of each result to the second retrieval formula is evaluated. Finally, the two results are merged, for example, the two retrieval methods can be assigned weights, for each retrieval result, the result under the two query methods can be weighted and summed by the weights of the two query methods to obtain a total score, and the total score is sorted according to the total score. Select the top several results as the second matching result. For example, if the weight of the first retrieval formula corresponding to the retrieval result is α, and the weight of the second retrieval formula corresponding to the retrieval result is β, the total score can be expressed as: α x similarity + β x relevance score.

[0090] In this embodiment, the queries by vector retrieval, keyword retrieval and mixed retrieval of the two can complement each other and improve the efficiency and accuracy of knowledge recall together.

[0091] In an exemplary embodiment, as shown in Figure 3 The above vector database is constructed by the following methods:

[0092] Step S310, a plurality of quality difference case documents are obtained;

[0093] Step S320, each quality difference case document is respectively sliced and each slice obtained is vectorized;

[0094] Step S330, the vectorized slices of each quality difference case document are used to construct a vector database.

[0095] In a specific implementation, the case documents about network quality problems can be collected from various channels, and the slice length is determined according to the content and structure of the documents. Each quality problem case document is sliced according to the slice length, so that each document can obtain multiple slices. Then, each slice is vectorized by using a pre-trained vectorization model, that is, the text slice is converted into a vector representation, and the vectorized slices of each quality problem case document are stored in a preset database to form a vector database.

[0096] In some embodiments, after obtaining the case documents, semantic annotation can also be performed, for example, annotation of key elements such as quality problem phenomena, root causes, and optimization schemes in the case documents, to facilitate subsequent analysis.

[0097] In some embodiments, to increase context information, overlapping slices (for example, 50% overlap between each slice) can be set when the quality problem case documents are sliced, to capture more semantic associations.

[0098] In this embodiment, the slices of the case documents are stored in the form of high-dimensional vectors through vectorization, so that similar cases and optimization schemes can be efficiently obtained through knowledge retrieval according to user input, the time cost of traditional text retrieval is greatly reduced, and the retrieval efficiency is improved.

[0099] In an example embodiment, the step S120 of querying the pre-constructed vector database according to the network quality problem includes: vectorizing the network quality problem; matching the network quality problem vector with each slice in the vector database; and screening target slices with a matching degree greater than a threshold from each slice as a first matching result.

[0100] In a specific implementation, when querying the vector database, the network quality problem is first converted into a high-dimensional vector representation. Then, the matching is performed by calculating the similarity (such as cosine similarity) between the network quality problem vector and the vectors of all slices in the vector database. A similarity threshold (for example, 0.75) can be predefined, and target slices with a similarity higher than the threshold are screened from the matched slices, and the screened target slices are returned as the first matching result.

[0101] In this embodiment, the query process is fast and efficient through vectorization and similarity retrieval, and the response speed is greatly improved compared with traditional text retrieval.

[0102] In an example embodiment, the method further includes: in the case of querying the first matching result of the network quality problem from the vector database, calling a first inference model to process the first matching result to obtain a target optimization scheme for the network quality problem of the target region.

[0103] In a specific implementation, if a first matching result of the network quality problem is queried in the vector database, a first reasoning model is called to process the first matching result. The first reasoning model is a reasoning model constructed for the matching result of the vector database. The first reasoning model can also be a large model, for example, a general large model or a large model fine-tuned through network optimization corpus. The large model is preset as a network optimization expert role and is endowed with certain skills and constraints. The large model is required to think of topN possible quality difference root causes and optimization schemes according to the matching result returned by the vector database, and the quality difference root causes and the optimization schemes have a corresponding relationship.

[0104] In some embodiments, considering that the quality difference optimization schemes are mostly in a unified format and have low humanization, a wireless expert quality difference optimization scheme output template can be defined. The first reasoning model is preset as a network optimization expert role and is endowed with certain information summarization ability. The first reasoning model is required to output the result in a specified format, and a generation example is provided for few-shot learning, so that the generated effect is more stable.

[0105] In this embodiment, a multi-thought chain reasoning mode is adopted. Different thought chain abilities are matched according to different types of knowledge recall. For example, for the vector database, a large model matched with the recall result of the vector database is trained to perform knowledge graph reasoning, so as to improve the accuracy of the target optimization scheme of reasoning.

[0106] Reference Figure 4 The flowchart of the method for determining the network quality difference optimization scheme is provided for another embodiment. In this embodiment, the following steps are included.

[0107] In step S410, the network quality problem of the target area is acquired.

[0108] In step S420, the network quality problem is processed by vectorization, and the obtained network quality problem vector is matched with each slice in the vector database.

[0109] In step S430, if there is a target slice in each slice that has a matching degree greater than a threshold with the network quality problem vector, the target slice is taken as a first matching result.

[0110] In step S440, a first reasoning model is called to process the first matching result, and a target optimization scheme for the network quality problem of the target area is obtained.

[0111] In step S450, if there is no target slice in each slice that has a matching degree greater than a threshold with the network quality problem vector, the network quality problem is processed by vectorization, and / or the keywords of the network quality problem are extracted.

[0112] Step S460: Based on the network quality problem vector and / or the keyword, query the graph database to obtain a second matching result.

[0113] Step S470: Call the second inference model to process the second matching result to obtain a target optimization solution for the poor network quality problem in the target area.

[0114] This method pre-builds a vector database, a graph database, and an inference model, and uses them in combination. It only needs to input the network quality problem to obtain an optimization plan, without first determining the root cause of the poor quality. This can improve the efficiency of determining the poor quality optimization plan. Furthermore, this method takes into account the efficient retrieval capability of the vector database and the deep reasoning capability of the graph database. It first queries the vector database, and then queries the graph database if no corresponding matching results are found. The dual recall of vectors and knowledge graphs can effectively ensure the balance between the efficiency and accuracy of determining the poor quality optimization plan.

[0115] In one embodiment, in order to facilitate those skilled in the art to understand the embodiments of the present application, the following will be described with reference to specific examples in conjunction with the accompanying drawings. Figure 5 , shows a schematic diagram of the architecture of a method for determining a network quality optimization solution, including a data processing module, a knowledge recall module, a reasoning module, and a generation module. The process of determining a network quality optimization solution through each module includes:

[0116] (1) Collecting raw data: Through the big data platform, obtain user measurement report (MR) data, base station cell performance report (KPI) data, alarm data, work parameter data, expert cases and work order return information in the sample area.

[0117] (2) Knowledge extraction and storage: On the one hand, the large model knowledge graph prompts the engineering to extract knowledge, count the number of similar knowledge, and form a four-tuple: <phenomenon description-root cause of quality difference-optimization solution-number of occurrences>, which is stored in the graph database; on the other hand, the case documents are sliced ​​and vectorized and stored in the vector database; knowledge is stored in two different ways.

[0118] (3) Vector database recall: Since vector retrieval is faster, vector database recall is given priority. The vector database is queried based on the user's poor network quality problem. If a matching result that meets the threshold of the user's problem is found, the result list is returned (such as selecting the topN results) and the reasoning of step (5) is entered. If not, the reasoning of step (4) is entered for graph database recall.

[0119] (4) Graph database recall: According to the user's network quality problem, query related knowledge entries based on vector / keyword / mixed retrieval and return a result list.

[0120] In some embodiments, considering the large overhead of the graph database, combined with the fast and slow thinking mode of human beings, after obtaining the user's network quality problem, the problem type can be determined, and if it is a simple problem, the vector database retrieval is preferred, and if it is a complex problem, the graph database retrieval is used.

[0121] (5) Multi-thought chain reasoning: Based on a general large model or a large model fine-tuned by network optimized corpus, and preset as a network optimization role. If the recall result belongs to the result returned by the vector database, construct the vector database thought chain reasoning ability, and let the large model deduce the related quality difference optimization scheme step by step according to the returned knowledge; if the recall result is the result returned by the graph database, construct the knowledge graph thought chain reasoning ability, and let the large model deduce the related quality difference optimization scheme according to the graph reasoning content and the number of occurrences.

[0122] (5.1) Vector database reasoning: The large model is preset as a network optimization expert role, and is given certain skills and constraints, requiring the large model to think of topN possible quality difference root causes and optimization schemes according to the results returned by the vector database, and the quality difference root cause and the optimization scheme need to have a corresponding relationship.

[0123] (5.2) Knowledge graph thought chain reasoning: The large model is preset as a network optimization expert role, and is given certain skills and constraints, requiring the large model to think of topN possible quality difference root causes and optimization schemes according to the results returned by the graph database and the number of occurrences, and the quality difference root cause and the optimization scheme need to have a corresponding relationship.

[0124] In this step, different thought chain reasoning is performed according to the different sources of recalled knowledge, and the advantages of the two different data formats are fully utilized to make the reasoning result more accurate.

[0125] (6) Hint engineering answer generation: Through few-shot (zero sample) hint engineering, the large model outputs the final quality difference optimization scheme according to the specified format of the current quality difference phenomenon, reference work order phenomenon, quality difference root cause, and quality difference optimization scheme, so that the generated effect is more stable.

[0126] In this step, according to the needs of front-line operation and maintenance personnel, the optimization scheme that meets the needs of front-line personnel is generated through the preset optimization scheme template.

[0127] The present application is directed to the current wireless quality difference optimization, which has the problems of high dependence on expert knowledge, low automation rate, and great optimization difficulty. A quality difference optimization scheme based on a large model is proposed. One aspect provides a large model to prompt engineering to extract knowledge, counts the number of similar knowledge, forms a four-tuple of <phenomenon description-quality difference root cause-optimization scheme-occurrence frequency>, and forms a graph database. On the other hand, the case documents are sliced and vectorized to form a vector database. Knowledge is stored in two different ways. During retrieval, multiple retrieval is performed, knowledge is selected and aggregated according to similarity, and finally the result is output by the large model prompt engineering. Compared with the prior art, the present application has the following beneficial effects:

[0128] 1. High efficiency: the generalization ability of the large model is strong, and the knowledge is automatically extracted by the large model using the prompt engineering method to construct the knowledge graph, reducing manual operation.

[0129] 2. High accuracy: a vector database and a four-tuple graph database are constructed, and the powerful reasoning ability of the large model can improve the accuracy.

[0130] 3. The dual recall mode of vector and knowledge graph can effectively balance the efficiency and accuracy of quality difference optimization.

[0131] 4. High humanization: using the generation ability of the large model, the root cause positioning and optimization scheme are generated in one step according to the specified format, which is more close to the needs of front-line optimization personnel.

[0132] 5. Strong sustainability: with the continuous iteration of knowledge, more knowledge is automatically extracted, and the recommendation will be more accurate.

[0133] It should be understood that although each step in the flowchart involved in each embodiment described above is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in each embodiment described above can include multiple steps or stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0134] Based on the same inventive concept, the embodiments of the present application also provide a network quality difference optimization scheme determination apparatus for implementing the network quality difference optimization scheme determination method described above. The apparatus provides a solution implementation similar to the implementation described in the above method, and therefore the specific limitations in one or more network quality difference optimization scheme determination apparatus embodiments provided below can refer to the limitations of the network quality difference optimization scheme determination method described above, which will not be described here again.

[0135] In one embodiment, as shown in Figure 6 a network quality difference optimization scheme determination apparatus is provided, comprising:

[0136] The acquisition module 610 is configured to acquire a network quality difference problem of a target region.

[0137] The first query module 620 is configured to query a pre-constructed vector database according to the network quality difference problem; the vector database stores information of multiple quality difference cases.

[0138] The second query module 630 is configured to, in a case where no first matching result of the network quality difference problem is queried from the vector database, query a pre-constructed graph database according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship between a quality difference phenomenon, a quality difference root cause, an optimization scheme and an occurrence frequency.

[0139] The reasoning module 640 is configured to call a second reasoning model to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target region.

[0140] In one embodiment, the apparatus further comprises a graph database construction module configured to acquire network data of a sample region; perform knowledge extraction on the network data to form multiple quality difference data combinations; each quality difference data combination comprises a quality difference phenomenon, a quality difference root cause and an optimization scheme; and count the occurrence frequency of each quality difference data combination to obtain a mapping relationship between the quality difference phenomenon, the quality difference root cause, the optimization scheme and the occurrence frequency, and form the graph database.

[0141] In one embodiment, the second query module 630 is further configured to perform vectorization processing on the network quality difference problem, and / or extract keywords of the network quality difference problem; query the graph database based on the network quality difference problem vector and / or the keywords to obtain the second matching result.

[0142] In one embodiment, the apparatus further comprises a vector database construction module configured to acquire multiple quality difference case documents; perform slicing processing on each quality difference case document respectively, and perform vectorization processing on each slice obtained; and construct the vector database according to the vectorized slices of each quality difference case document.

[0143] In one of the embodiments, the first query module 620 is further configured to perform vectorization processing on the network quality problem; match the network quality problem vector with each slice in the vector database; and filter out, as a first matching result, a target slice that has a matching degree greater than a threshold with the network quality problem vector from the slices.

[0144] In one of the embodiments, the reasoning module 640 is further configured to, in a case where the first matching result of the network quality problem is queried from the vector database, call the first reasoning model to process the first matching result, and obtain a target optimization scheme for the network quality problem in the target region.

[0145] The modules in the network quality optimization scheme determination apparatus described above can be implemented in whole or in part by software, hardware, and combinations thereof. The modules described above can be embedded in or independent of a processor in a computer device in hardware form, or stored in a memory in a computer device in software form, so as to be called and executed by a processor to perform the operations corresponding to the modules described above.

[0146] In one embodiment, a computer device is provided, which can be a terminal, and an internal structure diagram of the computer device can be as shown in Figure 7 The computer device includes a processor, a memory, a communication interface, a display screen, and an input device connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is configured to perform wired or wireless communication with an external terminal. The wireless communication can be achieved through WIFI, mobile cellular network, NFC (Near Field Communication), or other technologies. The computer program is executed by the processor to implement a network quality optimization scheme determination method. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or a key, trackball, or touchpad arranged on the shell of the computer device. The input device can also be an external keyboard, touchpad, or mouse, etc.

[0147] Those skilled in the art can understand that Figure 7 The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. A specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0148] In an embodiment, a computer device is also provided, comprising a memory and a processor, the memory storing a computer program, and the processor implementing the steps in the above method embodiments when executing the computer program.

[0149] In an embodiment, a computer readable storage medium is provided, storing a computer program, which, when executed by a processor, implements the steps in the above method embodiments.

[0150] In an embodiment, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the steps in the above method embodiments.

[0151] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant laws, regulations and standards of relevant countries and regions.

[0152] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when the computer program is executed, the processes of the above-mentioned embodiments of the methods can be included. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, etc., without being limited thereto.

[0153] Any combination of the technical features of the above embodiments can be made. In order to make the description simple, all possible combinations of the technical features in the above embodiments are not described, however, as long as the combination of the technical features does not exist, it should be considered as the scope of the present application.

[0154] The above embodiments only express several implementation manners of the present application, and the description is more specific and detailed, but it should not be understood as a limitation on the scope of the patent of the present application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the present application should be subject to the appended claims.

Claims

1. A method for determining a network quality optimization scheme, the method comprising: determining a network quality optimization scheme based on a network quality optimization scheme database. The method comprises: obtaining a network quality difference problem of a target region; querying a pre-constructed vector database according to the network quality difference problem; the vector database stores information of multiple quality difference cases; in a case where a first matching result of the network quality difference problem is not queried from the vector database, querying a pre-constructed graph database according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship between a quality difference phenomenon, a quality difference root cause, an optimization scheme and a number of occurrences; calling a second reasoning model to process the second matching result to obtain a target optimization scheme for the network quality difference problem of the target region.

2. The method of claim 1, wherein, The graph database is constructed by the following method: obtaining network data of a sample region; performing knowledge extraction on the network data to form multiple quality difference data combinations; each quality difference data combination comprises a quality difference phenomenon, a quality difference root cause and an optimization scheme; counting the number of occurrences of each quality difference data combination to obtain a mapping relationship between a quality difference phenomenon, a quality difference root cause, an optimization scheme and a number of occurrences, and to constitute the graph database.

3. The method of claim 2, wherein, The querying of the pre-constructed graph database according to the network quality difference problem to obtain a second matching result comprises: performing vectorization processing on the network quality difference problem, and / or extracting keywords of the network quality difference problem; querying the graph database based on a network quality difference problem vector and / or the keywords to obtain the second matching result.

4. The method of claim 1, wherein, The vector database is constructed by the following method: obtaining multiple quality difference case documents; respectively performing slicing processing on each quality difference case document, and performing vectorization processing on each slice obtained; constructing the vector database according to the vectorized slices of each quality difference case document.

5. The method of claim 4, wherein, The querying of the pre-constructed vector database according to the network quality difference problem comprises: performing vectorization processing on the network quality difference problem; matching a network quality difference problem vector with each slice in the vector database; screening, from the slices, a target slice with a matching degree greater than a threshold value with the network quality difference problem vector as the first matching result.

6. The method of claim 1, wherein, The method further comprises: in a case where the first matching result of the network quality difference problem is queried from the vector database, calling a first reasoning model to process the first matching result to obtain a target optimization scheme for the network quality difference problem of the target region.

7. A network quality optimization scheme determination apparatus, characterized by comprising: a network quality optimization scheme determination unit configured to determine a network quality optimization scheme based on a network quality optimization scheme determination rule. The device comprises: an obtaining module configured to obtain a network quality difference problem of a target region; a first querying module configured to query a pre-constructed vector database according to the network quality difference problem; the vector database stores information of multiple quality difference cases; a second querying module configured to, in a case where a first matching result of the network quality difference problem is not queried from the vector database, query a pre-constructed graph database according to the network quality difference problem to obtain a second matching result; the graph database stores a mapping relationship between a quality difference phenomenon, a quality difference root cause, an optimization scheme and a number of occurrences; An inference module is configured to invoke a second inference model to process the second matching result, and obtain a target optimization scheme for a network quality problem of the target region.

8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that, The processor implements the steps of the method for determining the network quality optimization scheme according to any one of claims 1 to 6 when executing the computer program.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method for determining the network quality optimization scheme according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program, when executed by the processor, implements the steps of the method for determining the network quality optimization scheme according to any one of claims 1 to 6. The computer program, when executed by the processor, implements the steps of the method for determining the network quality optimization scheme according to any one of claims 1 to 6.

Images