Open source data quality quantification calculation method fusing generative AI

Abstract

The invention relates to the fields of artificial intelligence, big data processing and the like, and provides an open source data quality quantification calculation method fusing generative AI, which comprises the following steps: constructing a data quality quantification index system, and configuring the weight of each data quality quantification index; aiming at the data quality quantification index system, fusing a generative AI to design each data quality quantification index quantification rule; and calculating the data quality based on the weight and the quantification rule. According to the method, a comprehensive data quality quantitative index system is constructed, the quantitative rule of each index is designed by fusing the generative model, automatic and efficient calculation of the data quality can be realized, and the evaluation efficiency of the data quality is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the fields of artificial intelligence, big data processing, etc., in particular to an open source data quality quantification calculation method fusing generative AI. BACKGROUND

[0002] In today's information technology, a large amount of open source data can be obtained through the Internet, news media, social software, etc., which can effectively help people know about emergencies, understand the whole picture of the event, and support people to make rational and correct judgments. However, with the rise of generative AI technology, a large amount of low-quality and false open source information has affected public opinion, confused the public, triggered trust crises, and interfered with people's cognition, judgment and decision-making. In order to reduce the influence of low-quality data, it is urgent to evaluate the data quality.

[0003] There are two methods for data quality evaluation: qualitative and quantitative. Qualitative analysis is heavily dependent on the subjective cognition of analysts, and the results are not reliable and the analysis results are slow. Therefore, current data quality evaluation is biased towards quantitative analysis, which starts from a specific hypothesis, numerically evaluates the indicators, and uses mathematical tools to calculate the numerical relationship between variables. The results are scientific and objective. Open source data quality quantitative analysis involves evaluation index system and index quantification rules. Traditional data quality quantitative evaluation completely relies on human participation in model construction, rule design and index quantification, which is inefficient and difficult to meet the massive data quality evaluation needs of the information age. Therefore, it is urgent to combine artificial intelligence, big data and other frontier technologies with expert experience to build an open source data quality evaluation method fusing generative AI, guided by expert experience, and using artificial intelligence, big data and other frontier technologies as a starting point to meet the needs of massive data quality evaluation and realize the discovery of low-quality and false open source data. SUMMARY

[0004] The present application takes generative AI as a starting point, uses the text semantic understanding and reasoning judgment capabilities of generative AI, and provides an open source data quality quantification calculation method fusing generative AI to solve the problems of low efficiency and difficulty in automation of open source data quality quantification methods dominated by humans.

[0005] The present application provides an open source data quality quantification calculation method fusing generative AI, which comprises:

[0006] Constructing a data quality quantification index system and configuring the weights of each data quality quantification index;

[0007] Designing each data quality quantification index quantification rule fusing generative AI for the data quality quantification index system;

[0008] Calculating data quality based on the weights and quantification rules.

[0009] In some embodiments, the construction data quality quantification index system comprises:

[0010] The framework of constructing the data quality quantification index system comprises a target layer, a criterion layer subordinate to the target layer, and a scheme layer subordinate to the criterion layer;

[0011] The weights of each data quality quantification index are configured based on the analytic hierarchy process.

[0012] In some embodiments, the target layer is open source data quality; the indicators of the criterion layer include integrity, consistency, and diversity; the scheme layer indicators of consistency include spatio-temporal consistency and content consistency, the scheme layer indicators of integrity include element completeness and content completeness, and the scheme layer indicators of diversity include occurrence frequency and data source.

[0013] In some embodiments, the fusion generative AI designs each data quality quantification index quantification rule, comprising:

[0014] The consistency indicator quantification rule comprises equivalent spatio-temporal consistency quantification rule and element consistency quantification rule;

[0015] The integrity indicator quantification rule comprises element completeness quantification rule and content completeness quantification rule;

[0016] The diversity indicator quantification rule comprises occurrence frequency quantification rule and data source quantification rule.

[0017] In some embodiments, the consistency indicator quantification rule comprises:

[0018] A fact database is constructed for the generative AI model and based on a similarity function;

[0019] Based on the fact database, the generative AI model is used to recall associated facts of the open source data to be evaluated based on the similarity function;

[0020] The event consistency is judged based on the associated fact set, and the judgment result includes conflict, no conflict, and cannot be judged;

[0021] Assuming that the conflict appears M times, the no conflict appears N times, and the cannot be judged appears n-M-N times in the event judgment result, the consistency quantification is performed according to the event consistency judgment result:

[0022] If then the spatio-temporal consistency C1 indicator quantification and the element consistency C2 indicator quantification V(C1) = V(C2) = 0.2;

[0023] If and N>n-M-N, the spatiotemporal consistency C1 index quantization and the element consistency C2 index quantization V(C1)=V(C2)=0.7;

[0024] If M=0, and N>n-M-N, the spatiotemporal consistency C1 index quantization and the element consistency C2 index quantization V(C1)=V(C2)=1.

[0025] If M=0, and N>n-M-N, the spatiotemporal consistency C1 index quantization and the element consistency C2 index quantization V(C1)=V(C2)=1.

[0026] If M=0, and N>n-M-N, the spatiotemporal consistency C1 index quantization and the element consistency C2 index quantization V(C1)=V(C2)=1.

[0027] In some embodiments, the element completeness quantification rule comprises:

[0028] For generative AI models, extract elements in the open source data to be evaluated, including time, place, subject, object, and behavior.

[0029] Add all elements in the open source data to be evaluated to obtain the number of elements in the open source data to be evaluated, denoted as num_ele, and then the element completeness C3 is quantified as follows:

[0030]

[0031] Where n represents the number of events of the open source data to be evaluated.

[0032] In some embodiments, the content integrity quantification rule comprises:

[0033] For generative AI models, judge the completeness of grammatical elements.

[0034] Let num_full be the number of complete grammatical elements in all events of the open source data to be evaluated, num_else be the number of incomplete elements, and num_full+num_else=n. According to this, the content integrity C4 quantification value is calculated:

[0035] If num_else≤0.1n, the content integrity C4 index quantization V(C4)=1.

[0036] If 0.1n<num_else≤0.2n, the content integrity C4 index quantization V(C4)=0.7.

[0037] If 0.2n<num_else≤0.3n, the content integrity C4 index quantization V(C4)=0.5. ​

[0038] If 0.3n < num_else ≤ 0.4n, the content integrity C4 index is quantified as V(C4) = 0.3;

[0039] If 0.4n < num_else, the content integrity C4 index is quantified as V(C4) = 0.

[0040] In some embodiments, the occurrence frequency quantification rule comprises:

[0041] Constructing an event set to be evaluated eva_event_set;

[0042] Based on the event set to be evaluated, using a generative AI model and a similarity function to find similar events to be evaluated, and recording the number of similar events to be evaluated as

[0043] The occurrence frequency of the open source data to be evaluated is quantified by using the interval division rule:

[0044] If then the occurrence frequency C5 index is quantified as V(C5) = 1;

[0045] If then the occurrence frequency C5 index is quantified as V(C5) = 0.75;

[0046] If then the occurrence frequency C5 index is quantified as V(C5) = 0.5;

[0047] If then the occurrence frequency C5 index is quantified as V(C5) = 0.25;

[0048] Wherein, n represents the number of events in the open source data to be evaluated, min() represents the minimum value function, and len() represents the set element number calculation function.

[0049] In some embodiments, the data source quantification rule comprises:

[0050] Dividing the data source level, including national government media class1, provincial and municipal unit media class2, commercial company media class3, and other information channels class4;

[0051] Quantifying the data source based on the data source level, tracing the source of the open source data to be evaluated event, and recording it as class, then the data source C6 quantification rule is as follows:

[0052] If class ∈ class1, then the data source C6 index is quantified as V(C6) = 1;

[0053] If class e class2, the data source C6 index quantification V(C6) = 0.75;

[0054] If class e class3, the data source C6 index quantification V(C6) = 0.5;

[0055] If class e class4, the data source C6 index quantification V(C6) = 0.25.

[0056] In some embodiments, the data quality is calculated based on the weight and the quantification rule, comprising:

[0057]

[0058] Wherein, R(event) is the data quality of the open source data event to be evaluated, C i is the i-th scheme layer index, i = 1, 2, 3, 4, 5, 6; W(C i ) is the weight of the scheme layer index C i , and V(C i ) is the quantification result of the scheme layer index C i .

[0059] In summary, due to the adoption of the above technical solutions, the beneficial effects of the present application are:

[0060] The present application constructs a comprehensive data quality quantification index system and integrates the quantification rules of the indexes designed by the generative model, so as to realize automatic and efficient calculation of data quality and improve the data quality evaluation efficiency. BRIEF DESCRIPTION OF DRAWINGS

[0061] Figure 1 A flowchart of a fusion generative AI open source data quality quantification calculation method provided by an embodiment of the present application

[0062] Figure 2 A framework diagram of a data quality evaluation index system in an embodiment of the present application.

[0063] Figure 3 A flowchart of configuring index weights in an embodiment of the present application.

[0064] Figure 4 A flowchart of consistency index quantification in an embodiment of the present application.

[0065] Figure 5 A flowchart of integrity index quantification in an embodiment of the present application.

[0066] Figure 6 A flowchart of diversity index quantification in an embodiment of the present application.

[0067] Figure 7 A schematic structural diagram of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0068] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0069] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort shall fall within the scope of protection of the present invention.

[0070] Example

[0071] like Figure 1 As shown, this embodiment provides a method for quantifying and calculating open source data quality by integrating generative AI, including the following steps:

[0072] Step 1: Build a data quality quantitative indicator system and configure the weight of each data quality quantitative indicator;

[0073] Step 2: Based on the data quality quantitative indicator system, integrate generative AI to design quantitative rules for each data quality quantitative indicator;

[0074] Step 3: Calculate data quality based on the weights and quantification rules.

[0075] The following describes in detail the open source data quality quantification calculation method of the fusion generative AI through specific examples.

[0076] Step 1: Build a data quality quantitative indicator system and configure the weights of each data quality quantitative indicator.

[0077] The data quality quantitative indicator system is a framework for data quality quantification, which directly affects the data quality quantification effect. It is constructed based on expert experience through interviews with experts, questionnaires, expert consultations, etc. Specifically:

[0078] Step 1.1: Build a framework for the data quality quantitative indicator system.

[0079] In this embodiment, in view of the characteristics of the generative AI in generating false data, a framework of the open source data quality quantification index system is constructed from three aspects of integrity, consistency and diversity, involving six dimensions of spatio-temporal consistency, element consistency, element completeness, content integrity, frequency of occurrence and data source, as shown in Figure 2 The framework of the open source data quality quantification index system includes a target layer, a criterion layer subordinate to the target layer, and a scheme layer subordinate to the criterion layer.

[0080] The target layer is the open source data quality.

[0081] The indexes of the criterion layer include integrity, consistency and diversity.

[0082] The consistency refers to checking whether there is a conflict in the content of the data, which can be specifically checked from the perspective of cross verification to check the conflict of the data in the two dimensions of space-time and content. Thus, the indexes of the scheme layer subordinate to the consistency include spatio-temporal consistency and content consistency.

[0083] The integrity refers to checking whether the data content is complete, which measures the data quality from the perspective of grammatical structure. Based on the 6 elements of intelligence (5W1H, Where, When, Who, What, Why, How) and the subject-predicate-object structure of the article, the intelligence elements and the subject-predicate-object structure in the data are extracted by the generative AI model, and the completeness of the data is measured from two dimensions of element completeness and content integrity. Thus, the indexes of the scheme layer subordinate to the integrity include element completeness and content integrity.

[0084] The diversity refers to checking whether the data source channel is reliable, which measures the data quality from the side, and measures the reliability of the data from two dimensions of frequency and source by tracing the data source and associating similar data. Thus, the indexes of the scheme layer subordinate to the diversity include frequency of occurrence and data source.

[0085] Step 1.2: Configure the weight of each data quality quantification index based on the analytic hierarchy process.

[0086] As shown in Figure 3 , based on the framework of the data quality quantification index system in Figure 2 , the analytic hierarchy process is used to construct the judgment matrix between indexes at each level through expert questionnaire, and the contribution of each data quality quantification index to the open source data quality, i.e. the weight, is calculated through consistency check. In this embodiment, the weight of each data quality quantification index is shown in Table 1.

[0087] Table 1, weight of each data quality quantification index:

[0088]

[0089]

[0090] Step 2: For the data quality quantification index system, fuse the quantitative rules of each data quality quantification index generated by the generative AI design.

[0091] Open source data quality quantification refers to quantifying the indexes in the data quality quantification index system to 0-1 numerical values according to the content, source, and other information of the open source data to be evaluated, combining existing facts, involving consistency, completeness, and diversity.

[0092] Step 2.1: Consistency index quantification rule. The consistency index mainly compares the open source data to be evaluated with existing facts, judges whether there is conflicting information based on the semantic understanding and thinking chain reasoning ability of the generative AI model, combines the quantification rules, realizes the quantification calculation of the consistency index, and the whole process contains four parts: constructing a fact database, recalling associated facts of the open source data to be evaluated, open source data consistency judgment, and consistency index quantification calculation, as shown in Figure 4 .

[0093] Step 2.1.1: Construct a fact database for the generative AI model based on the similarity function.

[0094] The news reports of national media are regarded as real data sources, and network crawlers are used to crawl news report data from designated national media websites. For the generative AI model, the following prompt template is constructed {“prompt”:”Please extract the event information from the following data: data.”}, from which the event sentence is obtained. Based on the semantic understanding ability of the generative AI model, the semantic vector of the event sentence is obtained from the Embedding layer, and the similarity function con() is used to calculate the similarity of two events.

[0095]

[0096] where sen ik = [x ik1 ,…,x ikm ,…,x ikp ], sen jl = [x jl1 ,…,x jlm ,…,x jlp ] respectively represent the semantic vectors of two event sentences, p represents the length of the semantic vector, and m represents the m-th vector component.

[0097] Let the judgment threshold be th con , if the semantic similarity con(sen ik , sen jl ) of event sen ik and event sen jl is greater than or equal to th con , it is considered that event senik and sen jl are the same event. Compare the similarity of all events to achieve semantic-based deduplication, and obtain a fact database, denoted as Event_set=[sen1,sen2,…,sen i ,…],wherein sen i represents the i-th event of the fact database.

[0098] Step 2.1.2: Based on the fact database, use the generative AI model and based on the similarity function, recall the associated facts of the to-be-evaluated open source data.

[0099] Based on the semantic information of the to-be-evaluated open source data, find out the associated event sentences from the fact database through semantic correlation. Extract events from the to-be-evaluated open source data event, denoted as eve_sen1,eve_sen2,…,eve_sen n , wherein eve_sen i is the i-th event of the to-be-evaluated open source data event. For any event eve_sen i and any element sen j in the fact database, use the Embedding layer of the generative AI model to obtain their semantic vectors, and use the similarity function con() to calculate the semantic similarity of their semantic vectors. Sort the semantic similarity of all facts in the fact database Event_set with the event eve_sen i from large to small, and take the topk facts as the associated facts of the event eve_sen i , forming the associated fact set of the event eve_sen i Con_eve_sen i =[sen i1 ,sen i2 ,…,sen iA ].

[0100] Step 2.1.3: Judge event consistency based on associated fact set. Based on the event eve_sen i in the to-be-evaluated open source data and the corresponding associated fact set Con_eve_sen i , construct the following prompt template {“prompt”:“Please judge whether the associated facts in the associated data set Con_eve_sen i are consistent with the event eve_sen i ”}. iThere is a contradiction, and choose one from the three results of conflict, no conflict, and cannot judge, and give the specific analysis process, return the result in Json format. Among them, no conflict means that the related facts in the related fact set Con_eve_sen i can support the authenticity of the event eve_sen i ; cannot judge means that the related facts in the related fact set Con_eve_sen i and the event eve_sen i neither conflict nor confirm; contradiction means that the event eve_sen i conflicts with the related facts in the related fact set Con_eve_sen i . Record all event judgment results in the open source data event to be evaluated as Jud_set=[α1,α2,…,α i ,…,α n ].

[0101] Step 2.1.4: According to the event consistency judgment result, the consistency is quantified, and the consistency index quantification is composed of the space-time consistency index quantification and the element consistency index quantification. In order to simplify the input length of the generated formula AI model, the quantification of space-time consistency C1 and element consistency C2 is equivalent.

[0102] Suppose that in the event judgment result Jud_set, conflict occurs M times, no conflict occurs N times, and cannot judge occurs n-M-N times:

[0103] If , it is considered that the open source data event to be evaluated has more conflicts with the fact database, and the space-time consistency C1 index quantification and the element consistency C2 index quantification V(C1)=V(C2)=0.2;

[0104] If and N≤n-M-N, it is considered that the open source data event to be evaluated has a small amount of conflict with the fact database, and most of the information is not supported by the fact database. Then, the space-time consistency C1 index quantification and the element consistency C2 index quantification V(C1)=V(C2)=0.5;

[0105] If and N>n-M-N, it is considered that the open source data event to be evaluated has a small amount of conflict with the fact database, and most of the information is supported by the fact database. Then, the space-time consistency C1 index quantification and the element consistency C2 index quantification V(C1)=V(C2)=0.7;

[0106] If M = 0 and N > n-M-N, it is considered that the open source data event to be evaluated has no conflict with the fact database, and most of the information is supported by the fact database, then the spatio-temporal consistency C1 index quantization and the element consistency C2 index quantization V(C1) = V(C2) = 1.

[0107] If M = 0 and N > n-M-N, it is considered that the open source data event to be evaluated has no conflict with the fact database, and most of the information is supported by the fact database, then the spatio-temporal consistency C1 index quantization and the element consistency C2 index quantization V(C1) = V(C2) = 1.

[0108] It should be noted that the specific values of the above judgment conditions and the specific values of the quantization results are preferred in the embodiment, and can be adjusted as needed.

[0109] Step 2.2: Integrity index quantization rule. The integrity index checks whether the data content is complete, measures the data quality from the grammar point of view, quantizes the element completeness and content completeness by extracting the intelligence elements and grammar structure in the data to be evaluated, and the process is as shown in Figure 5 .

[0110] Step 2.2.1: Element completeness quantization rule. Based on the element extraction result of the data to be evaluated, the element completeness index of the data to be evaluated is quantized.

[0111] Step 2.2.1.1: For generative AI model, extract elements in open source data to be evaluated. The elements in the open source data to be evaluated refer to the basic elements contained in the text information, including time, place, subject, object and behavior.

[0112] Let eve_sen i be any event of the open source data to be evaluated event, for the generative AI model, the following prompt template is constructed {“prompt”:“Please extract the time, place, subject, object, behavior, etc. from the following text, and return it in Json format: eve_sen i .”}. Traverse all events eve_sen1, eve_sen2, …, eve_sen n in the open source data to be evaluated event, and get the time, place, subject, object and behavior elements.

[0113] Step 2.2.1.2: Accumulate all elements in the open source data to be evaluated to get the number of elements in the open source data to be evaluated, denoted as num_ele, then the element completeness C3 is quantized as follows:

[0114]

[0115] Wherein, n represents the number of events of the open source data event to be evaluated.

[0116] Step 2.2.2: Content integrity quantification rule. Based on the content integrity judgment of the generative AI model to the open source data to be evaluated, the content integrity index of the open source data to be evaluated is quantified.

[0117] Step 2.2.2.1: Judgment of syntactic element integrity for generative AI model. Content integrity mainly judges whether the subject, predicate, object and other syntactic elements of the data are complete, and quantitatively calculates the content integrity based on the judgment result. eve_sen i is any event in the open source data event to be evaluated, facing the generative AI model, the following prompt template is constructed {“prompt”:“Judge whether the following text syntax element is complete, return complete or incomplete, for example: ‘I love’, incomplete; ‘I love China’, complete: eve_sen i .”} Traverse all events eve_sen1, eve_sen2, …, eve_sen n , get the judgment result of syntactic element integrity.

[0118] Step 2.2.2.2: Quantitative calculation of content integrity. Let the number of complete syntactic elements in all events of the open source data event to be evaluated be num_full, the number of incomplete elements be num_else, and num_full+num_else=n. According to this, the content integrity C4 quantization value is calculated:

[0119] If num_else≤0.1n, it means that very few events in the open source data event to be evaluated have incomplete syntax structure, and the syntax quality of the open source data event to be evaluated is very high, then the content integrity C4 index quantization V(C4) = 1;

[0120] If 0.1n<num_else≤0.2n, it means that there are a small number of events in the open source data event to be evaluated with incomplete syntax structure, and the syntax quality of the open source data event to be evaluated is high, then the content integrity C4 index quantization V(C4) = 0.7;

[0121] If 0.2n<num_else≤0.3n, it means that a certain number of events in the open source data event to be evaluated have incomplete syntax structure, and the syntax quality of the open source data event to be evaluated is general, then the content integrity C4 index quantization V(C4) = 0.5;

[0122] If 0.3n < num_else ≤ 0.4n, it indicates that there are more event information syntax structures in the to-be-evaluated open source data event, and it is considered that the syntax quality of the to-be-evaluated open source data event is poor, and the content integrity C4 index is quantized as V(C4) = 0.3.

[0123] If 0.4n < num_else, it indicates that there are many event information syntax structures in the to-be-evaluated open source data event, and it is considered that the syntax quality of the to-be-evaluated open source data event is very poor, and the content integrity C4 index is quantized as V(C4) = 0.

[0124] It should be noted that the specific values of the above judgment conditions and the specific values of the quantization results are preferred in this embodiment, and can be adjusted as needed.

[0125] Step 2.3: Diversity index quantization rule.

[0126] The diversity index mainly tests whether the data source is reliable, and measures the data quality from the side. By comparing the semantic content of the to-be-evaluated data, combining the data source level division standard, quantizing the data occurrence frequency and the data source, the quantization process is as shown in Figure 6 .

[0127] Step 2.3.1: Occurrence frequency quantization rule. By constructing a to-be-evaluated event set, find out the data similar to the to-be-evaluated open source data from the to-be-evaluated event set, and according to the data similar to the to-be-evaluated open source data, the occurrence frequency index is quantized.

[0128] Step 2.3.1.1: Constructing a to-be-evaluated event set. Let the to-be-evaluated open source data set be eva_data_set, and any to-be-evaluated open source data event be event i . i The events contained in the i-th to-be-evaluated open source data event are denoted as All events in the to-be-evaluated open source data set eva_data_set are aggregated to form a to-be-evaluated event set eva_event_set, and any eva_event is the j-th to-be-evaluated open source data event j .

[0129] Step 2.3.1.2: Based on the to-be-evaluated event set, find similar to-be-evaluated events by using the generative AI model and the similarity function.

[0130] For any event The semantic vector of the event is constructed by using the generative AI model Embedding layer. The similarity function con() is used to calculate the event The similarity of any event sen in the event set to be evaluated eva_event_set and event sen k .

[0131] The similarity threshold of the event to be evaluated is sim_th, if , it is considered that event is similar to event sen k . All similar events of event in the event set to be evaluated eva_event_set are obtained, and the similar events of event are recorded as event .

[0132] Step 2.3.1.3: Quantify the occurrence frequency index of the open source data event to be evaluated event i by using the interval division rule. n represents the number of events of the open source data event i to be evaluated, min() represents the minimum value function, and len() represents the set element number calculation function.

[0133] If , the occurrence frequency C5 index quantification V(C5) = 1.

[0134] If , the occurrence frequency C5 index quantification V(C5) = 0.75.

[0135] If , the occurrence frequency C5 index quantification V(C5) = 0.5.

[0136] If , the occurrence frequency C5 index quantification V(C5) = 0.25.

[0137] It should be noted that the specific values of the above judgment conditions and the specific values of the quantification results are preferred in this embodiment, which can be adjusted as needed.

[0138] Step 2.3.2: Data source quantification rule. Different data sources are divided into different levels, and the data source index is quantified according to the source level of the open source data to be evaluated.

[0139] Step 2.3.2.1: Divide the data source level. According to the host unit of the news media, the media is divided into four categories: national government media class1; provincial and municipal unit media class2; commercial company media class3; and other information channels class4.

[0140] Step 2.3.2.2: Quantify the data source based on the data source level. Trace the source of the open source data event to be evaluated event, recorded as class, then the data source C6 quantification rule is as follows.

[0141] If class e class1, the data source C6 indicator quantization V(C6) = 1;

[0142] If class e class2, the data source C6 indicator quantization V(C6) = 0.75;

[0143] If class e class3, the data source C6 indicator quantization V(C6) = 0.5;

[0144] If class e class4, the data source C6 indicator quantization V(C6) = 0.25.

[0145] It should be noted that the specific values of the above quantization results are preferred in this embodiment, and can be adjusted as needed.

[0146] Step 3: Calculate the data quality based on the weight and quantization rule.

[0147] The weights and quantization values of the spatiotemporal consistency, element consistency, element completeness, content integrity, occurrence frequency and data source are respectively denoted as W(C1), V(C1), W(C2), V(C2), W(C3), V(C3), W(C4), V(C4), W(C5), V(C5), W(C6), V(C6). The data quality of the open source data event to be evaluated is denoted as R(event), and the calculation formula is as follows:

[0148]

[0149] Wherein, C i is the i-th scheme layer indicator, i = 1, 2, 3, 4, 5, 6; W(C i ) is the weight of the scheme layer indicator C i , and V(C i ) is the quantization result of the scheme layer indicator C i .

[0150] Based on the same technical concept, the embodiment of the present application also provides an electronic device which can realize the open source data quality quantization calculation method flow provided by the above-mentioned embodiment of the present application. In an embodiment, the electronic device can be a server, a terminal device or other electronic device. As shown in the figure, the electronic device can include: Figure 7

[0151] At least one processor and a memory connected with the at least one processor, and the specific connection medium between the processor and the memory is not limited in the embodiment of the present application, Figure 7 The connection between the processor and the memory in the embodiment of the present application is taken as an example of connection through bus. The bus is​ Figure 7 The connection between the components is indicated by a thick line, and the connection between the components is only illustrative and is not limited. The bus can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, Figure 7 In the figure, only one thick line is used to represent the connection between the components, but it does not mean that there is only one bus or one type of bus. Alternatively, the processor can also be referred to as a controller, and the name is not limited.

[0152] In the embodiment of the application, the memory stores instructions executable by the at least one processor, and the at least one processor can execute the open source data quality quantification calculation method of the fusion generative AI discussed above by executing the instructions stored in the memory. The processor can realize Figure 7 The functions of each module in the device shown.

[0153] The processor is the control center of the device, and can utilize various interfaces and lines to connect each part of the entire control device. By running or executing instructions stored in the memory and calling data stored in the memory, the device processes data and performs various functions, thereby monitoring the entire device.

[0154] In an alternative design, the processor can include one or more processing units, and the processor can integrate an application processor and a modem processor, wherein the application processor mainly processes the operating system, user interface, and application program, etc., and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor can also not be integrated into the processor. In some embodiments, the processor and the memory can be implemented on the same chip, and in some embodiments, they can also be implemented on separate chips respectively.

[0155] The processor can be a general-purpose processor, such as a CPU, a digital signal processor, an application-specific integrated circuit, a field programmable gate array, or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, which can implement or execute the methods, steps and logic block diagrams disclosed in the embodiments of the application. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the open source data quality quantification calculation method of the fusion generative AI disclosed in the embodiments of the application can be directly embodied as execution completed by a hardware processor, or executed by a combination of hardware and software modules in the processor.

[0156] The memory, as a non-volatile computer readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules. The memory can include at least one type of storage medium, for example, can include flash memory, hard disk, multimedia card, card type memory, random access memory (RAM), static random access memory (SRAM), programmable read-only memory (PROM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), magnetic memory, magnetic disk, optical disk, etc. The memory is any other medium capable of carrying or storing desired program code in the form of instructions or data structures and capable of being accessed by a computer, but is not limited thereto. The memory in the embodiments of the present application can also be a circuit or any other device capable of realizing the storage function, used for storing program instructions and / or data.

[0157] By designing and programming the processor, the code corresponding to the open source data quality quantification calculation method of the fusion generative AI introduced in the foregoing embodiments can be fixed in the chip, so that the chip can execute the steps of the method of the foregoing embodiments at runtime. How to design and program the processor is a technology known to those skilled in the art, which will not be described here.

[0158] Based on the same inventive concept, the embodiments of the present application also provide a storage medium storing computer instructions, when the computer instructions run on a computer, the computer instructions make the computer execute the foregoing open source data quality quantification calculation method of the fusion generative AI.

[0159] In some optional embodiments, the present application also provides various aspects of an open source data quality quantification calculation method of the fusion generative AI, which can also be realized in the form of a program product, including program code, when the program product runs on the device, the program code is used to make the control device execute the steps of the open source data quality quantification calculation method of the fusion generative AI according to various exemplary embodiments of the present application described in the foregoing description of the present application.

[0160] It should be noted that, although several units or sub-units of the apparatus are mentioned in the above detailed description, such division is merely exemplary and not mandatory. Indeed, according to an embodiment of the application, the features and functionalities of two or more units described above can be embodied in one unit. Conversely, the features and functionalities of one unit described above can be further divided into units embodied by several units. Moreover, although the operations of the method of the application are described in a particular, sequential order, this is not necessarily the case. Indeed, certain of the steps can be performed in a different order than that described, or can be performed concurrently. Additionally or alternatively, certain steps can be omitted, combined, or further divided into multiple steps.

[0161] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0162] The present application is described in reference to the flowchart illustrations and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processing system or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams block or blocks.

[0163] The program code can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server.

[0164] In situations where the remote computing device is involved, the remote computing device can be connected to the user computing device through any kind of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (e.g., using an Internet service provider to connect through the Internet).

[0165] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0166] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions executed on the computer or other programmable apparatus provide steps for implementing the Figure 1 function specified in the flow or flows and / or blocks Figure 1 of the block or blocks.

[0167] The above only describes the preferred embodiments of the present application and is not intended to limit the present application. The present application can have various modifications and changes for those skilled in the art. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An open source data quality quantification calculation method integrated with generative AI, characterized by: include: Build a data quality quantitative indicator system and configure the weights of each data quality quantitative indicator; For the data quality quantitative indicator system, fusion generative AI is used to design quantitative rules for each data quality quantitative indicator; Based on the weights and quantification rules, the data quality is calculated.

2. The open source data quality quantification calculation method for integrating generative AI according to claim 1 is characterized in that: The construction of the data quality quantitative indicator system includes: Construct a framework for the data quality quantitative indicator system, including the target layer, the criterion layer below the target layer, and the solution layer below the criterion layer; The weights of each data quality quantitative indicator are configured based on the hierarchical analysis method.

3. The open source data quality quantification calculation method for integrating generative AI according to claim 2 is characterized in that: The target layer refers to the quality of open source data; the indicators of the criterion layer include integrity, consistency and diversity; the program layer indicators under consistency include spatiotemporal consistency and content consistency, the program layer indicators under integrity include element completeness and content completeness, and the program layer indicators under diversity include frequency of occurrence and data source.

4. The open source data quality quantification calculation method for integrating generative AI according to claim 3 is characterized in that: The fusion generative AI designs quantitative rules for various data quality quantitative indicators, including: Quantification rules for consistency indicators, including equivalent quantification rules for spatiotemporal consistency and quantification rules for element consistency; Quantification rules for completeness indicators, including quantification rules for element completeness and quantification rules for content completeness; Quantification rules for diversity indicators include quantification rules for frequency of occurrence and quantification rules for data sources.

5. The open source data quality quantification calculation method for integrating generative AI according to claim 4 is characterized in that: The consistency indicator quantification rules include: Build a fact database based on similarity functions for generative AI models; Based on the fact database, the generative AI model is used and the similarity function is used to recall the related facts of the switch data to be evaluated; Determine the consistency of events based on the associated fact set, with the results including conflict, no conflict, and indeterminate. Assume that in the event judgment results, conflicts occurred M times, non-conflicts occurred N times, and judgment could not be made nMN times. The consistency is quantified based on the event consistency judgment results: if Then the quantification of the spatiotemporal consistency C1 indicator and the element consistency C2 indicator is V(C1)=V(C2)=0.2; if And N≤nMN, then the quantification of the spatiotemporal consistency C1 indicator and the element consistency C2 indicator is V(C1)=V(C2)=0.5; if And N>nMN, then the quantification of the spatiotemporal consistency C1 indicator and the element consistency C2 indicator is V(C1)=V(C2)=0.7; If M = 0 and N ≤ nMN, then the quantification of the spatiotemporal consistency C1 indicator and the element consistency C2 indicator is V(C1) = V(C2) = 0.8; If M=0 and N>nMN, the spatiotemporal consistency C1 indicator quantification and the element consistency C2 indicator quantification V(C1)=V(C2)=1.

6. The open source data quality quantification calculation method for integrating generative AI according to claim 4 is characterized in that: The element completeness quantification rules include: Generative AI models extract elements from open source data to be evaluated, including time, location, subject, object, and behavior; All elements in the open source data to be evaluated are accumulated to obtain the number of elements in the open source data to be evaluated, which is recorded as num_ele. The element completeness C3 is quantified as follows: Here, n represents the number of events in the open source data to be evaluated.

7. The open source data quality quantification calculation method for integrating generative AI according to claim 4 is characterized in that: The content integrity quantification rules include: Determine the completeness of grammatical elements for generative AI models; Note the number of complete grammatical elements num_full and the number of incomplete grammatical elements num_else in all events of the open source data to be evaluated, and num_full + num_else = n. Based on this, calculate the quantitative value of content completeness C4: If num_else≤0.1n, then the content completeness C4 indicator quantization V(C4)=1; If 0.1n < num_else ≤ 0.2n, the quantization of the content integrity C4 index V(C4) = 0.7; If 0.2n < num_else ≤ 0.3n, the quantization of the content integrity C4 index V(C4) = 0.5; If 0.3n < num_else ≤ 0.4n, the quantization of the content integrity C4 index is V(C4) = 0.3; If 0.4n < num_else, the quantization of the content integrity C4 index V(C4) = 0.

8. The open source data quality quantification calculation method for integrating generative AI according to claim 4 is characterized in that: The above-mentioned quantization rules for the occurrence frequency include: Construct an event set to be evaluated eva_event_set; Based on the set of events to be evaluated, we use the generative AI model and similarity function to find similar events to be evaluated. The number of similar events to be evaluated is Quantify the occurrence frequency index of the open-source data to be evaluated by using the interval division rule: if Then the frequency C5 indicator quantification V(C5) = 1; if Then the frequency C5 index quantification V(C5) = 0.75; if Then the frequency C5 index quantification V(C5) = 0.5; if Then the frequency C5 index quantification V(C5) = 0.25; Where n represents the number of events in the open-source data to be evaluated, min() represents the function to obtain the minimum value, and len() represents the function to calculate the number of set elements.

9. The open source data quality quantification calculation method for integrating generative AI according to claim 4 is characterized in that: The above-mentioned quantization rules for the data source include: Divide the data source levels, including national government media class1, provincial and municipal unit media class2, commercial company media class3, and other information channels class4; Quantify the data source based on the data source level, trace the source of the open-source data event to be evaluated, denoted as class, and the quantization rule of the data source C6 is as follows: If class ∈ class1, the quantization of the data source C6 index V(C6) = 1; If class ∈ class2, the quantization of the data source C6 index V(C6) = 0.75; If class ∈ class3, the quantization of the data source C6 index V(C6) = 0.5; If class ∈ class4, the quantization of the data source C6 index V(C6) = 0.

25.

10. The open source data quality quantification calculation method for integrating generative AI according to claim 1 is characterized in that: The above-mentioned calculation of the data quality based on the weight and quantization rules includes: Among them, R(event) is the data quality of the open source data event to be evaluated, C i is the i-th solution level indicator, i=1,2,3,4,5,6; W(C i ) is the scheme level indicator C i The weight of V(C i ) is the scheme level indicator C i quantified results.

Images